Compressor control method, controller, air conditioner, heat management system and vehicle

Abstract

The invention discloses a compressor control method, a controller, an air conditioner, a heat management system and a vehicle, and the compressor control method comprises the steps that a compressor rotating speed control instruction and a historical state value of an air conditioning system where a compressor is located are obtained, and the compressor rotating speed control instruction and the historical state value are used for inputting a deep learning model; an output value of the deep learning model is obtained, wherein the output value represents the space temperature of the corresponding air conditioning system temperature adjusting space under the compressor rotating speed control instruction; and a target compressor rotating speed control instruction is determined according to the corresponding relation between the compressor rotating speed control instruction and the space temperature so as to control the compressor. By adopting the method, the workload of a data acquisition process can be reduced, multi-objective optimization is carried out on time and energy consumption in a temperature regulation process, meanwhile, the temperature regulation duration is shortened, a control strategy with the lowest energy consumption and the fastest time consumption is realized, and the control method has the advantages of good self-learning performance, relatively high robustness and the like.

Description

Technical Field

[0001] The present invention relates to the field of vehicles, and more particularly to a compressor control method, a controller, an air conditioning system, a thermal management system, and a vehicle. Background Art

[0002] In related technologies, currently, the compressor control method collects the temperature inside the vehicle and the low pressure of the air conditioning system to judge the temperature change trend; within one control cycle, according to the deviation value △T between the temperature inside the vehicle and the user-set temperature and the temperature change trend, the compressor capacity demand coefficient K is calculated, and then the compressor output frequency is calculated; within one compensation cycle, according to the deviation value △P between the low pressure of the air conditioning system and the target low pressure, the low pressure compensation control method is used to calculate the low pressure compensation value, and the compressor output frequency is changed according to the low pressure compensation value. This method relies on the temperature and pressure acquisition measuring points in the air conditioning system and their accuracy. Once the sensor fails, it will cause control errors; it requires temperature feedback, there is a certain time delay, and it is difficult to quickly meet the user's needs. Summary of the Invention

[0003] The present invention aims to at least solve one of the technical problems existing in the prior art. To this end, the first object of the present invention is to propose a compressor control method, which can learn the compressor speed control through a deep learning model to achieve the control strategy with the lowest energy consumption and the shortest time.

[0004] The second object of the present invention is to propose a controller.

[0005] The third object of the present invention is to propose an air conditioning system.

[0006] The fourth object of the present invention is to propose a thermal management system.

[0007] The fifth object of the present invention is to propose a vehicle.

[0008] To solve the above problems, an embodiment of the first aspect of the present invention provides a compressor control method, which is characterized by including: obtaining a compressor speed control instruction and historical state values of the air conditioning system where the compressor is located, the compressor speed control instruction and the historical state values being used as inputs to a deep learning model; obtaining an output value of the deep learning model, the output value representing the space temperature of the temperature adjustment space of the air conditioning system corresponding to the compressor speed control instruction; and determining a target compressor speed control instruction according to the corresponding relationship between the compressor speed control instruction and the space temperature to control the compressor.

[0009] The compressor control method according to an embodiment of the present invention uses a deep learning model to determine the optimal compressor speed control method through relevant feature data and historical state values. It can directly utilize the original data, avoid human interference, reduce the workload of data acquisition, and the deep learning model can achieve a relatively fast training speed relying on the number of computing cores. Moreover, the model occupies a small space, has high prediction accuracy and fast prediction speed, can be conveniently and quickly applied to different environments, has little dependence on sensors, and does not require temperature feedback, thus improving user satisfaction.

[0010] In some embodiments, determining the target compressor speed control instruction according to the correspondence between the compressor speed control instruction and the space temperature includes: obtaining a plurality of compressor speed control instruction strings, each of the compressor speed control instruction strings including a plurality of compressor speed control instructions for making the space temperature reach the expected temperature from the current temperature; the target compressor speed control instruction is the target compressor speed control instruction string determined from the plurality of compressor speed control instruction strings according to the temperature adjustment process consumption parameter value; wherein, the temperature adjustment process consumption parameter value includes the target parameter value consumed for controlling the compressor according to each compressor speed control instruction string to make the space temperature reach the expected temperature from the current temperature.

[0011] In some embodiments, the temperature adjustment process consumption parameter value includes the temperature adjustment time; the temperature adjustment time includes the time consumed for controlling the compressor according to each compressor speed control instruction string to make the space temperature reach the expected temperature from the current temperature.

[0012] In some embodiments, the target compressor speed control instruction is the compressor speed control instruction string corresponding to the plurality of compressor speed control instruction strings whose corresponding temperature adjustment time meets the temperature adjustment time condition.

[0013] In some embodiments, the temperature adjustment process consumption parameter value includes the compressor temperature adjustment power consumption; the compressor temperature adjustment power consumption includes the power consumption of the compressor for controlling the compressor according to each compressor speed control instruction string to make the space temperature reach the expected temperature from the current temperature.

[0014] In some embodiments, the target compressor speed control instruction is the compressor speed control instruction string corresponding to the plurality of compressor speed control instruction strings whose corresponding compressor temperature adjustment power consumption meets the compressor power consumption condition.

[0015] In some embodiments, the deep learning model is constructed with the parameter data related to temperature adjustment of the air conditioning system under different working conditions and different operation cycles as the input and the space temperature as the output.

[0016] In some embodiments, the following constraint conditions are satisfied in the construction of the deep learning model: the space temperature at each moment is maintained at the required temperature of the temperature-controlled space; each historical state value satisfies the safety constraints of the air-conditioning system; the compressor speed control instruction increases or decreases monotonically with time.

[0017] In some embodiments, the hyperparameters of the deep learning model include: the activation function is the Sigmoid function, the weight initialization method is the normal distribution, the loss function is the square loss, the optimization algorithm is the Adam optimization, and the BatchSize is 1000.

[0018] In some embodiments, the hyperparameters of the deep learning model further include: the learning rate is between 0.0001 and 0.001, the number of neurons is between 50 and 100, the number of network layers is between 1 and 3, and the associated time step is (1000 ± Δ).

[0019] An embodiment of the second aspect of the present invention provides a controller, which includes: a processor configured with the deep learning model; a memory communicatively connected to the processor; the memory stores a computer program executable by the processor, and when the processor executes the computer program, it implements the compressor control method described in the above embodiments.

[0020] According to the controller of the embodiment of the present invention, by executing the compressor control method of the above embodiment, the control speed is faster. The deep learning model can obtain a faster training speed depending on the number of computing cores, and the model occupies a small space, has high prediction accuracy and fast prediction speed, can be conveniently and quickly applied to different environments, has little dependence on sensors, does not require temperature feedback, and improves user satisfaction.

[0021] An embodiment of the third aspect of the present invention provides an air-conditioning system, which includes: a compressor; the controller described in the above embodiment, and the controller is connected to the compressor.

[0022] According to the air-conditioning system of the embodiment of the present invention, by adopting the controller of the above embodiment, faster control can be achieved, the dependence on sensors is small, temperature feedback is not required, and user satisfaction is improved.

[0023] An embodiment of the fourth aspect of the present invention provides a thermal management system, which includes the air-conditioning system described in the above embodiment.

[0024] According to the thermal management system of the embodiment of the present invention, by adopting the air-conditioning system of the above embodiment, faster control can be achieved, the dependence on sensors is small, temperature feedback is not required, and user satisfaction is improved.

[0025] An embodiment of the fifth aspect of the present invention provides a vehicle, which is characterized by including the air conditioning system described in the above embodiment.

[0026] For the vehicle according to the embodiment of the present invention, by adopting the air conditioning system of the above embodiment, faster control can be achieved, with less dependence on sensors, no need for temperature feedback, and user satisfaction can be improved.

[0027] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The above and / or additional aspects and advantages of the present invention will become apparent and be readily understood from the description of the embodiments in conjunction with the following drawings, where:

[0029] Figure 1 is a flowchart of a compressor control method according to an embodiment of the present invention;

[0030] Figure 2 is a flowchart of optimizing a control instruction string according to an embodiment of the present invention;

[0031] Figure 3 is a schematic diagram of data preprocessing according to an embodiment of the present invention;

[0032] Figure 4 is a schematic diagram of training a deep learning model according to an embodiment of the present invention;

[0033] Figure 5 is a flowchart of an optimization scheme for a compressor speed command of a deep learning model according to an embodiment of the present invention;

[0034] Figure 6 is a structural block diagram of a controller according to an embodiment of the present invention;

[0035] Figure 7 is a structural block diagram of an air conditioning system according to an embodiment of the present invention;

[0036] Figure 8 is a structural block diagram of a thermal management system according to an embodiment of the present invention;

[0037] Figure 9 is a structural block diagram of a vehicle according to an embodiment of the present invention.

[0038] REFERENCE SIGNS:

[0039] Controller 10; Air conditioning system 20; Thermal management system 30; Vehicle 40;

[0040] Processor 1; Memory 2; Compressor 3. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] Embodiments of the present invention will be described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. Embodiments of the present invention will be described in detail below.

[0042] Figure 1 is a flowchart of a compressor control method according to an embodiment of the present invention. As Figure 1 shown, the compressor control method includes steps S1 - S3.

[0043] S1, Obtain a compressor speed control instruction and historical state values of the air - conditioning system where the compressor is located. The compressor speed control instruction and the historical state values are used as inputs to a deep - learning model.

[0044] Specifically, a deep - learning model for the compressor control method can be pre - established. This deep - learning model can take the compressor speed control instruction and the historical state values of the air - conditioning system where the compressor is located as inputs, and the space temperature of the temperature - controlled space of the air - conditioning system, such as the occupant compartment or a room or other spaces, as the output, perform model learning and training, and obtain a model that meets the requirements and save it in the processor.

[0045] When controlling the compressor, after obtaining the compressor speed control instruction and the historical state values of the air - conditioning system where the compressor is located, input them into the deep - learning model, and analyze the data through the deep - learning model. The historical state values of the air - conditioning system where the compressor is located can be understood as the state values of the air - conditioning system during the temperature - control process at a time before the current moment, such as the compressor speed, pressures at each measuring point, temperature, etc.

[0046] Among them, deep learning is a type of machine learning. The concept of deep learning stems from the research on artificial neural networks. A multi - layer perceptron with multiple hidden layers is a deep - learning structure. Deep learning forms more abstract high - level representations (such as attribute categories or features) by combining low - level features to discover the distributed feature representations of data. Studying deep learning can establish a neural network that simulates the human brain for analysis and learning. It imitates the mechanism of the human brain to interpret data, such as images, sounds, and texts, etc. During the deep - learning process, by designing and establishing an appropriate number of neuron computing nodes and multi - layer operation hierarchical structures, selecting appropriate input and output layers, and through the learning and optimization of the network, a functional relationship from input to output is established. Although it is impossible to find the functional relationship between input and output 100%, it can approximate the real - world correlation relationship as much as possible to meet the automation requirements for processing complex affairs.

[0047] S2, Obtain the output value of the deep - learning model. The output value represents the space temperature of the temperature - controlled space of the air - conditioning system corresponding to the compressor speed control instruction.

[0048] Specifically, after obtaining the compressor speed control instruction and the historical state value of the air conditioning system where the compressor is located, input them into the deep learning model. The deep learning model analyzes the data to obtain an output value. The output value of the deep learning model represents the space temperature of the temperature adjustment space of the air conditioning system corresponding to the compressor speed control instruction. In some embodiments, the temperature adjustment space of the air conditioning system can be the passenger compartment of a vehicle, and the space temperature is the passenger compartment temperature. Or, the temperature adjustment space of the air conditioning system is a room or other spaces where the air conditioning system is used for temperature adjustment, which is not specifically limited herein.

[0049] S3. Determine the target compressor speed control instruction according to the correspondence between the compressor speed control instruction and the space temperature to control the compressor.

[0050] Specifically, after inputting the compressor speed control instruction and the historical state value of the air conditioning system where the compressor is located into the deep learning model, the output value of the deep learning model is obtained. At this time, the output value of the deep learning model is the space temperature of the temperature adjustment space of the air conditioning system corresponding to the compressor speed control instruction. According to the correspondence between the compressor speed control instruction and the space temperature, guide the compressor control instruction, that is, determine the target compressor speed control instruction. For example, increase, decrease or maintain based on the current compressor speed control instruction, etc. Control the operation of the compressor according to the output compressor speed control instruction, so that the space temperature of the temperature adjustment space of the air conditioning system reaches the corresponding temperature value.

[0051] According to the compressor control method of the embodiment of the present invention, by using a deep learning model, the best compressor speed control method is determined through relevant feature data and historical state values. The original data can be directly used to avoid human interference, reduce the workload of data acquisition. And the deep learning model can obtain a faster training speed depending on the number of computing cores, and the model occupies less space, has high prediction accuracy and fast prediction speed, can be conveniently applied to different environments, has less dependence on the performance of sensors, and does not require temperature feedback, thus improving user satisfaction.

[0052] In some embodiments, the control of the temperature adjustment process of the compressor can be optimized based on the correspondence between the space temperature obtained by the deep learning model and the compressor control instruction. Determining the target compressor speed control instruction according to the correspondence between the compressor speed control instruction and the space temperature includes: obtaining a plurality of compressor speed control instruction strings, each compressor speed control instruction string including a plurality of compressor speed control instructions for making the space temperature reach the expected temperature from the current temperature; the target compressor speed control instruction is the target compressor speed control instruction string determined from the plurality of compressor speed control instruction strings according to the parameter value consumed in the temperature adjustment process; wherein, the parameter value consumed in the temperature adjustment process includes the target parameter value consumed for controlling the compressor according to each compressor speed control instruction string to make the space temperature reach the expected temperature from the current temperature.

[0053] Among them, the parameter value consumed in the temperature adjustment process can be understood as the parameter corresponding to the substance that needs to be spent or consumed during the temperature adjustment process of the compressor, such as time or power consumption.

[0054] Specifically, collect the compressor speed control instruction string data to obtain multiple compressor speed control instruction strings. Each compressor speed control instruction string contains multiple compressor speed control instructions from the current temperature to the expected temperature, that is, a series of compressor control instructions during the process of controlling the space temperature to gradually reach the expected temperature from the current temperature. Then, analyze and calculate according to the multiple compressor speed control instructions to obtain the corresponding compressor speed control instruction string, that is, obtain the optimal compressor control. The target parameter value consumed by each compressor speed control instruction string to control the compressor to make the space temperature gradually reach the expected temperature from the current temperature can be understood as the change value of each parameter of the compressor to reach the expected temperature, such as time or power consumption.

[0055] In some embodiments, the parameter value consumed in the temperature adjustment process may include the temperature adjustment time; the temperature adjustment time includes the time consumed to control the compressor to make the space temperature reach the expected temperature according to each compressor speed control instruction string.

[0056] Specifically, during the temperature adjustment process, the time consumed by each compressor speed control instruction string to control the compressor to make the space temperature reach the expected temperature from the current temperature is different, that is, the consumed parameter value is different. Therefore, it is necessary to analyze multiple compressor speed control instruction strings to obtain the compressor speed control instructions that meet the conditions.

[0057] In some embodiments, the target compressor speed control instruction is the compressor speed control instruction string in which the corresponding temperature adjustment time in multiple compressor speed control instruction strings meets the temperature adjustment time condition.

[0058] Specifically, after inputting multiple compressor speed control instruction strings into the deep learning model, analyze to obtain the target compressor speed control instruction. The target compressor control instruction can be the compressor speed control instruction string in multiple compressor speed control instruction strings that takes the shortest time to make the space temperature reach the expected temperature, or the compressor speed control instruction string that takes a certain time limit or at a certain time to make the space temperature reach the preset temperature. The temperature adjustment time condition can also be set based on requirements.

[0059] In some embodiments, the parameter value consumed in the temperature adjustment process may include the compressor temperature adjustment power consumption; the compressor temperature adjustment power consumption includes the compressor power consumption to control the compressor to make the space temperature reach the expected temperature according to each compressor speed control instruction string.

[0060] Specifically, during the temperature adjustment process, each compressor speed control instruction string controls the compressor such that the compressor power consumption required for the space temperature to reach the expected temperature from the current temperature is different, that is, the consumption parameter values are different. Therefore, it is necessary to analyze multiple compressor speed control instruction strings to obtain the compressor speed control instructions that meet the conditions.

[0061] In some embodiments, the target compressor speed control instruction is the compressor speed control instruction string among multiple compressor speed control instruction strings for which the corresponding compressor temperature adjustment power consumption meets the compressor power consumption condition.

[0062] Specifically, after inputting multiple compressor speed control instruction strings into the deep learning model, the target compressor speed control instruction is obtained through analysis. The target compressor control instruction can be the compressor speed control instruction string among multiple compressor speed control instruction strings that has the lowest compressor power consumption when the space temperature reaches the expected temperature, or the compressor speed control instruction string with a power consumption lower than a certain threshold. The compressor power consumption condition can be set according to requirements.

[0063] In some embodiments, the temperature adjustment time and the compressor temperature adjustment power consumption can also be considered comprehensively. The target compressor control instruction is the compressor control instruction string among multiple compressor speed control instruction strings that meets the temperature adjustment time and the compressor temperature adjustment power consumption conditions. Similarly, the temperature adjustment time and the compressor power consumption conditions can be set as needed and are not specifically limited herein.

[0064] For example, using a deep learning model that has achieved prediction accuracy, input a certain compressor speed adjustment and historical data before adjustment. After calculation, the expected space temperature, such as the temperature of the passenger compartment, is obtained to achieve accurate estimation of the compressor speed - temperature and the purpose of optimizing the compressor speed control.

[0065] During the prediction process, input the current possible compressor adjustment speed and various data within a preset time period traced back. Expressed by the formula:

[0066] Tt = f({θ t , A t-1 , A t-2 , …, A t-m} T , T t-1 )

[0067] In the formula, Tt is the space temperature at the current time t predicted by the deep learning model at a certain input speed; Tt - 1 is the space temperature at time t - 1; θ t is the compressor adjustment speed input at the current time t, and A t-mis the historical state value for the previous m moments. That is, the input information needs to include the compressor speed at time t, the state parameters from the previous moment to the previous m moments, and the temperature at the previous moment. Then, using the trained deep learning model, the change in space temperature caused by this speed adjustment instruction can be predicted, thereby guiding the adjustment of the compressor speed.

[0068] Determine the required compressor speed at a predetermined temperature according to the established prediction model, optimize the control instruction string for the regulation time and process energy consumption, and then perform control according to this instruction, such as Figure 2 shown, including steps S4 - S7.

[0069] Step S4, predict the space temperature under a certain compressor speed instruction based on the deep learning model.

[0070] Step S5, predict the temperature at continuous times under the compressor speed instruction string based on the above strategy.

[0071] Step S6, select an optimization algorithm, with the optimization variable being the control instruction string and the optimization objective being the temperature regulation time and energy consumption.

[0072] Step S7, complete the optimization and give the control instruction for the best regulation time and process energy consumption.

[0073] According to the system temperature prediction model, we can predict the space temperature at several moments from the compressor speed instructions within the selected period. Expressed by the formula:

[0074]

[0075] It can be seen from the above formula that only based on the state values at historical moments and the compressor speed control instruction string within a certain period, the temperature values at several future moments can be predicted. Expressed by the formula:

[0076] [T t ,T t+1 ,…T t+n = F(A t-1 ,A t-2 ,…,A t-m ,[θ t ,θ t+1 ,…,θ t+n )

[0077] In the formula, F - the unit condition prediction combination function based on the deep learning model, F is composed of n f's. Thus, this function can be used to optimize the temperature regulation time and power consumption. Among them, the optimization method selects the genetic algorithm, and the optimization independent variable is the compressor speed adjustment control instruction string within a certain period, that is, [θ t ,θ t+1 ,…,θ t+n, the optimized cost function is the above function F, and the optimization objectives are the temperature reaching time n and the power consumption required for reaching the standard.

[0078] In some embodiments, the deep learning model is constructed with the parameter data related to temperature regulation of the air conditioning system under different working conditions and different operating cycles as the input and the space temperature as the output.

[0079] Specifically, data of the air conditioning system is collected. Based on the vehicle operation or bench test data, data is collected using each data monitoring point on the system, and data features related to the temperature regulation of the temperature control space are selected.

[0080] For example, according to the existing original historical record information or bench test operation data, various parameter data related to the in-vehicle temperature regulation under different operating conditions and different operating cycles are obtained; for example, operating data such as compressor speed, valve opening, refrigerant flow rate, pressure and temperature of each measurement point, etc. The collected data can be one or more of the above parameters. By adjusting different compressor speeds, controlling parameters such as valve opening and inlet air volume, the refrigerant side flow rate, temperature and pressure can be changed to affect the temperature of the temperature control space, and various data are collected.

[0081] Preprocessing work such as screening and expanding the obtained data of each group dimension is carried out. As Figure 3 shown: Data is extracted at the same time interval to ensure the continuity of information, such as a time interval of 1 - 2s; invalid points are removed according to the parameter constraint criterion, such as abnormal data with a valve opening exceeding 100% is removed; data is supplemented according to the data interpolation principle. For parameters with inheritance attributes such as valve position opening command, the value of the previous moment is inherited, otherwise linear interpolation is used.

[0082] In some embodiments, the following at least one constraint condition is satisfied in the construction of the deep learning model: the space temperature at each moment is maintained at the required temperature of the temperature control space; each historical state value satisfies the safety constraints of the air conditioning system; the compressor speed control command increases or decreases monotonically with time.

[0083] Specifically, the space temperature at each moment is maintained at the required temperature of the temperature control space; for example, the temperature of the occupant compartment at the nth moment is stabilized at the required temperature, that is, Tn = Tneed, where Tneed is the required temperature inside the vehicle.

[0084] Each historical state value satisfies the safety constraints of the air conditioning system. For example, the state changes from the tth moment to the (t + n)th moment satisfy the system safety constraints, and the threshold of the safety constraints can be determined by the user's safety level. Such as the speed change does not exceed 1000 rpm / s and the pressure change of each measurement point does not exceed 5 bar / s during the process.

[0085] The compressor speed control command increases or decreases monotonically with time, that is, the command does not reciprocate during the control process. In addition, it is allowed to remain unchanged. Through constrained multi-objective optimization based on the genetic algorithm, the fastest adjustment time noptim, the lowest power consumption, and the compressor speed control command string [θ t , θ t+1 , …, θ t+n optim can be obtained under the system state xt at the current moment and the temperature demand command Tneed given by the user. At this time, the optimization is completed.

[0086] In some embodiments, the hyperparameters of the deep learning model include: the activation function is the Sigmoid function, the weight initialization method is the normal distribution, the loss function is the squared loss, the optimization algorithm is Adam optimization, and the BatchSize is 1000.

[0087] Specifically, in information science, due to its properties such as monotonic increase and monotonic increase of the inverse function, the Sigmoid function is often used as the activation function of the neural network; the Adam (Adaptive Moment Estimation) algorithm is a gradient-based optimization algorithm that updates the model's parameters by calculating the gradient. As a widely used optimization algorithm, the Adam optimization algorithm has good performance balance and robustness; BatchSize refers to the size of the data set input each time the network is trained in deep learning. When training a neural network, the data is usually divided into small batches (Batch) for training. Each small batch contains multiple data samples, and the size of this small batch is BatchSize. Generally speaking, choosing an appropriate BatchSize can speed up the training speed and improve the training effect.

[0088] After selecting and determining the hyperparameters of the neural network, the deep learning model is trained, and the model training is completed under the preset convergence conditions. As Figure 4 shown, it specifically includes:

[0089] A deep learning model is established. The input variables of this model include the current compressor speed adjustment command and historical state values such as compressor speed, pressures at each measuring point, temperature, etc. It is expressed by the formula:

[0090] x t = {θ t , A t-1 , A t-2 , …, A t-m} T

[0091] In the formula, θ t —— the input at time t, A t-k ​—— The state values of the historical data for the previous k moments;

[0092] The model can calculate the temperature of the passenger compartment at the current moment based on xt and the temperature at the previous moment, which is expressed by the formula:

[0093] T t =f(x t ,T t-1 )

[0094] In the formula, Tt is the temperature of the passenger compartment predicted by the model at the current moment t; Tt-1 is the temperature of the passenger compartment at the previous moment; xt is the compressor speed command at the current moment t and the historical state values for the previous m moments; then, the hyperparameters to be used by the deep learning model are determined: The hyperparameters mainly include fixed parameters, such as the activation function being the Sigmoid function, the weight initialization method being the normal distribution, the loss function being the squared loss, the optimization algorithm being the Adam optimization, and the BatchSize being 1000.

[0095] In some embodiments, the hyperparameters of the deep learning model further include: the learning rate is between 0.0001 and 0.001, the number of neurons is between 50 and 100, the number of network layers is between 1 and 3, and the associated time step is (1000 ± Δ).

[0096] Specifically, the learning rate is between 0.0001 and 0.001, the number of neurons is between 50 and 100, the number of network layers is between 1 and 3, and the associated time step is (1000 ± Δ). The neural network model is adjusted by regulating the hyperparameters. The hyperparameters are set for the neural network according to the above ranges, and the processed data is divided into a 70% training set and a 30% test set; on the training set, a group of batch data is sequentially selected for network training. On the test set, the monitored data is input to predict the temperature Tti, and the true temperature on the test set corresponding to the input data is Toi. Calculate the error between the predicted value and the true value for each batch, that is

[0097]

[0098] The number of training times u of the neural network is large enough. When Δ i ≤0.05, the training ends.

[0099] Next, with reference to Figure 5 as shown below, an example of the compressor control method according to the embodiments of the present invention will be described, and the specific content is as follows.

[0100] Step S8, start.

[0101] Step S9, based on the vehicle operation or bench test data, data is collected using each data monitoring point on the system, and the data features related to the space temperature adjustment are selected.

[0102] Step S10, perform preprocessing operations such as data screening and expansion on the acquired data to reduce the neural network calculation training volume and improve the generalization ability of the model.

[0103] Step S11, based on the large amount of data obtained after processing, after selecting and determining the hyperparameters of the neural network, perform the training of the deep learning model, and complete the model training under the preset convergence conditions.

[0104] Step S12, use the deep learning model that has reached the prediction accuracy, input a certain regulating compressor speed and the historical data before regulation, and obtain the expected space temperature after calculation to achieve the accurate estimation of the compressor speed - temperature, and achieve the purpose of optimizing the compressor speed control.

[0105] Step S13, determine the required compressor speed at a predetermined temperature according to the prediction model, optimize the control instruction string for the regulation time and process energy consumption, and then perform the control according to this instruction.

[0106] Step S14, end.

[0107] As above, the method of the embodiment of the present invention provides an optimization scheme for the compressor speed command based on a deep learning model. For example, taking a vehicle air conditioner as an example, it includes: The first step, air conditioning system data acquisition. That is, based on the vehicle operation database or bench test data, use each monitoring point to collect data for several time periods, and select data features related to the occupant compartment temperature control; The second step, raw data preprocessing. Perform preprocessing operations such as data screening and expansion on the data obtained in the first step to reduce the neural network calculation training volume and improve the generalization ability of the model; The third step, based on the large amount of data obtained after the second step, after selecting and determining the hyperparameters of the neural network, perform the training of the deep learning model, and complete the model training under a certain convergence condition; The fourth step, use the deep learning model that has reached the prediction accuracy, input the current compressor speed regulation command and the historical state data before regulation, and obtain the occupant compartment temperature after calculation to achieve the accurate estimation of the compressor speed - occupant compartment temperature; The fifth step, determine the required compressor speed at a predetermined temperature according to the prediction model established in the fourth step, optimize the control instruction string for the regulation time and process energy consumption, and then perform the control according to this instruction.

[0108] This method makes full use of the advantages of the deep learning model in processing historical dependent data, without going through a large amount of preprocessing, without using empirical correction coefficients and formulas, has the advantages of fast model training, self - adaptability, high prediction accuracy, obtains a small training model, fast prediction speed, and has good application value.

[0109] Generally speaking, the compressor control method according to the embodiments of the present invention has the following differences compared with the related art:

[0110] The compressor speed command optimization scheme according to the embodiments of the present invention is faster than the traditional temperature monitoring-feedback speed control method. The deep learning model takes the compressor speed and characteristic data monitoring variables related to the occupant compartment temperature, such as pressure and temperature, as input variables, and the occupant compartment temperature as the output variable, and supplements the system information at historical moments. It can realize the compressor speed control at a predetermined temperature based on model prediction, and has advantages such as good self-learning ability and high robustness;

[0111] The method according to the embodiments of the present invention takes into account the historical dependence of the data of the heat pump air conditioning system, directly uses the original data, does not need to rely on empirical formulas and correction coefficients, can largely avoid human interference, and truly reflects the influence of historical data on the relationship between the compressor speed and the occupant compartment temperature; at the same time, it can avoid the influence of outliers and missing values, and does not need to perform preprocessing work such as clustering and dimensionality reduction like traditional machine learning methods in advance, so the workload in the data acquisition process can be reduced;

[0112] Since the compressor speed command optimization scheme according to the embodiments of the present invention adopts model predictive control, it can predict the temperature change and energy consumption level during the temperature adjustment process. Therefore, it can perform multi-objective optimization on the time and energy consumption during the temperature adjustment process. Applying it to a real vehicle can improve the energy consumption performance of the whole vehicle, shorten the temperature adjustment duration, and improve consumer satisfaction.

[0113] In addition, for the method according to the embodiments of the present invention, once the hyperparameters of the deep learning model are determined, the deep learning model based on GPU can obtain a relatively fast training speed depending on the number of computing cores, and the obtained deep learning model occupies a small space, has high prediction accuracy and fast prediction speed.

[0114] Based on the compressor control method of the above embodiments, the second aspect of the embodiments of the present invention provides a controller, as Figure 6 shown, the controller 10 includes: a processor 1 and a memory 2.

[0115] Wherein, the memory 2 stores a computer program executable by the processor 1, and when the processor 1 executes the computer program, it implements the compressor control method of the above embodiments.

[0116] In some embodiments, the controller can be a compressor controller or a domain controller of a vehicle, etc., and no specific limitation is made here.

[0117] The controller according to the embodiment of the present invention controls faster by executing the compressor control method of the above embodiment. The deep learning model can achieve a faster training speed depending on the number of computing cores, and the model occupies a small space, has high prediction accuracy and fast prediction speed. It can be conveniently and quickly applied to different environments, has little dependence on sensors, and does not require temperature feedback, thus improving user satisfaction.

[0118] The third aspect embodiment of the present invention provides an air conditioning system, as Figure 7 shown, the air conditioning system 20 includes: a compressor 3 and a controller 10.

[0119] In the embodiment, the air conditioning system can be a vehicle-mounted air conditioning system or a household or commercial air conditioning system such as a heat pump air conditioning system, etc.

[0120] Wherein, the controller 10 is connected to the compressor 3.

[0121] Specifically, the controller 10 in the air conditioning system 20 obtains the output value of the deep learning model, represents the output value as the target speed control instruction of the compressor 3, and when the controller 10 obtains the output value of the deep learning model, it controls the compressor 3 to operate with the output value as the target speed of the compressor.

[0122] According to the air conditioning system of the embodiment of the present invention, by adopting the controller of the above embodiment, faster control can be achieved, the dependence on sensors is small, temperature feedback is not required, and user satisfaction is improved. In addition, by controlling the compressor speed through the output value of the deep learning model, the control strategy of the lowest energy consumption and the fastest time is realized by comprehensively considering the temperature adjustment speed and the process energy consumption, thus improving user satisfaction.

[0123] The fourth aspect embodiment of the present invention provides a thermal management system, as Figure 8 shown, the thermal management system 30 includes: an air conditioning system 20.

[0124] Specifically, the thermal management system 30 adjusts the space temperature through the air conditioning system 20. During the adjustment process, the controller in the air conditioning system 20 obtains the output value of the deep learning model, represents the output value as the target speed control instruction of the compressor, and when the controller obtains the output value of the deep learning model, it controls the compressor to operate with the output value as the target speed of the compressor.

[0125] According to the thermal management system of the embodiment of the present invention, by adopting the air conditioning system of the above embodiment, faster control can be achieved, the dependence on sensors is small, temperature feedback is not required. In addition, by controlling the compressor speed through the output value of the deep learning model, the control strategy of the lowest energy consumption and the fastest time is realized by comprehensively considering the temperature adjustment speed and the process energy consumption, thus improving user satisfaction.

[0126] The fifth aspect of the present invention provides a vehicle, such as Figure 9 shown, the vehicle 40 includes: an air conditioning system 20.

[0127] Specifically, the vehicle 40 adjusts the temperature of the temperature control space of the vehicle through the air conditioning system 20. The space temperature of the temperature control space is the temperature of the passenger compartment. During the temperature adjustment process, the controller in the air conditioning system 20 obtains the output value of the deep learning model, and uses the output value as the target speed control command for the compressor. When the controller obtains the output value of the deep learning model, it controls the compressor to operate with the output value as the target speed of the compressor.

[0128] For the vehicle according to the embodiment of the present invention, the in-vehicle air conditioning system uses a deep learning model, which can achieve faster control, has less dependence on sensors, does not require temperature feedback, and improves user satisfaction. In addition, the optimal compressor speed control method is determined through interaction with the environment, and the temperature adjustment speed and process energy consumption are comprehensively considered during the learning process of the control method. During application, the control speed is faster, the dependence on sensor performance is less, the vehicle energy consumption is reduced, and user satisfaction is improved.

[0129] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "schematic embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example.

[0130] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and purposes of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A compressor control method, characterized in that, it includes: Obtain a compressor speed control instruction and historical state values of the air conditioning system where the compressor is located, and the compressor speed control instruction and the historical state values are used as inputs to a deep learning model; Obtain the output value of the deep learning model, and the output value represents the space temperature of the temperature adjustment space of the air conditioning system corresponding to the compressor speed control instruction; Determine a target compressor speed control instruction according to the corresponding relationship between the compressor speed control instruction and the space temperature to control the compressor.

2. The compressor control method according to claim 1, characterized in that, Determining a target compressor speed control instruction according to the corresponding relationship between the compressor speed control instruction and the space temperature includes: Obtain a plurality of compressor speed control instruction strings, and each of the compressor speed control instruction strings includes a plurality of compressor speed control instructions for making the space temperature reach the expected temperature from the current temperature; The target compressor speed control instruction is a target compressor speed control instruction string determined from the plurality of compressor speed control instruction strings according to the temperature adjustment process consumption parameter value; Wherein, the temperature adjustment process consumption parameter value includes a target parameter value consumed for controlling the compressor according to each of the compressor speed control instruction strings to make the space temperature reach the expected temperature from the current temperature.

3. The compressor control method according to claim 2, characterized in that, The temperature adjustment process consumption parameter value includes the temperature adjustment time; The temperature adjustment time includes the time consumed for controlling the compressor according to each of the compressor speed control instruction strings to make the space temperature reach the expected temperature from the current temperature.

4. The compressor control method according to claim 3, characterized in that, The target compressor speed control instruction is a compressor speed control instruction string corresponding to the plurality of compressor speed control instruction strings whose temperature adjustment time meets the temperature adjustment time condition.

5. The compressor control method according to any one of claims 2-4, characterized in that, The temperature adjustment process consumption parameter value includes the compressor temperature adjustment power consumption; The compressor temperature adjustment power consumption includes the compressor power consumption for controlling the compressor according to each of the compressor speed control instruction strings to make the space temperature reach the expected temperature from the current temperature.

6. The compressor control method according to claim 5, characterized in that, The target compressor speed control instruction is a compressor speed control instruction string corresponding to the plurality of compressor speed control instruction strings whose compressor temperature adjustment power consumption meets the compressor power consumption condition.

7. The compressor control method according to claim 1, characterized in that, The deep learning model is constructed with parameter data related to temperature adjustment of the air conditioning system under different working conditions and different operation cycles as inputs and the space temperature as an output.

8. The compressor control method according to claim 7, characterized in that, The following at least one constraint condition is satisfied in the construction of the deep learning model: The space temperature at each moment is maintained at the required temperature of the temperature adjustment space; Each historical state value satisfies the safety constraints of the air conditioning system; The compressor speed control instruction monotonically increases or decreases with time.

9. The compressor control method according to claim 1, characterized in that the hyperparameters of the deep learning model include: the activation function is the Sigmoid function, the weight initialization method is the normal distribution, the loss function is the squared loss, the optimization algorithm is the Adam optimization, and the BatchSize is 1000.

10. The compressor control method according to claim 9, characterized in that the hyperparameters of the deep learning model further include: the learning rate is between 0.0001 and 0.001, the number of neurons is between 50 and 100, the number of network layers is between 1 and 3, and the associated time step is (1000 ± Δ).

11. A controller, characterized in that comprising: a processor, the processor being configured with the deep learning model; a memory, communicatively connected to the processor; the memory stores a computer program executable by the processor, and when the processor executes the computer program, the compressor control method according to any one of claims 1-10 is implemented.

12. An air conditioning system, characterized in that comprising: a compressor; the controller according to claim 11, the controller being connected to the compressor.

13. A thermal management system, characterized in that comprising the air conditioning system according to claim 12.

14. A vehicle, characterized in that comprising the air conditioning system according to claim 12, and the space temperature of the temperature control space of the air conditioning system is the occupant compartment temperature.

Images