Cold chain hybrid electric vehicle energy consumption optimization control method based on personalized temperature control

Abstract

The invention discloses a cold chain hybrid power vehicle energy consumption optimization control method based on personalized temperature control, and belongs to the field of cold chain transport vehicle energy consumption optimization, and the method comprises the steps: obtaining the travel information of a cold chain hybrid power vehicle, and initializing the key parameters of the vehicle; calculating a cabin temperature difference and a cargo compartment temperature difference by using travel information and vehicle key parameters, and establishing a personalized temperature control energy consumption prediction model to carry out personalized temperature control; designing a reference SOC prediction model based on an adaptive neural network fuzzy inference system to predict the reference SOC of the battery; defining a multi-objective optimization problem considering fuel oil, electric energy and temperature control energy consumption in combination with the personalized temperature control energy consumption prediction model and the reference SOC prediction model, and constructing an energy management model based on adaptive dynamic programming; an energy management model is applied, power distribution among power sources is adjusted in real time, and the overall energy consumption of the vehicle is optimally controlled. According to the method, the problems of personalized temperature control and real-time energy consumption optimization of multi-dimensional information fusion in a complex and strong time-varying environment of the cold chain hybrid vehicle are solved.

Description

Technical Field

[0001] The present invention relates to the field of energy consumption optimization of cold chain transportation vehicles, and in particular to an energy consumption optimization control method for cold chain hybrid vehicles based on personalized temperature control. Background Art

[0002] Cold chain transportation vehicles are facing new development requirements of green and intelligent logistics and digital upgrading. The development of intelligent transportation systems and hybrid power technologies provides new possibilities for solving the above challenges. Introducing the above technologies into the cold chain transportation field can reduce energy consumption during transportation, lower transportation costs, meet personalized temperature requirements under different transportation conditions and demands, and ensure the comfort of drivers and the quality of transported goods. Since the refrigeration and air-conditioning systems in cold chain transportation vehicles are high-energy-consuming components, and their working effects will affect the quality of goods and energy consumption costs, different drivers and transported goods have different temperature requirements for the cockpit and cargo hold. Hybrid vehicles can reasonably and efficiently operate the power, refrigeration, and air-conditioning systems by optimizing the power of multiple power sources during transportation, achieving a multi-objective optimization control effect. Therefore, optimizing the energy consumption of cold chain transportation vehicles based on personalized temperature control is crucial for meeting personalized temperature requirements, reducing the overall vehicle energy consumption, and lowering transportation costs. Currently, the energy consumption optimization control of cold chain transportation vehicles usually adopts preset performance control or fixed rule control. These methods are generally based on historical data or fixed working conditions and use partial traffic information to optimize the energy consumption of cold chain hybrid vehicles.

[0003] Chinese Patent Application No. CN114228696A discloses a real-time optimization control method for cold chain hybrid vehicles considering the energy consumption of the refrigeration system, including an SOC reference curve planning module for planning the linear decrease of SOC with the progress of the journey; a PI control equivalent factor module for dynamically adjusting the equivalent factor to ensure that the battery SOC can follow the planned SOC reference curve; a vehicle energy consumption module for the energy consumption of the refrigeration system to describe the dynamic process of the energy loss of the intelligent system during transportation; and an equivalent fuel consumption minimization control module for solving the instantaneous optimal operating point to achieve the comprehensive control of battery energy consumption and vehicle fuel consumption, realizing real-time control during vehicle operation. However, in the process of implementing the technical solution of the present invention in the embodiments of the present application, the inventors found the following technical problems in the above technology:

[0004] First, although there are methods for intelligent temperature adjustment of the cold chain transportation cargo hold in the existing technology, the transportation temperature requirements of different goods are not distinguished, resulting in a decline in the quality of the goods. More importantly, it does not comprehensively consider it into the vehicle energy consumption, causing the transportation cost to remain high. Second, the impact of the driver's personalized temperature adjustment requirements during long-distance transportation on the air-conditioning energy consumption is ignored. Third, traditional optimization control cannot solve the problems of rapid solution and dynamic adaptation of multi-dimensional information fusion non-linear systems in complex strongly time-varying environments. As a result, the energy-saving effect of the whole vehicle is not obvious, the transportation cost remains high, and the adaptive and dynamic adjustment capabilities for different transported goods and transportation conditions are poor.

[0005] Therefore, it is necessary to develop an energy consumption optimization control method for cold chain hybrid vehicles based on personalized temperature control. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an energy consumption optimization control method for cold chain hybrid vehicles based on personalized temperature control. Based on personalized temperature requirements, a personalized temperature control energy consumption prediction model is established. The reference SOC of the battery is predicted by using an adaptive neural network fuzzy inference system. By constructing a non-linear multi-objective optimization problem and adopting an adaptive dynamic programming reinforcement learning algorithm, the network parameters are adjusted in real time, and the comprehensive optimization control includes the consumption of fuel, electric energy, and temperature control system, achieving the purpose of personalized temperature control and energy consumption optimization control of cold chain hybrid vehicles.

[0007] To solve the above technical problems, the technical solution adopted by the present invention is:

[0008] An energy consumption optimization control method for cold chain hybrid vehicles based on personalized temperature control, comprising the following steps:

[0009] S1, obtaining the travel information of the cold chain hybrid vehicle including the total travel mileage and the change of ambient temperature along the line, and initializing the key parameters of the vehicle;

[0010] S2, calculating the temperature difference between the cockpit and the cargo hold during the travel of the cold chain hybrid vehicle by using the travel information and the key parameters of the vehicle, establishing a personalized temperature control energy consumption prediction model, and performing personalized temperature control;

[0011] S3, designing a reference SOC prediction model based on an adaptive neural network fuzzy inference system by using the travel information and the key parameters of the vehicle, and predicting the reference SOC of the vehicle battery;

[0012] S4, combining the personalized temperature control energy consumption prediction model and the reference SOC prediction model, defining a multi-objective optimization problem that comprehensively considers fuel, electric energy, and temperature control energy consumption, and constructing an energy management model based on adaptive dynamic programming;

[0013] S5, applying the energy management model based on adaptive dynamic programming, adjusting the power distribution between various power sources in real time, and optimizing and controlling the overall energy consumption of the vehicle.

[0014] A further improvement of the technical solution of the present invention lies in: in S1, the travel information of the cold-chain hybrid vehicle is obtained through the vehicle networking technology; the key vehicle parameters include the vehicle SOC, the cabin and cargo hold temperature, the set desired temperature of the cabin and cargo hold, and the vehicle's own parameters.

[0015] A further improvement of the technical solution of the present invention lies in: in S2, specifically, according to the ambient temperature curve along the travel route of the cold-chain hybrid vehicle and the set desired temperature of the cabin and cargo hold, calculate the temperature difference in the cabin and cargo hold during the overall travel, and design a personalized temperature control energy consumption prediction model according to the predicted cabin and cargo hold temperature difference curve to predict the energy consumption required for personalized temperature control.

[0016] A further improvement of the technical solution of the present invention lies in: the personalized temperature control energy consumption prediction model is established using a backpropagation neural network, and is offline trained using a large amount of historical energy consumption data of the cabin air conditioner and the cargo hold refrigeration system. The trained personalized temperature control energy consumption prediction model is used for online prediction;

[0017] The personalized temperature control energy consumption prediction model based on the backpropagation neural network includes an input layer, a hidden layer, and an output layer; among them, the number of input layer nodes is 2, and the input sequence includes the cabin air conditioner temperature difference sequence and the cargo hold refrigeration system temperature difference sequence; the number of hidden layer nodes is 5; the number of output layer nodes is 1, and the output sequence is the instantaneous energy consumption of the temperature control system;

[0018] During the online prediction process, the vehicle initialization information and the cabin and cargo hold temperature difference curve are obtained, and the instantaneous energy consumption of the temperature control system is predicted in real time, and the overall energy consumption curve of the cold-chain hybrid vehicle travel is obtained by cumulative calculation.

[0019] A further improvement of the technical solution of the present invention lies in: in S3, specifically, considering the influence of the uncertain driving conditions of the cold-chain hybrid vehicle on energy consumption, using the historical optimal SOC data and part of the current traffic information as inputs, and offline training the reference SOC prediction model using the adaptive neural network fuzzy inference system to obtain an approximate optimal adaptive reference SOC curve;

[0020] The reference SOC prediction model is a system with four inputs and one output. The four inputs are the current driving distance D, the current driving speed V, the current instantaneous energy consumption of the temperature control system, and the optimal SOC at the previous step under the current training condition. One output is the current optimal SOC under this training condition. The optimal SOC is obtained by global optimization calculation using the DP algorithm during the operation of the training condition.

[0021] Based on the reference SOC prediction model, the vehicle control system can obtain an approximate optimal adaptive reference SOC according to the current traffic information during online operation.

[0022] A further improvement of the technical solution of the present invention lies in: In S4, specifically, it is necessary to consider the performance of SOC tracking the reference SOC to ensure approximate optimality. Therefore, the objective function of the optimization problem is defined as:

[0023]

[0024] In the formula, t end is the travel end time, is the instantaneous fuel consumption rate, P mot (t) and P temp (t) are the current motor power and the power of the temperature control system respectively, H l is the lower calorific value of fuel, SOC(t) and SOC ref (t) are the current SOC value and the reference SOC value respectively, and λ, μ, ω are weight conversion coefficients;

[0025] In the adaptive dynamic programming algorithm, it is necessary to discretize the parameters for optimization training. Therefore, the objective function is rewritten as:

[0026]

[0027] In the formula, γ is a discount factor in (0, 1), Δt is a control time domain step of the optimization algorithm, and k ∈ [t 0 , t end -1].

[0028] A further improvement of the technical solution of the present invention lies in: The constraint conditions of the optimization problem are:

[0029]

[0030] A further improvement of the technical solution of the present invention lies in: In S5, specifically, the energy management model based on adaptive dynamic programming includes an action network AN, a dynamic model, and two evaluation networks CN1 and CN2; among them, the action network AN and the evaluation networks CN1 and CN2 are constructed by a backpropagation neural network; after the energy management model based on adaptive dynamic programming is offline trained, it is loaded into the vehicle controller for online application. Each time a sample is taken, the inputs required for the control strategy include external traffic information and the vehicle's own state to achieve real-time adjustment of the power distribution of each power source.

[0031] A further improvement of the technical solution of the present invention lies in: During the training process of CN1 and CN2 in the energy management model based on adaptive dynamic programming, the actual value is close to the target value Therefore, the error functions of CN1 and CN2 are the difference between the approximate target value and the actual value, denoted as:

[0032]

[0033] Converge the error function to 0, so let

[0034]

[0035] where ε c is an error value close to 0 and E c , and use the gradient descent algorithm to train the network weights;

[0036] Similarly, the training objective of AN is to find the optimal control value Combined with the iterative update of network parameters until the optimal control is obtained;

[0037] Use the optimal search method to find the optimal control value:

[0038]

[0039] where ε a is an error value very close to 0.

[0040] Due to the adoption of the above technical solution, the technical progress achieved by the present invention is:

[0041] 1. An energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control provided by the present invention, by establishing a personalized temperature control energy consumption prediction model, comprehensively considering the energy consumption of personalized temperature control, and intelligently adjusting the cabin and cargo hold temperatures on the premise of ensuring the quality of goods and the comfort of drivers, while ensuring the temperature control effect of the cabin and cargo hold, optimizing the energy consumption of the cold chain hybrid vehicle, and reducing the cost of goods transportation.

[0042] 2. An energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control provided by the present invention, based on the vehicle networking system to obtain the temperature information along the way of the cold chain hybrid vehicle in real time, predict the personalized temperature control energy consumption of the cabin and cargo hold, and fully consider it into the energy consumption optimization problem of the hybrid vehicle, reasonably predict the reference SOC and implement the optimization of the overall energy consumption including fuel, electric energy, and temperature control system.

[0043] 3. An energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control provided by the present invention, based on considering the characteristics of high dynamics and strong randomness of the external environment, uses reinforcement learning and artificial intelligence methods such as adaptive dynamic programming and neural networks for multi-objective optimization, effectively solving the problems of rapid solution and real-time adaptation of multi-dimensional information fusion in complex and strongly time-varying environments. Brief Description of the Drawings

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings;

[0045] Figure 1 It is the control flowchart of an energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control provided in the embodiment of the present invention;

[0046] Figure 2 It is the overall architecture schematic diagram of an energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control provided in the embodiment of the present invention. Detailed implementation manners

[0047] It should be noted that the terms "including" and "having" and any variations thereof in the description and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0048] The following will further elaborate on the present invention in conjunction with the drawings and embodiments:

[0049] As Figure 1 shown, an energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control includes the following steps:

[0050] S1. Obtain the travel information of the cold chain hybrid vehicle including the total transportation mileage and the change of ambient temperature along the line, and initialize the key vehicle parameters;

[0051] Specifically, obtain the travel information such as the total transportation mileage of the cold chain hybrid vehicle and the change of ambient temperature along the line through vehicle networking technology; the key vehicle parameters are the vehicle's SOC (the full English name is State of charge, and the Chinese is the state of charge), the actual temperatures of the cockpit and the cargo hold, the set desired temperatures of the cockpit and the cargo hold, and the vehicle's own parameters, etc. The above-obtained information and parameter initialization are for real-time optimization control applications.

[0052] S2. Calculate the temperature differences in the cockpit and the cargo hold during the travel of the cold chain hybrid vehicle using the travel information and the key vehicle parameters, establish a personalized temperature control energy consumption prediction model, and perform personalized temperature control;

[0053] Specifically, according to the ambient temperature curve along the itinerary of the cold-chain hybrid vehicle and the desired temperature set for the cockpit and cargo hold, calculate the temperature difference between the cockpit and cargo hold during the overall itinerary. Based on the predicted temperature difference curve between the cockpit and cargo hold, design a personalized temperature control (cockpit air conditioner and cargo hold refrigeration system) energy consumption prediction model to predict the energy consumption required for personalized temperature control, so as to be further considered in the overall vehicle energy consumption optimization control.

[0054] Furthermore, the personalized temperature control energy consumption prediction model is established using a backpropagation neural network and is offline trained using a large amount of historical energy consumption data of the cockpit air conditioner and cargo hold refrigeration system. The trained personalized temperature control energy consumption prediction model is used for online prediction. The personalized temperature control energy consumption prediction model based on the backpropagation neural network includes an input layer, a hidden layer, and an output layer; among them, the number of nodes in the input layer is 2, and the input sequence includes the cockpit air conditioner temperature difference sequence and the cargo hold refrigeration system temperature difference sequence; the number of nodes in the hidden layer is 5; the number of nodes in the output layer is 1, and the output sequence is the instantaneous energy consumption of the temperature control system. During the online prediction process, vehicle initialization information and the temperature difference curve between the cockpit and cargo hold are obtained, and the instantaneous energy consumption of the temperature control system is predicted in real time. By cumulative calculation, the overall energy consumption curve of the cold-chain hybrid vehicle itinerary is obtained.

[0055] S3. Design a reference SOC prediction model based on an adaptive neural network fuzzy inference system using the itinerary information and vehicle key parameters to predict the reference SOC of the vehicle battery;

[0056] Specifically, considering the impact of the uncertain driving conditions of the cold-chain hybrid vehicle on energy consumption, using the historical optimal SOC data and part of the current traffic information as inputs, the reference SOC prediction model is offline trained using an adaptive neural network fuzzy inference system to obtain an approximately optimal adaptive reference SOC curve. Among them, the reference SOC prediction model is a system with four inputs and one output. The four inputs are the current driving distance D, the current driving speed V, the current instantaneous energy consumption of the temperature control system, and the previous optimal SOC under the current training condition. One output is the current optimal SOC under this training condition. The optimal SOC is obtained by global optimization calculation using the DP algorithm during the operation of the training condition. Based on this reference SOC prediction model, the vehicle control system can obtain an approximately optimal adaptive reference SOC according to the current traffic information during online operation.

[0057] S4. Combine the personalized temperature control energy consumption prediction model and the reference SOC prediction model, define a multi-objective optimization problem that comprehensively considers fuel, electric energy, and temperature control energy consumption, and construct an energy management model based on adaptive dynamic programming;

[0058] Specifically, it is necessary to consider the performance of SOC tracking the reference SOC to ensure approximate optimality. Therefore, the objective function of the optimization problem is defined as:

[0059]

[0060] In the formula, t end is the end time of the journey, is the instantaneous fuel consumption rate, P mot (t) and P temp (t) are the current motor power and the power of the temperature control system respectively, H l is the low calorific value of fuel, SOC(t) and SOC ref (t) are the current SOC value and the reference SOC value respectively, and λ, μ, ω are weight conversion coefficients.

[0061] In the adaptive dynamic programming algorithm, the parameters need to be discretized for optimization training. Therefore, the objective function is rewritten as:

[0062]

[0063] In the formula, γ is a discount factor in (0, 1), Δt is a control time domain of the optimization algorithm, and k ∈ [t 0 , t end - 1].

[0064] The constraint conditions of the optimization problem are:

[0065] SOC min ≤ SOC(t) ≤ SOC max

[0066] P motmin ≤ P mot (t) ≤ P motmax

[0067] P tempmin ≤ P temp (t) ≤ P tempmax

[0068] S5. Apply the energy management model based on adaptive dynamic programming to adjust the power distribution between power sources in real time and optimize the overall energy consumption of the vehicle.

[0069] Specifically, the energy management model based on adaptive dynamic programming includes an action network AN, a dynamic model, and two evaluation networks CN1 and CN2; among them, the action network AN and the evaluation networks CN1 and CN2 are constructed by a backpropagation neural network. After the energy management model based on adaptive dynamic programming is trained offline, it is loaded into the vehicle controller for online application. Each time a sample is taken, the inputs required for the control strategy include external traffic information and the vehicle's own state to achieve real-time adjustment of the power distribution of each power source.

[0070] Furthermore, during the training process of CN1 and CN2 in the energy management model based on adaptive dynamic programming, the actual value is close to the target value Therefore, the error functions of CN1 and CN2 are the difference between the approximate target value and the actual value, denoted as:

[0071]

[0072] It is necessary to make the error function converge to 0, so let

[0073]

[0074] where ε c is an error value close to 0 and E c , and the gradient descent algorithm is used to train the network weights. Similarly, the training objective of AN is to find the optimal control value Combined with the iterative update of network parameters until the optimal control is obtained. The optimal search method is used to find the optimal control value:

[0075]

[0076] where ε a is an error value very close to 0.

[0077] To better implement the above method, as Figure 2 shown, the present invention further describes the overall architecture of an energy consumption optimization control method for a cold chain hybrid vehicle based on personalized temperature control as follows:

[0078] (1) The energy consumption optimization control method obtains the required power of the vehicle power system and the trip network connection information according to the driving cycle condition, and offline trains three parts. One is the reference SOC prediction model based on the adaptive neural network fuzzy inference system, which is a four-input and one-output model; the second is the personalized temperature control energy consumption prediction model based on the backpropagation neural network, which is offline trained using a large amount of historical energy consumption data of the air conditioner and refrigeration system; the third is the energy management model based on adaptive dynamic programming, which constructs the action network AN and the evaluation networks CN1 and CN2 using the backpropagation neural network, and is loaded into the vehicle controller for online application after offline training.

[0079] (2) In the energy consumption optimization control method, a multi-objective optimization problem considering fuel, electric energy, and temperature control energy consumption is defined. At the same time, the performance of SOC tracking the reference SOC needs to be considered to ensure approximate optimality. The objective function of the optimization problem is defined to include four performance indicators, and an energy management model based on adaptive dynamic programming is constructed accordingly.

[0080] (3) After the offline training of the reference SOC prediction model, the personalized temperature control energy consumption prediction model, and the energy management model in the energy consumption optimization control method, they are loaded into the vehicle controller for online application. When the reference SOC prediction model runs online, the control system can obtain an approximate optimal reference SOC based on the current traffic information; during the online prediction process of the personalized temperature control energy consumption prediction model, the vehicle initialization information and the cabin-cargo hold temperature difference curve are obtained to predict the instantaneous energy consumption of the temperature control system in real time; when the energy management model is applied online, the required inputs include external traffic information and the vehicle's own state to achieve real-time adjustment of the power distribution of each power source and optimize the overall energy consumption control of the vehicle.

[0081] In summary, through the connected vehicle technology, the embodiments of the present invention obtain the transportation task and the environmental temperature change information, establish a personalized temperature control energy consumption prediction model, perform personalized temperature control, design a reference SOC prediction model based on an adaptive neural network fuzzy inference system to predict the reference SOC of the vehicle battery, define a multi-objective optimization problem that comprehensively considers fuel, electric energy, and temperature control energy consumption, and use an energy management model based on adaptive dynamic programming for multi-objective optimization, thereby realizing the energy consumption optimization of the cold chain hybrid vehicle under personalized temperature control, reducing the cargo transportation cost, and effectively solving the problems of fast solution and real-time adaptation of multi-dimensional information fusion in a complex and strongly time-varying environment in the prior art.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing energy consumption of cold chain hybrid vehicles based on personalized temperature control, characterized in that: The following steps are involved: S1, obtaining the trip information of the cold chain hybrid vehicle including the total transportation mileage and the change of ambient temperature along the route, and initializing the key parameters of the vehicle; S2, using the trip information and key vehicle parameters to calculate the cabin temperature difference and cargo compartment temperature difference during the cold chain hybrid vehicle trip, establish a personalized temperature control energy consumption prediction model, and perform personalized temperature control; S3, using the trip information and key vehicle parameters to design a reference SOC prediction model based on an adaptive neural network fuzzy inference system to predict the reference SOC of the vehicle battery; S4, combining the personalized temperature control energy consumption prediction model and the reference SOC prediction model, defines a multi-objective optimization problem that comprehensively considers fuel, electricity and temperature control energy consumption, and constructs an energy management model based on adaptive dynamic programming; S5 uses an energy management model based on adaptive dynamic programming to adjust the power distribution between various power sources in real time and optimize the overall energy consumption of the vehicle.

2. According to claim 1, a cold chain hybrid vehicle energy consumption optimization control method based on personalized temperature control is characterized in that: In S1, the travel information of the cold chain hybrid vehicle is obtained through the Internet of Vehicles technology; the key vehicle parameters include the vehicle SOC, the cabin and cargo compartment temperature, the cabin and cargo compartment set expected temperature and the vehicle's own parameters.

3. The method for optimizing the energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 1 is characterized in that: In S2, it specifically includes calculating the temperature difference of the cabin and cargo compartment during the overall journey according to the ambient temperature curve along the journey of the cold chain hybrid vehicle and the expected temperature set in the cabin and cargo compartment, designing a personalized temperature control energy consumption prediction model according to the predicted cabin and cargo compartment temperature difference curve, and predicting the energy consumption required for personalized temperature control.

4. The method for optimizing the energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 3 is characterized in that: The personalized temperature control energy consumption prediction model is established by using a back propagation neural network, and a large amount of historical energy consumption data of the cabin air conditioning and cargo compartment refrigeration systems are used for offline training. The trained personalized temperature control energy consumption prediction model is used for online prediction; The personalized temperature control energy consumption prediction model based on back propagation neural network includes input layer, hidden layer and output layer; the number of nodes in the input layer is 2, and the input sequence includes the cabin air conditioning temperature difference sequence and the cargo hold refrigeration system temperature difference sequence; the number of nodes in the hidden layer is 5; the number of nodes in the output layer is 1, and the output sequence is the instantaneous energy consumption of the temperature control system; During the online prediction process, the vehicle initialization information and the cabin-cargo compartment temperature difference curve are obtained, the instantaneous energy consumption of the temperature control system is predicted in real time, and the overall energy consumption curve of the cold chain hybrid vehicle is obtained by cumulative calculation.

5. The method for optimizing energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 1 is characterized in that: Specifically, in S3, considering the impact of uncertain driving conditions of cold chain hybrid vehicles on energy consumption, the historical optimal SOC data and part of the current traffic information are used as inputs, and the reference SOC prediction model is trained offline using an adaptive neural network fuzzy inference system to obtain an approximately optimal adaptive reference SOC curve; The reference SOC prediction model is a four-input and one-output system, wherein the four inputs are the current driving distance D, the current driving speed V, the instantaneous energy consumption of the current temperature control system and the optimal SOC of the previous step under the current training condition, and the one output is the current optimal SOC under this training condition, and the optimal SOC is obtained by global optimization calculation using the DP algorithm when the training condition is running; Based on the reference SOC prediction model, the control system can obtain the approximately optimal adaptive reference SOC according to the current traffic information when the vehicle is running online.

6. The method for optimizing the energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 1 is characterized in that: In S4, specifically, it is necessary to consider the performance of SOC tracking the reference SOC to ensure approximate optimality, so the objective function of the optimization problem is defined as: Where, t end The end time of the trip. is the instantaneous fuel consumption rate, P mot (t) and P temp (t) are the current motor power and temperature control system power, H l is the lower calorific value of fuel, SOC(t) and SOC ref (t) are the current SOC value and the reference SOC value, respectively, λ, μ, ω are weight conversion coefficients; In the adaptive dynamic programming algorithm, the parameters need to be discretized for optimization training, so the objective function is rewritten as: Where γ is a discount factor in (0, 1), and Δt is a control time domain step size k∈[t0,t end -1].

7. The method for optimizing the energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 6 is characterized in that: The constraints of the optimization problem are:

8. The method for optimizing the energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 1 is characterized in that: In S5, specifically, the energy management model based on adaptive dynamic programming includes an action network AN, a dynamic model and two evaluation networks CN1 and CN2; wherein, the action network AN and the evaluation networks CN1 and CN2 are constructed through back propagation neural networks; after the offline training of the energy management model based on adaptive dynamic programming is completed, it is loaded into the vehicle controller for online application, and the input required for the control strategy at each sampling includes external traffic information and the vehicle's own status, so as to realize real-time adjustment of the power distribution of each power source.

9. The method for optimizing the energy consumption of cold chain hybrid vehicles based on personalized temperature control according to claim 8 is characterized in that: In the energy management model based on adaptive dynamic programming, the actual value of CN1 and CN2 is used in the training process Close to target value Therefore, the error function of CN1 and CN2 is the difference between the approximate target value and the actual value, which is expressed as: Make the error function converge to 0, so let In the formula, ε c is close to 0 and E c The error value is used to train the network weights using the gradient descent algorithm; Similarly, the training goal of AN is to find the optimal control value Combined with iterative update of network parameters, until optimal control is obtained; Use the best search method to find the optimal control value: In the formula, ε a is an error value very close to 0.

Images