Method for controlling air cooling flow based on DDPG algorithm to perform thermal management on motor of electric power-assisted two-wheeled vehicle

Abstract

This invention discloses a thermal management method and system for electric power-assisted two-wheeler motors based on the DDPG algorithm to control air cooling flow, belonging to the field of thermal management technology for electric vehicles. The method includes: initializing the system during startup, loading basic motor parameters and DDPG algorithm parameters; real-time acquisition of data such as motor temperature, ambient temperature, and load power to construct a state vector; dynamically correcting the efficiency-temperature curve; inputting the state vector into the DDPG algorithm, calculating a reward value based on the difference between energy saving and energy consumption adjustment, and temperature fluctuations; outputting the optimal target temperature through experience playback and soft updates; and having the temperature adjustment device execute actions and periodically calibrate the curve to form a closed loop. This invention, using the above method, enables the motor to always operate in the high-efficiency range with the lowest overall energy consumption, optimizing energy consumption to extend the driving range, reducing motor losses to extend service life, and exhibiting a high degree of intelligence, reducing manual intervention.

Description

Technical Field

[0001] This invention relates to the field of thermal management technology for electric vehicles, specifically to a method and system for thermal management of the motor temperature of an electric-assisted two-wheeler that integrates motor efficiency-temperature characteristics and a deep deterministic strategy gradient (DDPG) algorithm, in order to optimize the energy consumption of the motor and improve working efficiency and driving range. Background Technology

[0002] In the field of thermal management technology for electric vehicles, electric-assisted two-wheelers have become an important choice for short-distance travel due to their flexibility, convenience, and low-carbon environmental friendliness. The energy consumption and efficiency of their motors directly affect the driving range, making them a core concern for users. The efficiency of electric-assisted two-wheeler motors has a significant characteristic relationship with their temperature, i.e., an efficiency-temperature characteristic curve. However, the riding conditions of electric-assisted two-wheelers are complex, with frequent starts and stops, load changes (such as climbing hills or varying loads), and fluctuations in ambient temperature. These factors can easily cause the motor temperature to deviate from its optimal range, resulting in energy waste, reduced efficiency, and shortened driving range.

[0003] Existing thermal management technologies for electric-assisted two-wheeler motors mostly employ simple heat dissipation or heating methods, failing to fully integrate the motor's efficiency-temperature characteristics for precise control. Furthermore, these technologies struggle to dynamically balance the energy consumed during temperature regulation with the energy saved due to improved motor efficiency, often resulting in excessive energy consumption for temperature control, thus negatively impacting overall energy efficiency.

[0004] The Deep Deterministic Policy Gradient (DDPG) algorithm has significant advantages in dynamic optimization of continuous action spaces, enabling it to autonomously find the optimal strategy by learning complex dynamic characteristics. However, currently, there is no technical solution that integrates motor efficiency-temperature characteristics with the DDPG algorithm for thermal management of electric-assisted two-wheeler motors to optimize energy consumption, improve work efficiency, and extend driving range.

[0005] To address the above problems, this invention proposes a method and system for thermal management of electric-assisted two-wheeler motor temperature based on the DDPG algorithm. By comprehensively evaluating energy savings, the optimal temperature is determined, achieving the goal of overall energy saving and constructing an efficient, energy-saving, and environmentally friendly thermal management system for electric-assisted two-wheelers. Summary of the Invention

[0006] This invention provides a method and system for thermal management of electric-assisted two-wheeled vehicle motors based on the DDPG algorithm. By updating the motor efficiency-temperature characteristic curve in real time, the optimal temperature and efficiency under the current operating conditions are determined to optimize the motor's range and working efficiency.

[0007] To achieve the above objectives, the present invention provides an active thermal management method for improving the operating efficiency of an electric motor, comprising the following steps: S1. During the start-up phase of the electric-assisted two-wheeler, the thermal management system first completes the initialization operation. It loads the basic parameters of the motor, including rated power, heat capacity, heat dissipation coefficient, and basic efficiency-temperature characteristic curves; at the same time, it initializes the DDPG algorithm module and the maximum adjustable power range of the temperature regulation device (such as the cooling fan). S2. Collect key data in real time through sensors: current motor temperature, ambient temperature, real-time load power, and real-time speed. After data preprocessing, construct the current state vector. S3. Based on the collected real-time data and the operating conditions of the motor, the initial efficiency-temperature characteristic curve is dynamically corrected to ensure that the curve accurately reflects the change of motor efficiency with temperature under the current operating conditions, providing a basis for subsequent determination of the optimal temperature range. S4. Input the real-time constructed state vector into the DDPG algorithm, and combine it with the reward value in the current state (the reward value is calculated based on the difference between energy saving from motor efficiency improvement and energy consumption from temperature regulation) and temperature fluctuations. Calculate the optimal target temperature in the current state through an experience replay mechanism and soft updates of the target network. S5. The temperature control device executes corresponding adjustment actions based on the optimal target temperature calculated by the DDPG algorithm. During the adjustment process, the sensor continuously collects changes in motor temperature and related parameters, and feeds the new status data back to the system, forming a closed-loop control. S6. At the same time, based on factors such as motor aging and long-term changes in riding conditions, the efficiency-temperature characteristic curve is regularly calibrated to ensure that the thermal management strategy can adapt to the long-term operating characteristics of the motor, always optimize the energy consumption of the motor, and improve work efficiency and range.

[0008] An active thermal management system for improving motor efficiency and energy utilization includes... The system initialization and data acquisition module is responsible for parameter configuration during system startup and data sensing during operation. It loads basic motor parameters (rated power, thermal capacity, etc.) and initial parameters for the DDPG algorithm (network weights, experience pool capacity, etc.); and collects data such as motor temperature, ambient temperature, load power, and speed in real time through temperature sensors, current / voltage sensors, and speed sensors, and performs preprocessing (noise reduction and standardization).

[0009] The temperature characteristic curve dynamic correction module dynamically optimizes the correlation model between motor efficiency and temperature. Based on real-time collected efficiency data and corresponding temperatures, it continuously corrects the initial efficiency-temperature curve to ensure that the curve reflects the true characteristics under current operating conditions, providing a basis for determining the optimal temperature.

[0010] The DDPG algorithm decision module generates the optimal temperature regulation strategy. It includes an Actor network (outputting continuous regulation actions, such as heating / cooling power) and a Critic network (evaluating the benefits of actions). Combining the state vector and efficiency-temperature curve, it calculates the optimal regulation action that balances energy consumption and energy saving through empirical replay and soft updates of the target network, and outputs the target power command.

[0011] The temperature regulation execution module executes the temperature regulation actions determined by the algorithm.

[0012] The feedback optimization module feeds back the adjusted motor status data to the data acquisition module. At the same time, based on motor aging and long-term operating condition changes, it periodically calibrates the efficiency-temperature curve and continuously optimizes the DDPG algorithm strategy to ensure the long-term energy-saving effect of the system.

[0013] The DDPG control algorithm includes: The Actor network takes the current state of the motor as input and outputs continuous temperature regulation actions, realizing a direct mapping from state to action; The Critic network takes the current state and the actions output by the Actor network as input and evaluates the long-term benefits (i.e., the overall energy-saving effect) of the state-action group, providing a basis for optimizing the Actor network. The target Actor network and the target Critic network have the same structure as the main Actor / Critic network, respectively, and are used to stabilize the training process. They reduce training fluctuations through a "soft update" mechanism (slowly synchronizing the main network parameters).

[0014] Meanwhile, DDPG introduces an experience replay mechanism, storing the "state-action-reward-next state" data generated during motor operation in an experience pool. During training, data is randomly sampled to update the network, avoiding the impact of sample correlation on model convergence.

[0015] Preferably, in S3, the optimal temperature is calculated during riding based on real-time collected data and the efficiency-temperature curve, specifically as follows: Receive motor status data collected by sensors and construct a state vector; The Actor network outputs an initial adjustment action based on the current state; After an action is executed, the benefits of the action are evaluated through a Critic network, the reward generated by the action (overall energy-saving effect) is calculated, and the new motor state is obtained. Store the "Status-Action-Reward-New Status" data into the experience replay pool; Data is randomly sampled from the experience pool, and the target network is softly updated simultaneously. Repeat the above process to allow the algorithm to gradually learn the optimal temperature regulation strategy and achieve dynamic optimization of motor energy consumption.

[0016] The state vector includes T motor (Current motor temperature); T env (Ambient temperature); P load (Current load power of the motor); or current (Current motor efficiency); or opt (Optimal efficiency under current load); T opt (Optimal efficiency temperature under current load).

[0017] The action space is defined as continuous action, corresponding to the output power of the temperature control device.

[0018] Preferably, adding OU noise to the Action encourages exploration.

[0019] The reward function is: R = R save - R reg - R fluct It is used to calculate whether the current action has achieved the energy-saving effect.

[0020] in: Energy saving benefit item R save This reflects the energy savings resulting from the increased efficiency of the motor under the current adjustment. R save = k 1×( P load / or current - P load / or opt )×Δ t In the formula: P load / or current Input power at the current efficiency; P load / or opt : Input power at optimal efficiency; Δ t Adjustment cycle; k 1: Energy saving benefit coefficient (weight).

[0021] like or current → or opt ,but R save Approaching 0; if the current efficiency is lower than optimal, then R save For positive rewards, the larger the difference, the higher the reward.

[0022] Adjusting energy consumption penalty R reg This reflects the energy consumption of temperature regulation itself. R reg = k 2×∣ a ∣×Δ t In the formula: | a |: Adjust the absolute value of the power (power difference of the cooling water pump); k 2: Adjust the energy consumption penalty coefficient (weight, needs to be adjusted with...) k 1. Maintain balance, avoiding over-regulation or under-regulation.

[0023] Temperature fluctuation penalty R fluct Avoid frequent and large temperature fluctuations (which can affect motor lifespan and stability). R fluct = k 3×∣ T motor ( t )- T motor ( t -1)∣where: k 3: Fluctuation penalty coefficient; the more drastic the temperature change, the greater the penalty.

[0024] The soft update of the target network includes calculating the parameters of the target neural network; The Actor target network is updated using the gradient ascent method, where the gradient is... ; The Critit target network is updated using gradient descent, where the gradient is... ; Update the parameters of the target neural network; Actor Target Network: ; Critict target network: .

[0025] Preferably, the S5 includes a feedback optimization module that continuously iterates and optimizes the DDPG model parameters during riding, and updates the control strategy based on real-time feedback data.

[0026] An electronic device includes sensors and a memory. The sensors include a temperature sensor for acquiring real-time motor temperature and ambient temperature; a current sensor and a voltage sensor for calculating real-time load power of the motor; and a speed sensor for acquiring motor speed.

[0027] Therefore, the present invention employs the above-mentioned active thermal management method, system, electronic device, and storage medium for optimizing the driving range and working efficiency of a motor during operation, and the technical effects are as follows: (1) Optimize energy consumption and extend driving range: The DDPG algorithm dynamically balances temperature regulation energy consumption and efficiency improvement energy saving, and combines the motor efficiency-temperature characteristic curve to achieve precise control, so that the motor always operates in the high-efficiency range with the lowest overall energy consumption. Compared with the traditional method, it not only reduces energy waste caused by blind adjustment, but also reduces energy consumption per unit distance, and significantly extends the driving range with the same battery capacity.

[0028] (2) Reduce motor losses and extend service life: Dynamic temperature management avoids the motor being in a low-temperature and inefficient state or a high-temperature and overload state for a long time, reduces problems such as winding aging and core loss, and stable operating temperature reduces component fatigue damage, extends motor life and reduces maintenance costs.

[0029] (3) High level of intelligence and reduced manual intervention: The system automatically collects data and makes autonomous decisions on adjustment strategies through algorithms to achieve full automation of thermal management, thereby improving the intelligence level and ease of use of the vehicle. Attached Figure Description

[0030] Figure 1 This is an overall flowchart of an active thermal management method for optimizing the driving range and working efficiency of an electric motor during operation, according to the present invention. Figure 2 This is a structural diagram of an active thermal management method for optimizing the driving range and working efficiency of an electric motor during operation, according to the present invention. Detailed Implementation

[0031] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0032] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0033] Example 1 like Figure 1 As shown, the present invention discloses an active thermal management method for improving motor efficiency during motor operation, comprising the following steps: S1. When the electric-assisted two-wheeler starts, the thermal management system loads the basic parameters of the motor (rated power, heat dissipation coefficient, heat capacity, initial efficiency-temperature characteristic curve), completes the initialization and self-test of the DDPG algorithm module (Actor / Critic network), temperature sensor (accuracy ±0.5℃), current / voltage sensor (to calculate load power), and speed sensor, with an experience pool capacity of 10,000 entries, initial weights of the main / target network (random initialization), and a learning rate of 0.001.

[0034] S2. During the ride, the sensor collects data every 10 seconds and preprocesses it (denoise removal, standardization) to construct a state vector.

[0035] S3. The motor temperature is collected. The current efficiency is lower than the initial curve. It is determined that the initial curve deviates from the actual working conditions. Based on the real-time collected "temperature-efficiency" data, the initial curve is corrected.

[0036] The S4 and DDPG algorithms generate the optimal temperature solution based on the updated curve, and the temperature adjustment module executes the adjustment to the optimal temperature.

[0037] S5. After long-term riding, the efficiency-temperature curve is deeply calibrated, and the experience pool and network parameters of the DDPG algorithm are updated simultaneously to ensure that the strategy adapts to the aging characteristics of the motor and can still maintain optimal energy consumption.

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for thermal management of an electric-assisted two-wheeled vehicle motor based on the DDPG algorithm to control air cooling flow, characterized in that, Includes the following steps: S1. System initialization: Load the rated power, heat capacity, heat dissipation coefficient, and initial efficiency-temperature characteristic curve parameters of the motor, and initialize the DDPG algorithm module and temperature, current, voltage, and speed sensors; S2. Data Acquisition and Processing: Real-time acquisition of motor current temperature, ambient temperature, load power and speed data; after preprocessing the data, constructing a state vector containing the motor current temperature, ambient temperature, load power, current efficiency, optimal efficiency under current load and optimal efficiency temperature; S3. Dynamically correct the efficiency-temperature characteristic curve: Based on the real-time collected motor efficiency and temperature data, the initial efficiency-temperature characteristic curve is dynamically corrected to reflect the current actual working conditions. S4, DDPG algorithm decision: The state vector is input into the DDPG algorithm module, and the reward value calculated based on energy saving benefits, energy consumption adjustment penalties and temperature fluctuation penalties is combined with the experience playback and target network soft update mechanism to output the optimal target temperature and the corresponding action command of the temperature regulation device. S5. Execution and Feedback: The temperature regulation device executes the action command to regulate the motor temperature and feeds back the regulated motor status data to the system, forming a closed-loop control. S6. Periodic calibration: Based on motor aging and long-term operating condition changes, periodically perform in-depth calibration on the efficiency-temperature characteristic curve and DDPG algorithm strategy.

2. The method according to claim 1, characterized in that, The DDPG algorithm module mentioned in step S4 includes an Actor network, a Critic network, a target Actor network, and a target Critic network; The Actor network is used to output continuous temperature regulation actions based on the current state vector; The Critic network is used to evaluate the long-term benefits of state-action pairs; The target Actor network and the target Critic network stabilize the training process through a soft update mechanism.

3. The method according to claim 1 or 2, characterized in that, The formula for calculating the reward value R in step S4 is as follows: R = R save - R reg - R fluct ; Among them, R save The value of the energy-saving benefit is... k 1×( P load / η current - P load / η opt )×Δ t k1 is the energy-saving benefit coefficient, P load For load power, η current For the current efficiency, η opt For optimal efficiency, Δt is the adjustment period; R reg To adjust the energy consumption penalty, its value is k2×|a|×Δt, where k2 is the energy consumption penalty coefficient and |a| is the absolute value of the adjusted power; R fluct The temperature fluctuation penalty is calculated as k3×|Tmotor(t)-Tmotor(t-1)|, where k3 is the fluctuation penalty coefficient and T is the temperature fluctuation penalty coefficient. motor (t) represents the motor temperature at the current moment, T motor (t-1) represents the motor temperature at the previous moment.

4. The method according to claim 1, characterized in that, The experience replay mechanism described in step S4 is as follows: the "state-action-reward-next state" data generated during motor operation is stored in the experience pool, and data is randomly sampled from the experience pool during training to update the network.

5. The method according to claim 1, characterized in that, The action command mentioned in step S4 is a continuous power output signal applied to the air-cooled flow control device.

6. The method according to claim 5, characterized in that, OU noise is introduced when generating the action instructions to encourage exploration.

7. The method according to claim 1, characterized in that, The data acquisition frequency in step S2 is once every 10 seconds.

8. An active thermal management system for implementing the method according to any one of claims 1-7, characterized in that, include: The system initialization and data acquisition module is used to load system parameters and acquire motor operating status data in real time. The dynamic temperature characteristic curve correction module is used to dynamically correct the efficiency-temperature characteristic curve of the motor based on real-time data. The DDPG algorithm decision module is used to calculate and output the optimal temperature regulation action command based on the current state vector and the corrected efficiency-temperature characteristic curve. The temperature regulation execution module is used to execute the action command and control the temperature regulation device to work; The feedback optimization module is used to form closed-loop feedback control and periodically perform in-depth calibration of the system.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method as described in any one of claims 1-7. The electronic device according to claim 9 is characterized in that it further includes a temperature sensor, a current sensor, a voltage sensor and a speed sensor for collecting motor operating status data.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.

Images