Series hybrid tracked vehicle energy management method based on predicted vehicle speed changes

Abstract

The application discloses a series hybrid tracked vehicle energy management strategy based on predicted vehicle speed change. It comprises the following steps: (1) establishing a longitudinal dynamics model of the series hybrid tracked vehicle; (2) establishing an A-ECMS energy management strategy model for the series hybrid tracked vehicle; (3) predicting the speed change value by using the LSTM method, dividing the change value into two categories of positive and negative, respectively adopting the Kmeans method to classify the predicted speed change values under different symbols, and obtaining the threshold value; (4) for the speed change value whose absolute value is greater than the threshold value under the corresponding symbol, adjusting the equivalent factor of the corresponding A-ECMS, and comparing the obtained comprehensive fuel consumption with the comprehensive fuel consumption obtained by the A-ECMS, and taking the smaller value corresponding to the APU and the power battery power distribution. The application improves the matching of the device output power and the demand power according to the predicted vehicle speed, and simultaneously realizes the further reduction of the comprehensive fuel consumption value.

Description

Technical Field

[0001] This invention belongs to the field of energy management for series hybrid tracked vehicles, specifically relating to an energy management method for series hybrid tracked vehicles based on predicted vehicle speed changes. Background Technology

[0002] Vehicles powered by traditional energy sources have drawbacks such as high pollution and high fuel consumption. Series hybrid vehicles use an engine and a power battery as power sources, which can reduce pollution and fuel consumption.

[0003] Common energy management strategies include rule-based energy management strategies and optimization-based energy management strategies. While rule-based energy management strategies are easy to implement, they rely too heavily on experience and have poor adaptability to operating conditions. Optimization-based energy management strategies are divided into global optimization strategies and instantaneous optimization strategies, which optimize for specific objectives but have longer computation times.

[0004] The equivalent fuel consumption minimization strategy can be applied in real time during vehicle operation. By solving for the optimal solution in real time, it can achieve the goal of finding the optimal solution for the entire simulation condition, which has great practicality.

[0005] During training, the original RNN is prone to gradient explosion or gradient vanishing problems as the training time increases and the number of network layers increases, making it unable to process long sequence data and thus unable to obtain information from long-distance data. Summary of the Invention

[0006] The purpose of this invention is to provide an energy management strategy for a series hybrid tracked vehicle based on predicted vehicle speed changes. This strategy combines LSTM and A-ECMS, further adjusts the equivalent factor based on predicted vehicle speed changes, and determines the final equivalent factor based on overall fuel consumption. This improves the matching between the device's output power and the required power, while reducing overall fuel consumption.

[0007] The technical solution to achieve the purpose of this invention is: an energy management strategy for a series hybrid tracked vehicle based on predicted vehicle speed changes, comprising the following steps:

[0008] Step (1): Establish the longitudinal dynamics model of the series hybrid tracked vehicle;

[0009] Step (2): Establish an A-ECMS energy management strategy model for series hybrid tracked vehicles;

[0010] Step (3): Use the LSTM method to predict the velocity change value, divide the change value into positive and negative categories, use the Kmeans method to classify the predicted velocity change value under different signs, and obtain the threshold value among them;

[0011] Step (4): For speed change values ​​whose absolute value is greater than the corresponding lower threshold value, adjust the equivalent factor of the corresponding A-ECMS, and compare the obtained comprehensive fuel consumption with the comprehensive fuel consumption obtained from A-ECMS, and take the APU and power battery power distribution corresponding to the smaller value.

[0012] Furthermore, the longitudinal dynamic model in step (1) is as follows:

[0013]

[0014] in, For the power required by the vehicle, The vehicle's current speed. For vehicle quality, The rolling resistance coefficient, air density, For windward area, The air drag coefficient, To accelerate the vehicle, It is the angle between the road surface and the horizontal plane.

[0015] Furthermore, step (2) establishes the A-ECMS energy management strategy model as follows:

[0016]

[0017]

[0018] in,

[0019] in, For equivalent fuel consumption, For engine fuel consumption, This is the equivalent fuel consumption of the power battery. As an equivalent factor, For the output power of the power battery, Because it is a low-calorific-value fuel, P is the APU output power. d (t) represents the required power. The optimal equivalent factor obtained through offline iteration. For adjustment coefficients, The initial SOC of the power battery is given by SOC(t), and the real-time SOC of the power battery is given by SOC(t). and It is determined by the deviation between the final SOC value and the initial SOC value and the overall fuel consumption value.

[0020] Furthermore, step (3) specifically involves:

[0021] Define the LSTM network architecture and hyperparameters, and train and test the reference working condition data after normalization, adjusting the parameters until the test results reach the fit.

[0022] The velocity change threshold was obtained using the K-means clustering algorithm as follows: Two categories were set for the change values: "medium-high" and "medium-low". The predicted velocity change values ​​were then divided into positive and negative categories, and K-means clustering was performed on each category. The median value was used as the threshold for differentiation. Data with absolute values ​​greater than the corresponding sign threshold were analyzed using P-ECMS. The equivalent factor formula in P-ECMS, representing the predicted vehicle speed change, is as follows:

[0023]

[0024] in, The adaptive equivalence factor obtained from the aforementioned A-ECMS is... This is the P-ECMS adjustment coefficient. The threshold value for a positive velocity change. The threshold value for negative velocity change. To predict the maximum absolute value of the velocity change when the velocity change is positive, To predict the maximum absolute value of the velocity change when the velocity change is negative, It is determined by the deviation between the final SOC value and the initial SOC value and the overall fuel consumption value.

[0025] Furthermore, the method for comparing the overall fuel consumption in step (4) "comparing the obtained overall fuel consumption with the overall fuel consumption obtained from A-ECMS" is as follows:

[0026]

[0027] In the formula, The equivalent fuel consumption for electrical energy, in L; Battery energy consumption, in kWh; It is diesel fuel with a low calorific value of 42600 kJ / kg; The density of diesel fuel is 0.84 kg / L. The average operating efficiency of the engine is taken as 34% here; The average power generation efficiency of the ISG motor is taken as 90% here; The total fuel consumption is expressed in liters (L). This represents engine fuel consumption, expressed in liters (L).

[0028] Furthermore, the objective function of the energy management strategy model is:

[0029]

[0030] in, Based on the output power distribution or ;

[0031] Furthermore, the constraints of the energy management strategy model are as follows:

[0032]

[0033] in, This represents the minimum state of charge of the power battery. This represents the maximum state of charge (SOC) of the power battery. This is the minimum power that the power battery can achieve. This represents the maximum power that the power battery can achieve. This represents the maximum engine speed. This represents the minimum engine torque when the motor reverses direction. for The maximum value of the engine torque can be found in the table below. This is the minimum adjustable speed for engine speed control. This represents the maximum adjustable torque for engine speed regulation. This is the minimum adjustable torque value for engine speed regulation. This represents the maximum adjustable torque for engine speed regulation.

[0034] Compared with the prior art, the significant advantages of this invention are:

[0035] In vehicles where APU and power battery are used as power sources, the matching between the device's output power and the required power is further improved based on the predicted vehicle speed, taking into account the stability of the power battery's SOC, while also achieving a further reduction in overall fuel consumption. Attached Figure Description

[0036] Figure 1 This is a structural diagram of a vehicle system provided in an example of the present invention.

[0037] Figure 2 This is a diagram of the LSTM network structure provided in the example of the present invention.

[0038] Figure 3 This is a flowchart of an energy management strategy for a series hybrid tracked vehicle based on predicted vehicle speed changes, provided in an example of the present invention.

[0039] Figure 4 This describes the LSTM network application process in an example of the present invention.

[0040] Figure 5 This describes a complex working condition for offline training of the LSTM network provided in the examples of this invention.

[0041] Figure 6 The operating conditions used in the simulation provided in the examples of this invention. Detailed Implementation

[0042] The present invention will now be described in further detail with reference to the accompanying drawings.

[0043] Figure 1 The diagram shows the vehicle system structure, divided into a front power chain and a rear power chain. Energy sources are concentrated in the front power chain, while consumption sources are concentrated in the rear power chain. Energy sources include the APU (Automatic Power Unit) and the battery, while consumption sources include electrical accessories and the drive motor. The APU consists of the engine and the ISG (Independent Shutter Controller) motor. When the engine starts, the ISG motor reverse-drives the engine to complete the start-up; after the engine starts, it drives the ISG motor to output electrical energy. When the vehicle is driving, the APU and battery in the front power chain provide the power source. When the vehicle brakes, the battery in the front power chain recovers the energy generated during braking.

[0044] Figure 2 The diagram shows an LSTM network, which includes three types of gate structures: forget gate, input gate, and output gate. The forget gate determines whether to retain information based on the information from the previous hidden state and the current state; the input gate updates the cell state and incorporates new information discovered by the neural network; and the output gate determines the value of the next hidden state.

[0045] Figure 3 This is a flowchart of the energy management strategy for a series hybrid tracked vehicle based on predicted vehicle speed changes. The equivalent factor is adjusted based on the vehicle speed change value predicted by LSTM, and the resulting comprehensive fuel consumption is compared with the comprehensive fuel consumption obtained from A-ECMS. The smaller value corresponds to the power allocation of the APU and the power battery.

[0046] The specific implementation method mainly includes the following steps:

[0047] Step 1: Establish the longitudinal dynamics model of the series hybrid tracked vehicle;

[0048] (1)

[0049] in, For the power required by the vehicle, The vehicle's current speed. For vehicle quality, The rolling resistance coefficient, air density, For windward area, The air drag coefficient, To accelerate the vehicle, It is the angle between the road surface and the horizontal plane.

[0050] Step 2: Establish an A-ECMS energy management strategy model tailored to the characteristics of series hybrid tracked vehicles;

[0051] First, the ECMS energy management strategy aims to minimize the equivalent fuel consumption (fuel consumption plus electricity consumption) under simulated operating conditions. Its objective function is:

[0052] (2)

[0053] in, For equivalent fuel consumption, For engine fuel consumption, This is the equivalent fuel consumption of the power battery.

[0054] This can be achieved by solving for the optimal solution at each instant, as shown in the following formula:

[0055] (3)

[0056] (4)

[0057] in, As an equivalent factor, For the output power of the power battery, Because it is a low-calorific-value fuel, This refers to the APU's output power.

[0058] The above formula is the basic formula of the ECMS method. Since series hybrid vehicles need to consider the SOC stability of the power battery, A-ECMS adds SOC feedback-based adjustment to the traditional ECMS, which can enhance the adaptability of the energy management strategy and improve the SOC stability of the power battery. The formula is as follows:

[0059] (5)

[0060] in, The optimal equivalent factor obtained through offline iteration. For adjustment coefficients, This represents the initial state of charge (SOC) of the power battery. and It is determined by the deviation between the final SOC value and the initial SOC value and the overall fuel consumption value.

[0061] Step 3: Based on the goals of matching the output power of the booster device with the required power and reducing overall fuel consumption, the equivalent factor of A-ECMS is further adjusted using the vehicle speed change values ​​predicted by LSTM. Since adjusting all speed change values ​​is not very meaningful, the threshold values ​​corresponding to positive and negative speed change values ​​are obtained through the K-means method, and P-ECMS is only performed for cases where the absolute value of the speed change value is greater than the corresponding threshold value.

[0062] The main steps include:

[0063] This paper first uses a driving condition with rich speed variations as a reference condition to train the LSTM network parameters offline. Then, the network is applied to the simulated driving condition, and the network parameters are further updated based on the real-time vehicle speed. Since reducing the equivalence factor promotes a decrease in APU output power, and increasing the equivalence factor promotes an increase in APU output power, the equivalence factor can be adjusted based on the predicted vehicle speed change value to improve the matching between APU output power and required power, as the APU is the main power output device. If the predicted vehicle speed change value is positive, the equivalence factor can be increased to further promote an increase in APU output power. If the predicted vehicle speed change value is negative, the equivalence factor can be decreased to further promote a decrease in APU output power.

[0064] First, the LSTM network architecture and hyperparameters are defined. Then, the reference working condition data is normalized and used for training and testing. The parameters are adjusted until the test results achieve a good fit. Finally, the LSTM network is applied to this working condition. The predicted vehicle speed change value is represented. Figure 4 For the LSTM network application process, the speed change values ​​at the current time and the previous time are input to update the LSTM network parameters. At the same time, the LSTM network is input after normalization based on the LSTM parameters, and the output value is output after inverse normalization to predict the vehicle speed change value.

[0065] We set up two categories for the change values: "medium-high" and "medium-low". We then performed K-means clustering on both positive and negative predicted velocity change values, using the median value as a threshold for differentiation. Only data with absolute values ​​greater than the corresponding threshold were analyzed using P-ECMS, as shown in the following formula:

[0066] (6)

[0067] in, The adaptive equivalence factor obtained from the aforementioned A-ECMS is... This is the P-ECMS adjustment coefficient. The threshold value for a positive velocity change. The threshold value for negative velocity change. , These represent the maximum absolute value of the velocity change under the corresponding predicted velocity change conditions. It is determined by the deviation between the final SOC value and the initial SOC value and the overall fuel consumption value.

[0068] Step 4: For speed change values ​​whose absolute values ​​are greater than the corresponding sign threshold values, based on the combined fuel consumption values ​​obtained from A-ECMS and P-ECMS, select the APU and power battery output power distribution corresponding to the smaller value as the output distribution at the current moment.

[0069] The method for comparing overall fuel consumption values ​​is as follows:

[0070] (7)

[0071] In the formula, The equivalent fuel consumption for electrical energy, in L; Battery energy consumption, in kWh; It is diesel fuel with a low calorific value of 42600 kJ / kg; The density of diesel fuel is 0.84 kg / L. The average operating efficiency of the engine is taken as 34% here; The average power generation efficiency of the ISG motor is taken as 90% here; The total fuel consumption is expressed in liters (L). This represents engine fuel consumption, expressed in liters (L).

[0072] In this invention, the state variable of the energy management strategy is the power battery SOC, and the control variable is the engine operating point, which is a vector composed of speed and torque.

[0073] In this invention, the objective function of the energy management strategy is:

[0074] (8)

[0075] in, Based on the output power distribution or .

[0076] In this invention, the constraints of the energy management strategy are:

[0077] (9)

[0078] in, This represents the minimum state of charge of the power battery. This represents the maximum state of charge (SOC) of the power battery. This is the minimum power that the power battery can achieve. This represents the maximum power that the power battery can achieve. This represents the maximum engine speed. This represents the minimum engine torque when the motor reverses direction. for The maximum value of the engine torque can be found in the table below. This is the minimum adjustable speed for engine speed control. This represents the maximum adjustable torque for engine speed regulation. This is the minimum adjustable torque value for engine speed regulation. This represents the maximum adjustable torque for engine speed regulation.

Claims

1. A series hybrid tracked vehicle energy management method based on predicted vehicle speed changes, characterized by, Includes the following steps: Step (1): Establish the longitudinal dynamics model of the series hybrid tracked vehicle; Step (2): Establish an A-ECMS energy management strategy model for series hybrid tracked vehicles; Step (3): Predict the velocity change value using the LSTM method, divide the change value into positive and negative categories, use the Kmeans method to classify the predicted velocity change value under different signs, and obtain the threshold value; for velocity change values ​​whose absolute value is greater than the threshold value under the corresponding sign, adjust the equivalent factor of the corresponding A-ECMS. Step (4): Compare the obtained comprehensive fuel consumption with the comprehensive fuel consumption obtained from A-ECMS, and take the smaller value corresponding to the APU and power battery power distribution. Step (3) is as follows: Define the LSTM network architecture and hyperparameters, and train and test the reference working condition data after normalization, adjusting the parameters until the test results reach the fit. The velocity change threshold was obtained using the K-means clustering algorithm as follows: Two categories were set for the change values: "medium-high" and "medium-low". The predicted velocity change values ​​were then divided into positive and negative categories, and K-means clustering was performed on each category. The median value was used as the threshold for differentiation. Data with absolute values ​​greater than the corresponding sign threshold were analyzed using P-ECMS. The equivalent factor formula in P-ECMS, representing the predicted vehicle speed change, is as follows: ， in, The adaptive equivalence factor obtained from the aforementioned A-ECMS is... This is the P-ECMS adjustment coefficient. The threshold value for a positive velocity change. The threshold value for negative velocity change. To predict the maximum absolute value of the velocity change when the velocity change is positive, To predict the minimum absolute value of the velocity change when the velocity change is negative. It is determined by the deviation between the final SOC value and the initial SOC value and the overall fuel consumption value.

2. The management method according to claim 1, characterized in that, The longitudinal dynamic model in step (1) is as follows: ， in, For the power required by the vehicle, The vehicle's current speed. For vehicle quality, The rolling resistance coefficient, air density, For windward area, The air drag coefficient, To accelerate the vehicle, It is the angle between the road surface and the horizontal plane.

3. The energy management method of claim 2, wherein, Step (2) establishes the A-ECMS energy management strategy model as follows: ， ， wherein , in, For equivalent fuel consumption, For engine fuel consumption, This is the equivalent fuel consumption of the power battery. For the output power of the power battery, Because it is a low-calorific-value fuel, P is the APU output power. d (t) represents the required power. The optimal equivalent factor obtained through offline iteration. For adjustment coefficients, The initial SOC of the power battery is given by SOC(t), and the real-time SOC of the power battery is given by SOC(t). and It is determined by the deviation between the final SOC value and the initial SOC value and the overall fuel consumption value.

4. The energy management method of claim 3, wherein, The calculation method for the combined fuel consumption in step (4) "Compare the obtained combined fuel consumption with the combined fuel consumption obtained from A-ECMS" is as follows: ， In the formula, The equivalent fuel consumption for electrical energy, in L; Battery energy consumption, in kWh; It is diesel fuel with a low calorific value of 42600 kJ / kg; The density of diesel fuel is 0.84 kg / L. The average operating efficiency of the engine is taken as 34% here; The average power generation efficiency of the ISG motor is taken as 90% here; The total fuel consumption is expressed in liters (L). This represents engine fuel consumption, expressed in liters (L).

5. The energy management method according to claim 4, characterized in that, The objective function of the energy management strategy model is: ， in, Based on the output power distribution or .

6. The energy management method according to claim 5, characterized in that, The constraints of the energy management strategy model are: ， in, This represents the minimum state of charge of the power battery. This represents the maximum state of charge (SOC) of the power battery. This is the minimum power that the power battery can achieve. This represents the maximum power that the power battery can achieve. This represents the maximum engine speed. This represents the minimum engine torque when the motor reverses direction. for The maximum value of the engine torque can be found in the table below. This is the minimum adjustable speed for engine speed control. This represents the maximum adjustable torque for engine speed regulation. This is the minimum adjustable torque value for engine speed regulation. This represents the maximum adjustable torque for engine speed regulation.

Images