New energy automobile engine-generator speed ratio collaborative optimization method

Abstract

The invention belongs to the technical field of new energy automobile transmission system optimization design, and provides a new energy automobile engine-generator speed ratio collaborative optimization method, which comprises the following steps: calculating a speed ratio range through an engine optimal working curve in combination with constraint conditions, and adopting an improved multi-objective optimization algorithm to optimize the speed ratio range. Rapid matching and efficient optimization of the speed ratio are achieved, so that the fuel economy and the dynamic property of the whole vehicle are improved, the development speed is increased, and the development cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of optimized design of transmission systems for new energy vehicles, and in particular relates to a method for collaboratively optimizing the speed ratio of an engine and a generator for a new energy vehicle. Background Art

[0002] With the advancement of global carbon neutrality goals and the rapid development of diversified new energy vehicle technologies, parallel-parallel hybrid systems and extended-range electric vehicles have become the mainstream technology routes in the market because they take into account both driving range and low-carbon characteristics. Although the configurations are significantly different, the core optimization goals of the two types of systems are highly consistent - to maximize energy conversion efficiency by optimizing the speed ratio of the power coupling device. In these two configurations, the engine and generator achieve power distribution through planetary gears, clutches or motor coupling, and their speed ratio directly determines the working efficiency of the engine, the energy conversion efficiency of the generator and the power response performance of the entire vehicle. However, the existing technology still faces the following key problems in practical applications:

[0003] 1. Single-objective optimization: Existing speed ratio optimization methods often focus on a single objective (such as minimizing fuel consumption), ignoring the coordination of multiple objectives such as power, noise and vibration (NVH), and battery charge and discharge efficiency. This makes it difficult to meet users' requirements for comprehensive performance.

[0004] 2. Insufficient algorithm adaptability: Traditional optimization algorithms (such as gradient descent and dynamic programming) are prone to falling into local optimality when dealing with high-dimensional nonlinear constraints, and their high computational complexity makes it difficult to meet the requirements;

[0005] 3. Insufficient consideration of constraints: Fuel economy, NVH performance, manufacturing cost and other objectives are mutually constrained. In existing designs, certain constraints are usually abandoned for design calculations, and there is a lack of a systematic weight allocation mechanism. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides a method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle to solve the problems in the prior art. The technical solution adopted by the present invention is:

[0007] A method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle comprises the following steps:

[0008] Step 1: determine the engine operating point;

[0009] Step 2: determine the constraints, which include speed limit, bus voltage and generator output current limit, engine and generator external characteristic limit, and generator minimum torque limit;

[0010] Step three: construct a multi-objective optimization algorithm.

[0011] Furthermore, step one includes: drawing an equivalent fuel consumption curve based on equivalent fuel consumption, determining the corresponding minimum equivalent fuel consumption rate operating point for each given power value by calculation, performing trajectory fitting on the minimum energy consumption operating point corresponding to each output power, and obtaining an optimal efficiency curve with optimal energy efficiency characteristics.

[0012] Furthermore, in step 2:

[0013] Speed ​​limit: The lowest speed of the engine's optimal working range is n Emin , the maximum speed n in the optimal working area Emax , the maximum speed of the generator is n Mmax , the speed ratio satisfies:

[0014]

[0015] Furthermore, in step 2:

[0016] Limitations on bus voltage and generator output current: DC bus maximum voltage U max , DC bus minimum voltage U min , the allowable speed of the generator meets:

[0017]

[0018] Where, I dc k is the generator's allowed output current; α 、k β is the correlation coefficient.

[0019] Furthermore, in step 2:

[0020] Engine and generator external characteristic limitations: As the speed increases, there is an intersection between the speed of the generator and the engine. Before the intersection, the engine's power external characteristic curve is the limit, and after the intersection, the generator's power external characteristic is the limit.

[0021] Furthermore, in step 2:

[0022] Generator minimum torque limit: For a series-parallel hybrid electric vehicle, the minimum torque meets the following requirements:

[0023]

[0024] Where, T E_st is the starting torque required by the engine; n idle is the engine idle speed; t s is the acceleration time; I is the engine moment of inertia.

[0025] Furthermore, in step 3, constructing a multi-objective optimization algorithm includes an initialization phase, an evolutionary iteration phase, and a termination phase, wherein:

[0026] Initialization phase: In this phase, the algorithm randomly generates an initial population P0 of N, where each individual represents a candidate solution;

[0027] Evolutionary Iteration Phase: In this phase, the following steps are performed:

[0028] Generate offspring population Q t : Select from P through binary tournament t Select the individuals of the parent generation, and then perform crossover and mutation operations on the parent generation to generate the offspring population Q t ;

[0029] Merging populations: via R t =P t ∪Q t Operation, merge the population, the size of the merged population is 2N;

[0030] Non-dominated sorting and congestion calculation: R t The individuals in the front are stratified according to the dominance relationship, and then the distribution density of individuals in the same front in each target space is calculated;

[0031] Generate a new generation of population: fill individuals from high to low according to the frontier level. If a frontier cannot be fully accommodated, select individuals with high crowding degree in the frontier until N positions are filled.

[0032] Termination phase: The loop reaches the maximum number of iterations or the Pareto front has converged.

[0033] Furthermore, in step three, constructing a multi-objective optimization algorithm includes:

[0034] Step 1: Construct a DBO-BP neural network agent model based on the dung beetle optimization algorithm and BP neural network.

[0035] Step 2: Dynamic update mechanism of the agent model: The neural network agent model is dynamically updated according to the number of iterations. The optimal key individuals are evaluated every 10 generations, and the DBO-BP neural network agent model is updated based on the evaluation results and the working condition data corresponding to the key individuals.

[0036] Step 3: Determine the objective function:

[0037] Fuel consumption is selected to represent the energy consumption of the engine, cumulative power loss is selected to represent the performance of the generator, and engine speed is selected to represent the NVH performance of the engine-generator set. The objective function of the DBO-BP-NSGA multi-objective optimization algorithm is:

[0038]

[0039] Furthermore, the first step includes:

[0040] The position of the DBO optimal individual is expressed as:

[0041] x i (t+1)=x i (t)+αkx i (t-1)+bΔx

[0042] Δx=|x i (t)-X w |

[0043] Where t represents the current number of iterations, x i (t) represents the position information of the i-th dung beetle at the t-th iteration, k∈(0,0.2] represents the deflection coefficient, which is a constant value, b represents a constant value belonging to (0,1), α is the natural coefficient, X w represents the global worst position, and Δx is used to simulate the change of light intensity;

[0044] If the optimization direction is restricted, it will be repositioned and the position will be updated as follows:

[0045] x i (t+1)=x i (t)+tanθ|x i (t)-x i (t+1)|

[0046] Where θ is the deflection angle, |x i (t)-x i (t+1)| represents the difference between the position of the i-th dung beetle at the t-th iteration and its position at the t-1-th iteration;

[0047] During the iteration process, the location update information is defined as:

[0048] |x i (t+1)=X b +S×g×(x i (t)-X * |+|x i (t)-X b |)

[0049] Where, X b represents the global optimal position, g is a random vector of size 1×D, S represents a constant value, X * is the local optimal position at the current moment, X b Indicates the global optimal position; during the optimization process, the position is continuously updated, and finally the global optimal position X is output b ;

[0050] The optimization parameters of the DBO algorithm are set to the initial weight ω and learning rate η of the BP network, namely:

[0051] x i (t) = f(ω, η)

[0052] The fitness function is set as the root mean square error of the BP neural network for the training set and the test set, that is:

[0053] finteness=RMSE[predict(train)]+RMSE[predict(test)]

[0054]

[0055] Where RMSE(k) is the root mean square error of the prediction time domain of the kth sampling point; n is the number of sampling points; t p The length of the prediction time.

[0056] Furthermore, in the first step, when pre-training the model, the engine operating points under the combined working conditions of the global light vehicle test cycle and the Chinese light vehicle test cycle are selected as the data set to pre-train the DBO-BP neural network.

[0057] The present invention has the following beneficial effects:

[0058] (1) The present invention calculates the speed ratio range by combining multiple constraints with the engine's optimal operating curve, and adopts an improved multi-objective optimization algorithm to achieve rapid speed ratio matching and efficient optimization. This method comprehensively considers multiple constraints and is suitable for parallel-parallel hybrid systems and extended-range electric vehicles. It can improve the fuel economy and power performance of the entire vehicle, increase development speed, and reduce development costs.

[0059] (2) The present invention comprehensively considers multiple constraints and is applicable to parallel-parallel hybrid power systems and extended-range electric vehicles; an agent model is used to improve the multi-objective optimization algorithm, and the agent model can be dynamically updated during the optimization process, thereby improving the running speed and calculation accuracy of the multi-objective optimization algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Fuel consumption curve and optimal working curve for the engine;

[0061] Figure 2 Flowchart of the DBO-BP-NSGA multi-objective optimization algorithm;

[0062] Figure 3 The speed ratio is affected;

[0063] Figure 4 Generator operating points before and after optimization;

[0064] Figure 5 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0065] The following is a combination of the embodiments of the present invention Figure 1-Figure 5 , the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0066] like Figure 5 A method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle comprises the following steps:

[0067] Step 1: Determine the engine operating point:

[0068] In practice, the engine's operating point varies depending on driving conditions and control strategies. To simplify the design process and improve the applicability of the method, the engine's optimal efficiency operating curve is used as the engine's operating point.

[0069] The equivalent fuel consumption curve is used to draw the equivalent fuel consumption curve. For each given power value, the corresponding minimum equivalent fuel consumption rate operating condition point is determined by calculation. The trajectory fitting of the minimum energy consumption operating condition point corresponding to each output power can obtain the optimal efficiency curve that represents the optimal energy efficiency characteristics of the system. The engine equivalent fuel consumption curve and the optimal operating curve in this example are as follows: Figure 1 This power curve is based on the consideration of minimum energy consumption. In actual situations, it can be appropriately offset according to the vehicle's mode requirements.

[0070] Step 2: Determine the constraints:

[0071] The actual operating range of the engine-generator is mainly constrained by the following aspects:

[0072] (1) Speed ​​limit: The actual operating speed range is limited by the performance of both the engine and the generator. The lowest speed in the optimal operating range of the engine is n Emin , the maximum speed n in the optimal working area Emax , the maximum speed of the generator is n Mmax , then the speed ratio must satisfy:

[0073]

[0074] (2) Limitation of bus voltage and generator output current: In order to ensure the safety, reliability, efficient operation and equipment life of the vehicle power system, and to prevent thermal runaway, device damage and safety hazards caused by overvoltage and overcurrent, the DC bus voltage is usually limited. The maximum DC bus voltage U max , DC bus minimum voltage U min , then the allowable speed of the generator must meet:

[0075]

[0076] Where, I dc k is the generator's allowed output current (A); α 、k β is the correlation coefficient, which can be obtained by fitting the generator external characteristic curve.

[0077] (3) Limitations of the external characteristics of the engine and generator: The engine and generator have different operating characteristics. Generally speaking, at low speeds, the external characteristic power of the generator is greater than that of the engine, and at high speeds, the external characteristic power of the engine is greater than that of the generator. As the speed increases, there is usually an intersection between the two. During design, it is necessary to try to make the operating range fall within the coverage of the external characteristic power curves of the two. That is, before the intersection, the engine's external characteristic power curve is used as the limit, and after the intersection, the generator's external characteristic power curve is used as the limit. If it is unavoidable that the operating range exceeds the limit, the generator will overload. In this case, the overload operation time needs to be controlled in the control strategy.

[0078] (4) Minimum generator torque limit: For a series-parallel hybrid vehicle, the generator also has to take on the engine speed regulation work and must meet the requirement of quickly starting the engine from a stationary state to a stable idle state within 0.4 seconds. Therefore, the minimum torque must meet the following requirements:

[0079]

[0080]

[0081] Where, T E_st is the starting torque required by the engine (N·m); n idle is the engine idle speed (r / min), generally 750-850; t s is the acceleration time (s), which is taken as 0.4; I is the engine moment of inertia (kg·m 2 ), generally 0.2~0.5.

[0082] Step 3: Construction of multi-objective optimization algorithm

[0083] NSGA-II is a classic multi-objective optimization algorithm, often used to solve complex optimization problems with multiple conflicting objective functions. Its core concept is to balance convergence and diversity of solutions through non-dominated sorting and congestion distance strategies, thereby approaching the Pareto optimal solution set for the problem. The algorithm process is divided into three phases: initialization, evolutionary iteration, and termination:

[0084] (1) Initialization phase:

[0085] In this stage, the algorithm randomly generates an initial population P0 of N, where each individual represents a candidate solution.

[0086] (2) Evolutionary Iteration Stage:

[0087] In this stage, for each time instant t,

[0088] Generate offspring population Q t :

[0089] First, through the binary tournament selection from P t Select the individuals of the parent generation, and then perform crossover and mutation operations on the parent generation to generate the offspring population Q t .

[0090] Merge populations:

[0091] By R t =P t ∪Q t Operation, merge the population, the size of the merged population is 2N.

[0092] Non-dominated sorting and congestion calculation:

[0093] First, R t The individuals in the frontier are stratified according to the dominance relationship, and then the distribution density of individuals in the same frontier in each target space is calculated.

[0094] Generate a new generation of population:

[0095] Individuals are filled in from high to low according to the frontier level. If a certain frontier cannot be fully accommodated, individuals with high congestion in the frontier are selected until N positions are filled.

[0096] (3) Termination stage

[0097] The loop has reached the maximum number of iterations or the Pareto front has converged.

[0098] NSGA-II has been widely used in various fields. However, it is susceptible to the "curse of dimensionality" in high-dimensional target spaces, where its crowding mechanism can fail. Furthermore, the computational cost of non-dominated sorting for large populations is high. To address this issue, this paper proposes a DBO-BP-NSGA multi-objective optimization algorithm using a DBO-BP neural network as a surrogate model. By introducing the DBO-BP neural network surrogate model, an approximate model of the objective function is established, replacing some of the true evaluations and thus reducing computational time. Furthermore, the surrogate model can pre-evaluate the fitness of solutions during the non-dominated sorting phase, allowing for rapid screening of potential high-quality solutions.

[0099] Step 1: Construct a Dung Beetle Optimization Algorithm (DBO)-BP neural network agent model based on the Dung Beetle Optimization Algorithm (DBO) and BP neural network:

[0100] The BP neural network is a feedforward artificial neural network based on the backpropagation mechanism of errors. It uses a gradient descent algorithm to adjust network weights layer by layer to minimize the error between predicted and true values. It excels in nonlinear modeling, pattern recognition, and forecasting tasks, and is widely used in engineering. The detailed principles are not elaborated here. However, BP neural networks have strict restrictions on the selection of initial weights. During backpropagation, the gradient can be scaled too small or too large layer by layer due to the chain rule, resulting in slow or oscillatory updates of shallow weights and difficulty in model convergence. Fixed learning rates require manual adjustment; excessively large ones lead to oscillations, while too small ones lead to slow convergence. Furthermore, when the error surface is complex, gradient descent is prone to falling into local optima and failing to find the global optimum. Therefore, the DBO algorithm, with its adaptive global optimization capability, is integrated to optimize the initial weights and learning rate parameters of the BP neural network, improving the training efficiency and prediction accuracy of the BP neural network prediction model.

[0101] The Dung Beetle Optimizer (DBO) is an algorithm that simulates the biological behavior of dung beetles. It has strong optimization capabilities and fast convergence. During operation, the position of the optimizing individual can be expressed as:

[0102] x i (t+1)=x i (t)+αkx i (t-1)+bΔx

[0103] Δx=|x i (t)-X w |

[0104] Where t represents the current number of iterations, x i (t) represents the position information of the i-th dung beetle at the t-th iteration, k∈(0, 0.2] represents the deflection coefficient, which is customized. b represents a fixed value belonging to (0, 1), α is a natural coefficient, which is assigned to -1 or 1, and Xw represents the global worst position, and Δx is used to simulate the change of light intensity.

[0105] If the optimization direction is restricted, it will be repositioned and the position will be updated as follows:

[0106] x i (t+1)=x i (t)+tanθ|x i (t)-x i (t+1)|

[0107] Where θ is the deflection angle, which belongs to [0,π]. |x i (t)-x i (t+1)| represents the difference between the position of the i-th dung beetle at the t-th iteration and its position at the t-1-th iteration.

[0108] During the iteration process, the location update information is defined as:

[0109] |x i (t+1)=X b +S×g×(|x i (t)-X * |+|x i (t)-X b |)

[0110] Where, X b represents the global optimal position, g is a random vector of size 1×D, which obeys the normal distribution, S represents a constant value, X * is the local optimal position at the current moment, X b Indicates the global optimal position. During the optimization process, the position is continuously updated, and the global optimal position X is finally output. b .

[0111] In this example, the optimization parameters of the DBO algorithm are set to the initial weight ω and learning rate η of the BP network, that is:

[0112] x i (t) = f(ω, η)

[0113] The fitness function is set as the root mean square error of the BP neural network for the training set and the test set, that is:

[0114] finteness=RMSE[predict(train)]+RMSE[predict(test)]

[0115]

[0116] Where RMSE(k) is the root mean square error of the prediction time domain of the kth sampling point; n is the number of sampling points; t p The length of the prediction time.

[0117] In the pre-training model, the engine operating points under the combined operating conditions of the "World Light Vehicle Test Cycle (WLTC) + China Light Vehicle Test Cycle (CLTC)" are selected as the data set, 80% as the training set and 20% as the test set to pre-train the DBO-BP neural network.

[0118] Step 2: Dynamic update mechanism of proxy model:

[0119] Actual design operating conditions are complex and diverse, and pre-trained neural network proxy models are unlikely to meet all requirements. Therefore, a dynamic proxy model update mechanism is introduced to ensure that the model better matches the operating conditions during actual speed ratio design calculations. During the execution of the improved multi-objective optimization algorithm, the neural network proxy model is dynamically updated based on the number of iterations. In this example, the optimal key individuals are evaluated every 10 generations. The DBO-BP neural network proxy model is then updated based on the evaluation results and the corresponding operating condition data of the key individuals.

[0120] Step 3: Determine the objective function:

[0121] Fuel consumption is selected to represent the energy consumption of the engine, cumulative power loss is selected to represent the performance of the generator, and engine speed is selected to represent the NVH performance of the engine-generator set. The objective function of the DBO-BP-NSGA multi-objective optimization algorithm is:

[0122] The flow chart of DBO-BP-NSGA multi-objective optimization algorithm is as follows Figure 2 As shown. First, NSGA-II is initialized, and the second generation subgroup is generated based on the generated first generation subgroup and the pre-trained proxy model; then the NSGA-II iteration process is carried out, and the offspring generated in each iteration will be corrected by the neural network proxy model to avoid falling into the local optimum; finally, the termination condition is determined based on the maximum number of iterations or the convergence of the Pareto front. In addition, during the iteration process, the neural network proxy model will be dynamically updated according to the number of iterations, and the neural network proxy model will be updated according to the evaluation results of the optimal key individuals every 10 generations. In this example, the initial population of the DBO-BP-NSGA multi-objective optimization algorithm is set to 50, the number of iterations is set to 200, the crossover probability is set to 0.9, and the mutation probability is set to 0.1. The optimization results are shown in Figure 3 As shown in Figure 2, it can be seen that with the change of speed ratio, the cumulative power loss of the generator shows a trend of first decreasing and then increasing. The working points of the generator before and after optimization are as follows: Figure 4As shown in the figure, it can be seen that the engine operating point is more concentrated in the high-efficiency range after optimization, the cumulative power loss is reduced by 0.91 after optimization, and the energy consumption of the generator is improved.

[0123] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various deformations, modifications, and substitutions made to the technical solutions of the present invention by ordinary technicians in this field should fall within the scope of protection determined by the claims of the present invention.

Claims

1. A method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle, characterized in that: The following steps are involved: Step 1: determine the engine operating point; Step 2: determine the constraints, which include speed limit, bus voltage and generator output current limit, engine and generator external characteristic limit, and generator minimum torque limit; Step three: construct a multi-objective optimization algorithm to achieve speed ratio optimization.

2. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: Step 1 includes: drawing an equivalent fuel consumption curve based on equivalent fuel consumption, determining the corresponding minimum equivalent fuel consumption rate operating point for each given power value through calculation, and fitting the trajectory of the minimum energy consumption operating point corresponding to each output power to obtain the best efficiency curve with optimal energy efficiency characteristics.

3. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: In step 2: Speed ​​limit: The lowest speed of the engine's optimal working range is n Emin , the maximum speed n in the optimal working area Emax , the maximum speed of the generator is n Mmax , the speed ratio satisfies:

4. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: In step 2: Limitations on bus voltage and generator output current: DC bus maximum voltage U max , DC bus minimum voltage U min , the allowable speed of the generator meets: Where, I dc k is the generator's allowed output current; α 、k β is the correlation coefficient.

5. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: In step 2: Engine and generator external characteristic limitations: As the speed increases, there is an intersection between the speed of the generator and the engine. Before the intersection, the engine's power external characteristic curve is the limit, and after the intersection, the generator's power external characteristic is the limit.

6. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: In step 2: Generator minimum torque limit: For a series-parallel hybrid electric vehicle, the minimum torque meets the following requirements: Where, T E_st is the starting torque required by the engine; n idle is the engine idle speed; t s is the acceleration time; I is the engine moment of inertia.

7. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: In step 3, the construction of the multi-objective optimization algorithm includes the initialization phase, the evolutionary iteration phase, and the termination phase, where: Initialization phase: In this phase, the algorithm randomly generates an initial population P0 of N, where each individual represents a candidate solution; Evolutionary Iteration Phase: In this phase, the following steps are performed: Generate offspring population Q t : Select from P through binary tournament t Select the individuals of the parent generation, and then perform crossover and mutation operations on the parent generation to generate the offspring population Q t ; Merge populations: via R t =P t ∪Q t Operation, merge the population, the size of the merged population is 2N; Non-dominated sorting and congestion calculation: R t The individuals in the front are stratified according to the dominance relationship, and then the distribution density of individuals in the same front in each target space is calculated; Generate a new generation of population: fill individuals from high to low according to the frontier level. If a frontier cannot be fully accommodated, select individuals with high crowding degree in the frontier until N positions are filled. Termination phase: The loop reaches the maximum number of iterations or the Pareto front has converged.

8. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 1, characterized in that: In step three, building a multi-objective optimization algorithm includes: Step 1: Construct a DBO-BP neural network agent model based on the dung beetle optimization algorithm and BP neural network; Step 2: Dynamic update mechanism of the agent model: The neural network agent model is dynamically updated according to the number of iterations. The optimal key individuals are evaluated every 10 generations, and the DBO-BP neural network agent model is updated based on the evaluation results and the working condition data corresponding to the key individuals. Step 3: Determine the objective function: Fuel consumption is selected to represent the energy consumption of the engine, cumulative power loss is selected to represent the performance of the generator, and engine speed is selected to represent the NVH performance of the engine-generator set. The objective function of the DBO-BP-NSGA multi-objective optimization algorithm is:

9. A method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 8, characterized in that the first step include: The position of the DBO optimal individual is expressed as: x i (t+1)=x i (t)+αkx i (t-1)+bΔx Δx=|x i (t)-X w | Where t represents the current number of iterations, x i (t) represents the position information of the i-th dung beetle at the t-th iteration, k∈(0, 0.2 represents the deflection coefficient, which is a constant value, b represents a constant value belonging to (0, 1), α is the natural coefficient, X w represents the global worst position, and Δx is used to simulate the change of light intensity; If the optimization direction is restricted, it will be repositioned and the position will be updated as follows: x i (t+1)=x i (t)+tanθ|x i (t)-x i (t+1)| Where θ is the deflection angle, |x i (t)-x i (t+1)| represents the difference between the position of the i-th dung beetle at the t-th iteration and its position at the t-1-th iteration; During the iteration process, the location update information is defined as: |x i (t+1)=X b +S×g×(|x i (t)-X * |+|x i (t)-X b |) Where, X b represents the global optimal position, g is a random vector of size 1×D, S represents a constant value, X * is the local optimal position at the current moment, X b Indicates the global optimal position; during the optimization process, the position is continuously updated, and finally the global optimal position X is output b ; The optimization parameters of the DBO algorithm are set to the initial weight ω and learning rate η of the BP network, namely: x i (t)=f(ω,η) The fitness function is set as the root mean square error of the BP neural network for the training set and the test set, that is: finteness=RMSE[predict(train)]+RMSE[predict(test)] Where RMSE(k) is the root mean square error of the prediction time domain of the kth sampling point; n is the number of sampling points; t p The length of the prediction time.

10. The method for collaboratively optimizing the engine-generator speed ratio of a new energy vehicle according to claim 8, characterized in that: In the first step, when pre-training the model, the engine operating points under the combined conditions of the global light vehicle test cycle and the Chinese light vehicle test cycle are selected as the data set to pre-train the DBO-BP neural network.

Images