Forward design and system integration method for fuel cell hybrid power locomotive

Abstract

The invention discloses a fuel cell hybrid power locomotive forward design and system integration method, which comprises the following steps: S100, establishing a locomotive typical working condition model according to locomotive operation data and line conditions in combination with constraint conditions, and solving a fuel cell train hybrid power system forward integration feasible region, all feasible power system and hydrogen storage system configuration combinations are established; s200: according to the obtained feasible region, combining with a locomotive operation strategy, performing forward design on the fuel cell locomotive, predicting the power demand of the configured locomotive, calculating the purchase cost, the operation hydrogen consumption cost and the degradation cost of each configuration component in the locomotive, further performing weighting to obtain the total cost of the whole life cycle, and determining an optimal parameter combination with the lowest total cost; and S300, according to the determined parameter combination, carrying out locomotive system integration by using digital twinning. According to the invention, the operation efficiency of the fuel cell hybrid power locomotive can be further improved, and the lightweight and modular design of the whole locomotive is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of fuel cells, and in particular relates to a forward design and system integration method for a fuel cell hybrid power locomotive. Background Art

[0002] In recent years, the concepts of green, low-carbon and sustainable development have begun to take root in people's hearts, and more and more companies and scholars have begun to focus on the development and construction of green railways. Among the many new energy trains, hydrogen fuel cell trains have the advantages of high environmental protection, less infrastructure, long endurance, low noise and strong environmental adaptability, and are currently the mainstream research direction. However, due to the weak load-changing capacity of fuel cells, they are often combined with lithium batteries to form a fuel cell hybrid power system to achieve the effect of "peak shaving and valley filling". In addition, lithium batteries can also recover energy during train braking, further improving operating efficiency. At present, fuel cell trains have been put into operation in many countries, reflecting the environmental advantages and commercial potential of fuel cell hybrid trains around the world.

[0003] However, due to the complex internal structure of fuel cell hybrid trains, the difficulty of parameter matching is significantly increased compared to traditional trains. At present, the configuration of each power source in a fuel cell hybrid train often adopts a "reverse design" approach, that is, it is modified according to the existing locomotive, and then verified whether it meets the line information and locomotive parameter requirements. This method is more dependent on the experience of practitioners, and it is difficult to achieve the optimal solution under the configuration requirements; in addition, once the line information and locomotive parameters change, it is often necessary to reconfigure the locomotive manually, and its adaptability is poor. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a forward design and system integration method for a fuel cell hybrid locomotive.

[0005] To achieve the above object, the technical solution adopted by the present invention is: a forward design and system integration method for a fuel cell hybrid locomotive, comprising the steps of:

[0006] S100: Based on the locomotive operation data and line conditions, combined with the constraints including locomotive space constraints, axle weight constraints and bus voltage constraints, a typical operating condition model of locomotive acceleration start, maximum speed and uniform speed driving, low speed and uniform speed climbing and emergency rescue is established, the feasible domain of forward integration of the fuel cell train hybrid power system is solved, and all feasible power system and hydrogen storage system configuration combinations are established;

[0007] S200: Based on the parameter feasible domain obtained in step S100 and in combination with the locomotive operation strategy, the fuel cell locomotive is forward designed, the power demand of the configured locomotive is predicted, the purchase cost, operating hydrogen consumption cost and degradation cost of each configured component in the locomotive are calculated, and then the total cost of the whole life cycle is obtained by weighting, and the optimal parameter combination is determined with the lowest total cost;

[0008] S300: Based on the parameter combination determined in step S200, the locomotive system is integrated using the digital twin.

[0009] Furthermore, the constraints mentioned in step S100 are:

[0010] Locomotive space constraints and axle weight constraints:

[0011]

[0012] Bus voltage constraints:

[0013]

[0014] Where N fc is the number of fuel cells; N sbat With N pbat are the number of lithium batteries connected in series and in parallel respectively; k fc , k bat , k H2_TANK are the additional mass volume coefficients of fuel cells, lithium batteries and hydrogen storage systems; M FC_S 、M BAT_S 、M H2_TANK are the masses of the single fuel cell, lithium battery and hydrogen storage system respectively; V FC_S 、V BAT_S 、V H2_TANK are the volumes of the single fuel cell, lithium battery and hydrogen storage system respectively; P , E , M , V They are power, energy, mass and volume redundancy coefficients respectively; M max_axie V is the train axle weight limit; max is the train volume limit; U BAT_S is the voltage of a single lithium battery; U bus is the bus voltage.

[0015] Furthermore, in step S100, a typical operating condition model of locomotive acceleration start, maximum speed and uniform speed driving, low speed and uniform speed climbing and emergency rescue is established:

[0016] (1) Locomotive acceleration start:

[0017] When a train is running on a straight track at maximum acceleration, it needs to overcome running resistance including basic resistance and acceleration resistance. The power and energy calculation of this process is expressed as:

[0018]

[0019] In the formula, F A , P A 、E A are the traction force, traction power and traction energy of the train during acceleration; M total is the total mass of the train; v turn is the turning speed of the train. When the train speed is lower than the turning speed, the train maintains a uniformly accelerated state. In this state, the acceleration is a i Indicates that when the speed is higher than the turning speed, the traction motor power has reached the maximum value P A,vturn , the train is in a uniform power running state, in which the acceleration is a v Indicates; η M is the traction motor efficiency; γ is the train turning coefficient; F fb is the basic resistance of the train; v is the speed of the train, t is the time, t a The time required for the train to accelerate to its maximum speed;

[0020] (2) Maximum speed for driving at a constant speed on a straight road:

[0021] When the train is on a straight track and running at a constant speed at the highest speed, the train traction system only needs to overcome the basic resistance. The power and energy calculation of this process is expressed as:

[0022]

[0023] In the formula, F C , P C 、E C are the traction force, traction power and traction energy of the train at a constant speed; x e Designing driving range for fuel cell trains;

[0024] (3) Slow and steady speed climbing:

[0025] When the urban train is climbing a slope, a higher traction force is required; the traction power and energy required for the urban train running on a slope are calculated as follows:

[0026]

[0027] In the formula, F R , P R 、E R are the traction force, traction power and traction energy of the train when climbing; F gbis the slope resistance of the train, P TR is the climbing power of the train, F fb is the basic resistance of train operation, t r is the climbing time of the train;

[0028] (4) Emergency rescue conditions

[0029] When the fuel cell stops supplying power due to a fault, the lithium battery will power the entire vehicle and drive it to the nearest rescue station. Considering the straight section and the slope section, the traction energy demand of the fuel cell urban train under emergency rescue conditions is calculated according to the following formula:

[0030]

[0031] In the formula, S tot and S scope Represent the total mileage of self-rescue and the length of the ramp respectively; v max and v scope Respectively represent the maximum operating speed and ramp operating speed; E E The energy required for fuel cell trains in emergency rescue conditions.

[0032] Furthermore, step S100 needs to be trained using a generative adversarial network model based on transfer learning. To improve the performance of the model, random noise is replaced with target domain data, and the LRelu activation function is used in the convolution operation. The formula is:

[0033]

[0034] x is the neuron input.

[0035] Furthermore, the objective function of the adversarial neural network model includes classification loss and domain discriminator loss, expressed as:

[0036]

[0037] In the formula, θ f ,θ d ,θ y are the feature extractor, classifier and domain discriminator parameters respectively;

[0038] Define cross entropy as the loss function for classification loss in source domain data, the formula is:

[0039]

[0040] In the formula, is the source domain data distribution; f f (x i ) is the feature obtained after the source domain data passes through the feature extractor; g y (ff (x i )) is the classifier category probability; y i For sample x i , when the category estimate is equal to x i If the true values ​​are the same, it is 1; if they are different, it is 0. Represents the data point x i Expected value;

[0041] The domain discriminator loss is expressed as:

[0042]

[0043] In the formula, is the target domain data distribution; g d (f f (x i )) is the output of the domain discriminator; d i is the binary label value of the i-th sample.

[0044] Furthermore, the fuel cell locomotive is forward designed, including an inner loop of energy management and an outer loop of optimized matching.

[0045] Furthermore, the energy management inner loop is:

[0046] According to the control system logic analysis, the train power demand model is established:

[0047]

[0048] In the formula, K 1 , K 2 are the state feedback gain vector, K 1 =[k 1 ,k 2 ,k 3 ] T , K 2 =[k 4 ,k 5 ,k 6 ] T ;q 1 ,q 2 are the feedforward gains; c f 、c b are the component controller gains, representing the information transfer between controllers; P rf , P rb , P rw are the power requirements of the fuel cell, lithium battery and vehicle respectively; P f , P b , P w are the actual output power of the fuel cell, lithium battery and vehicle respectively; η f, η b are the efficiencies of fuel cells and lithium batteries respectively;

[0049] State space variable x = [x 1 ,x 2 ,x 3 ] T , specifically:

[0050]

[0051] In the formula, SOC k is the SOC reference value of lithium battery, t f is the train arrival time;

[0052] For ease of calculation, it is assumed that the SOC change rate of the lithium battery is linearly related to the charging and discharging power of the lithium battery, expressed as:

[0053] ΔSOC=-b h P rb ;

[0054] Where b h These are parameters related to the battery hardware system and are determined by the system itself;

[0055] The optimization objectives of the outer loop of the optimization matching include operating cost and acquisition cost. Therefore, the optimization matching problem is described as:

[0056] minCost total (X) = min[Cost g (X)+Cost y (X)N num N year ];

[0057] In the formula, X is the configuration combination in the feasible domain; Cost g Cost y are the purchase cost and operation cost of the locomotive respectively. total Configure the total cost for the train; N num With N year They are the annual operating conditions and service life of the locomotive, and the optimal configuration result is selected according to the actual operation requirements of the line.

[0058] Furthermore, in the energy management inner loop, it is assumed that the control system pole is p 1 、p 2 、p 3 , then the controller feedback gain obtained by the feedback control system pole configuration is:

[0059]

[0060] bh It is the battery hardware system related parameters;

[0061] From the above formula, we can see that k 4 , k 5 By k 1 , k 2 Linear representation, and k 6 =0, so only k needs to be optimized 1 , k 2 , k 3 That's it.

[0062] Furthermore, in the inner loop of the energy management, an objective function is established according to the power following cumulative error, the fuel cell hydrogen consumption and the lithium battery equivalent hydrogen consumption:

[0063]

[0064] In the formula, C fc (t) is the real-time hydrogen consumption of the fuel cell; C bat is the equivalent hydrogen consumption of lithium battery; β is the penalty factor, SOC t is the lithium battery SOC at the end time; ΔP and ΔSOC kt are the errors of train power and lithium battery SOC at time t respectively;

[0065] In order to ensure that the pole pi of the closed-loop system is always located in the left half plane, and the output power of the fuel cell and lithium battery and the SOC of the lithium battery must meet the variation range, the constraint expression of the system is:

[0066]

[0067] real(p i ) represents the real part of the pole pi; P fmin With P fmax are the upper and lower limits of fuel cell output power respectively; P bmin With P bmax They are the upper and lower limits of lithium battery output power; SOC min With SOC max They are the upper and lower limits of lithium battery SOC; SOC tmin With SOC tmax They are the upper and lower limits of the lithium battery SOC at the end of the operation.

[0068] Furthermore, in step S300, the digital twin system is applied to the hydrogen locomotive system integration, and the digital twin system architecture includes a data acquisition module, a physical field calculation module, and a visualization display module;

[0069] (1) Data acquisition module: Use measurement equipment to systematically measure the physical hardware in the locomotive system, process and exchange the collected data through the processing interface using a standard consistent protocol and format, and output data in the same format;

[0070] (2) Physical field calculation module: It is the core part of the digital twin system. It constructs and calculates the digital twin model based on the physical mechanism and formulas through the data imported from the data acquisition module, so as to calculate the corresponding results;

[0071] (3) Visualization display module: After the digital twin calculation results are output to the data cloud, the visualization platform imports the data, and the display module reads the calculation results from the platform and rematches them.

[0072] The beneficial effects of adopting this technical solution are:

[0073] Firstly, the forward design and integration method of fuel cell hybrid locomotive can be used to realize vehicle configuration under limited locomotive space and load conditions, realize locomotive lightweight and modular design, and improve operating efficiency; in addition, combined with historical data, the generative adversarial neural network model based on transfer learning proposed in this paper is used for training, which can quickly and accurately obtain the feasible domain of locomotive parameter configuration; finally, the digital twin technology is used to integrate the vehicle system, which can provide comprehensive support for the design, operation and maintenance of fuel cell locomotives and improve the overall performance and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic flow chart of a forward design and system integration method for a fuel cell hybrid locomotive according to the present invention;

[0075] Figure 2 Schematic diagram of a generative adversarial neural network model based on transfer learning in an embodiment of the present invention;

[0076] Figure 3 It is a structural diagram of a forward design method for a fuel cell locomotive in an embodiment of the present invention;

[0077] Figure 4 Schematic diagram of a control system of a hydrogen fuel cell hybrid locomotive in an embodiment of the present invention. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described below with reference to the accompanying drawings.

[0079] In this embodiment, see Figure 1 As shown, the present invention proposes a forward design and system integration method for a fuel cell hybrid locomotive, comprising the steps of:

[0080] S100: Based on the locomotive operation data and line conditions, combined with the constraints including locomotive space constraints, axle weight constraints and bus voltage constraints, a typical operating condition model of locomotive acceleration start, maximum speed and uniform speed driving, low speed and uniform speed climbing and emergency rescue is established, the feasible domain of forward integration of the fuel cell train hybrid power system is solved, and all feasible power system and hydrogen storage system configuration combinations are established;

[0081] S200: Based on the parameter feasible domain obtained in step S100 and in combination with the locomotive operation strategy, the fuel cell locomotive is forward designed, the power demand of the configured locomotive is predicted, the purchase cost, operating hydrogen consumption cost and degradation cost of each configured component in the locomotive are calculated, and then the total cost of the whole life cycle is obtained by weighting, and the optimal parameter combination is determined with the lowest total cost;

[0082] S300: Based on the parameter combination determined in step S200, the locomotive system is integrated using the digital twin.

[0083] As an optimization solution of the above embodiment, the constraints mentioned in step S100 are:

[0084] Locomotive space constraints and axle weight constraints:

[0085]

[0086] Bus voltage constraints:

[0087]

[0088] Where N fc is the number of fuel cells; N sbat With N pbat are the number of lithium batteries connected in series and in parallel respectively; k fc , k bat , k H2_TANK are the additional mass volume coefficients of fuel cells, lithium batteries and hydrogen storage systems; M FC_S 、M BAT_S 、M H2_TANK are the masses of the single fuel cell, lithium battery and hydrogen storage system respectively; V FC_S 、V BAT_S 、V H2_TANK are the volumes of the single fuel cell, lithium battery and hydrogen storage system respectively; P , E , M , V They are power, energy, mass and volume redundancy coefficients respectively; M max_axie V is the train axle weight limit; max is the train volume limit; U BAT_S is the voltage of a single lithium battery; U busis the bus voltage.

[0089] In step S100, a typical operating condition model of locomotive acceleration start, maximum speed and uniform speed driving, low speed and uniform speed climbing and emergency rescue is established:

[0090] (1) Locomotive acceleration start:

[0091] When a train is running on a straight track at maximum acceleration, it needs to overcome running resistance including basic resistance and acceleration resistance. The power and energy calculation of this process is expressed as:

[0092]

[0093] In the formula, F A , P A 、E A are the traction force, traction power and traction energy of the train during acceleration; M total is the total mass of the train; v turn is the turning speed of the train. When the train speed is lower than the turning speed, the train can maintain a uniform acceleration state. In this state, the acceleration is a i Indicates that when the speed is higher than the turning speed, the traction motor power has reached the maximum value P A,vturn , the train is in a uniform power running state, in which the acceleration is a v Indicates; η M is the traction motor efficiency; γ is the train turning coefficient; F fb is the basic resistance of the train; v is the speed of the train, t is the time, t a The time required for the train to accelerate to its maximum speed;

[0094] (2) Maximum speed for driving at a constant speed on a straight road:

[0095] When the train is on a straight track and running at a constant speed at the highest speed, the train traction system only needs to overcome the basic resistance. The power and energy calculation of this process is expressed as:

[0096]

[0097] In the formula, F C , P C 、E C are the traction force, traction power and traction energy of the train at a constant speed; x e Designing driving range for fuel cell trains;

[0098] (3) Slow and steady speed climbing:

[0099] When the urban train is climbing a slope, a higher traction force is required. The traction power and energy required for the urban train running on a slope are calculated as follows:

[0100]

[0101] In the formula, F R , P R 、E R are the traction force, traction power and traction energy of the train when climbing; F gb is the slope resistance of the train, P TR is the climbing power of the train, F fb is the basic resistance of train operation, t r is the climbing time of the train;

[0102] (4) Emergency rescue conditions

[0103] When the fuel cell stops supplying power due to a fault, the lithium battery will power the entire vehicle and drive it to the nearest rescue station. Considering the straight section and the slope section, the traction energy demand of the fuel cell urban train under emergency rescue conditions is calculated according to the following formula:

[0104]

[0105] In the formula, S tot and S scope Represent the total mileage of self-rescue and the length of the ramp respectively; v max and v scope Respectively represent the maximum operating speed and ramp operating speed; E E The energy required for fuel cell trains in emergency rescue conditions.

[0106] See also Figure 2 The transfer learning-based generative adversarial neural network model shown in the figure is different from the traditional generative adversarial network. This method replaces random noise with target domain data to avoid the sample generation process. The generator does not need to generate new samples, but is used to extract features. At the same time, in order to avoid the gradient loss problem caused by the large difference in feature size, the LRelu activation function can be used in the convolution operation to replace random noise with target domain data. The LRelu activation function is used in the convolution operation. The formula is:

[0107]

[0108] x is the neuron input.

[0109] The objective function of the adversarial neural network model includes classification loss and domain discriminator loss, expressed as:

[0110]

[0111] In the formula, θ f ,θ d ,θ yare the feature extractor, classifier and domain discriminator parameters, respectively.

[0112] Define cross entropy as the loss function for classification loss in source domain data, the formula is:

[0113]

[0114] In the formula, is the source domain data distribution; f f (x i ) is the feature obtained after the source domain data passes through the feature extractor; g y (f f (x i )) is the classifier category probability; y i For sample x i , when the category estimate is equal to x i If the true values ​​are the same, it is 1; if they are different, it is 0. Represents the data point x i Expected value;

[0115] The domain discriminator loss is expressed as:

[0116]

[0117] In the formula, is the target domain data distribution; g d (f f (x i )) is the output of the domain discriminator; d i is the binary label value of the i-th sample.

[0118] After training with the above neural network, the locomotive forward integration feasible domain can be obtained. Different from the traditional locomotive parameter matching method which only considers the purchase cost when configuring the locomotive, a fuel cell locomotive forward design method is proposed in step S200, which can calculate the operating cost when configuring the locomotive, and further improve the economic efficiency of locomotive configuration. The structure of the fuel cell locomotive forward design method is as follows: Figure 3 shown.

[0119] Forward design of fuel cell locomotives includes an inner loop of energy management and an outer loop of optimized matching.

[0120] The energy management inner loop is:

[0121] See also Figure 4 , according to the control system logic analysis, establish the train power demand model:

[0122]

[0123] In the formula, K 1 , K2 are the state feedback gain vector, K 1 =[k 1 ,k 2 ,k 3 ] T , K 2 =[k 4 ,k 5 ,k 6 ] T ;q 1 ,q 2 are the feedforward gains; c f 、c b are the component controller gains, representing the information transfer between controllers; P rf , P rb , P rw are the power requirements of the fuel cell, lithium battery and vehicle respectively; P f , P b , P w are the actual output power of the fuel cell, lithium battery and vehicle respectively; η f , η b are the efficiencies of fuel cells and lithium batteries respectively;

[0124] State space variable x = [x 1 ,x 2 ,x 3 ] T , specifically:

[0125]

[0126] In the formula, SOC k is the SOC reference value of lithium battery, t f is the train arrival time;

[0127] For ease of calculation, it is assumed that the SOC change rate of the lithium battery is linearly related to the charging and discharging power of the lithium battery, expressed as:

[0128] ΔSOC=-b h P rb ;

[0129] Where b h These are parameters related to the battery hardware system and are determined by the system itself;

[0130] Preferably, in the energy management inner loop, it is assumed that the control system pole is p 1 、p 2 、p 3 , then the controller feedback gain obtained by the feedback control system pole configuration is:

[0131]

[0132] b h is the battery hardware system related parameter; from the above formula, we can know that k 4 , k 5 By k 1 , k 2 Linear representation, and k 6 =0, so only k needs to be optimized 1 , k 2 , k 3 That's it.

[0133] In the inner loop of energy management, the objective function is established according to the power following cumulative error, fuel cell hydrogen consumption and lithium battery equivalent hydrogen consumption:

[0134]

[0135] In the formula, C fc (t) is the real-time hydrogen consumption of the fuel cell; C bat is the equivalent hydrogen consumption of lithium battery; β is the penalty factor, SOC t is the lithium battery SOC at the end time; ΔP and ΔSOC kt are the errors of train power and lithium battery SOC at time t respectively;

[0136] In order to ensure that the pole pi of the closed-loop system is always located in the left half plane, and the output power of the fuel cell and lithium battery and the SOC of the lithium battery must meet the variation range, the constraint expression of the system is:

[0137]

[0138] real(p i ) represents the real part of the pole pi; P fmin With P fmax are the upper and lower limits of fuel cell output power respectively; P bmin With P bmax They are the upper and lower limits of lithium battery output power; SOC min With SOC max They are the upper and lower limits of lithium battery SOC; SOC tmin With SOC tmax They are the upper and lower limits of the lithium battery SOC at the end of the operation.

[0139] The optimization objectives of the outer loop of the optimization matching include operating cost and acquisition cost. Therefore, the optimization matching problem is described as:

[0140] minCost total (X) = min[Cost g (X)+Cost y (X)N num Nyear ];

[0141] In the formula, X is the configuration combination in the feasible domain; Cost g Cost y are the purchase cost and operation cost of the locomotive respectively. total Configure the total cost for the train; N num With N year They are the annual operating conditions and service life of the locomotive, and the optimal configuration result is selected according to the actual operation requirements of the line.

[0142] As an optimization scheme of the above embodiment, in step S300, after completing the locomotive configuration selection, in order to facilitate the coordination between the various systems of the locomotive, reduce the R&D cycle and cost between prototype design and actual production, and realize the modular design of hydrogen locomotives, the digital twin system is applied to the system integration of hydrogen locomotives. The digital twin system architecture includes a data acquisition module, a physical field calculation module, and a visualization display module;

[0143] (1) Data acquisition module: Use measurement equipment to systematically measure the physical hardware in the locomotive system, process and exchange the collected data through the processing interface using a standard consistent protocol and format, and output data in the same format; after standardizing the data interface and format through this module, the data transmission efficiency can be effectively improved and the stable operation of the platform can be ensured. At the same time, in order to ensure the real-time and accuracy of the digital twin system, the data acquisition module needs to select a suitable network transmission protocol to ensure the transmission efficiency during transmission and the stability and integrity of the data.

[0144] (2) Physical field calculation module: It is the core part of the digital twin system. It constructs and calculates the digital twin model based on the physical mechanism and formulas through the data imported from the data acquisition module, and then calculates the corresponding results. Based on the calculation results, the data can be uploaded to the data cloud for subsequent use.

[0145] (3) Visualization display module: After the digital twin calculation results are output to the data cloud, the visualization platform imports the data, and the display module reads the calculation results from the platform and rematches them.

[0146] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A forward design and system integration method for a fuel cell hybrid locomotive, characterized in that: Includes steps: S100: Based on the locomotive operation data and line conditions, combined with the constraints including locomotive space constraints, axle weight constraints and bus voltage constraints, a typical operating condition model of locomotive acceleration start, maximum speed and uniform speed driving, low speed and uniform speed climbing and emergency rescue is established, the feasible domain of forward integration of the fuel cell train hybrid power system is solved, and all feasible power system and hydrogen storage system configuration combinations are established; S200: Based on the parameter feasible domain obtained in step S100 and in combination with the locomotive operation strategy, the fuel cell locomotive is forward designed, the power demand of the configured locomotive is predicted, the purchase cost, operating hydrogen consumption cost and degradation cost of each configured component in the locomotive are calculated, and then the total cost of the whole life cycle is obtained by weighting, and the optimal parameter combination is determined with the lowest total cost; S300: Based on the parameter combination determined in step S200, the locomotive system is integrated using the digital twin.

2. A fuel cell hybrid locomotive forward design and system integration method according to claim 1, characterized in that: The constraints mentioned in step S100 are: Locomotive space constraints and axle weight constraints: Bus voltage constraints: Where N fc is the number of fuel cells; N sbat With N pbat are the number of lithium batteries connected in series and in parallel respectively; k fc , k bat , k H2_TANK are the additional mass volume coefficients of fuel cells, lithium batteries and hydrogen storage systems; M FC_S 、M BAT_S 、M H2_TANK are the masses of the single fuel cell, lithium battery and hydrogen storage system respectively; V FC_S 、V BAT_S 、V H2_TANK are the volumes of the single fuel cell, lithium battery and hydrogen storage system respectively; P , E , M , V They are power, energy, mass and volume redundancy coefficients respectively; M max_axie V is the train axle weight limit; max is the train volume limit; U BAT_S is the voltage of a single lithium battery; U bus is the bus voltage.

3. A fuel cell hybrid locomotive forward design and system integration method according to claim 1, characterized in that: In step S100, a typical operating condition model of locomotive acceleration start, maximum speed and uniform speed driving, low speed and uniform speed climbing and emergency rescue is established: (1) Locomotive acceleration start: When a train is running on a straight track at maximum acceleration, it needs to overcome running resistance including basic resistance and acceleration resistance. The power and energy calculation of this process is expressed as: In the formula, F A , P A 、E A are the traction force, traction power and traction energy of the train during acceleration; M total is the total mass of the train; v turn is the turning speed of the train. When the train speed is lower than the turning speed, the train maintains a uniformly accelerated state. In this state, the acceleration is a i Indicates that when the speed is higher than the turning speed, the traction motor power has reached the maximum value P A,vturn , the train is in a uniform power running state, in which the acceleration is a v Indicates; η M is the traction motor efficiency; γ is the train turning coefficient; F fb is the basic resistance of the train; v is the speed of the train, t is the time, t a The time required for the train to accelerate to its maximum speed; (2) Maximum speed for driving at a constant speed on a straight road: When the train is on a straight track and running at a constant speed at the highest speed, the train traction system only needs to overcome the basic resistance. The power and energy calculation of this process is expressed as: In the formula, F C , P C 、E C are the traction force, traction power and traction energy of the train at a constant speed; x e Designing driving range for fuel cell trains; (3) Slow and steady speed climbing: When the urban train is climbing a slope, a higher traction force is required; the traction power and energy required for the urban train running on a slope are calculated as follows: In the formula, F R , P R 、E R are the traction force, traction power and traction energy of the train when climbing; F gb is the slope resistance of the train, P TR is the train climbing power, F fb is the basic resistance of train operation, t r is the climbing time of the train; (4) Emergency rescue conditions When the fuel cell stops supplying power due to a fault, the lithium battery will power the vehicle and drive it to the nearest rescue station. Considering the straight sections and slope sections, the traction energy demand of fuel cell urban trains under emergency rescue conditions is calculated according to the following formula: In the formula, S tot and S scope Represent the total mileage of self-rescue and the length of the ramp respectively; v max and v scope Respectively represent the maximum operating speed and ramp operating speed; E E The energy required for fuel cell trains in emergency rescue conditions.

4. A fuel cell hybrid locomotive forward design and system integration method according to claim 1, characterized in that: Step S100 needs to be trained using a generative adversarial network model based on transfer learning. To improve the performance of the model, random noise is replaced with target domain data, and the LRelu activation function is used in the convolution operation. The formula is: x is the neuron input.

5. A fuel cell hybrid locomotive forward design and system integration method according to claim 4, characterized in that: The objective function of the adversarial neural network model includes classification loss and domain discriminator loss, expressed as: In the formula, θ f ,θ d ,θ y are the feature extractor, classifier and domain discriminator parameters respectively; Define cross entropy as the loss function for classification loss in source domain data, the formula is: In the formula, is the source domain data distribution; f f (x i ) is the feature obtained after the source domain data passes through the feature extractor; g y (f f (x i )) is the classifier category probability; y i For sample x i , when the category estimate is equal to x i If the true values ​​are the same, it is 1; if they are different, it is 0. Represents the data point x i Expected value; The domain discriminator loss is expressed as: In the formula, is the target domain data distribution; g d (f f (x i )) is the output of the domain discriminator; d i is the binary label value of the i-th sample.

6. A fuel cell hybrid locomotive forward design and system integration method according to claim 1, characterized in that: Forward design of fuel cell locomotives includes an inner loop of energy management and an outer loop of optimized matching.

7. A fuel cell hybrid locomotive forward design and system integration method according to claim 6, characterized in that: The energy management inner loop is: According to the control system logic analysis, the train power demand model is established: Where K1 and K2 are state feedback gain vectors, K1 = [k1, k2, k3] T , K2=[k4,k5,k6] T ; q1, q2 are feedforward gains respectively; c f 、c b are the component controller gains, representing the information transfer between controllers; P rf , P rb , P rw are the power requirements of the fuel cell, lithium battery and vehicle respectively; P f , P b , P w They are the fuel cell, lithium battery and actual output power of the vehicle; η f , η b are the efficiencies of fuel cells and lithium batteries respectively; State space variable x = [x1, x2, x3] T , specifically: In the formula, SOC k is the SOC reference value of lithium battery, t f is the train arrival time; For ease of calculation, it is assumed that the SOC change rate of the lithium battery is linearly related to the charging and discharging power of the lithium battery, expressed as: ΔSOC=-b h P rb ; Where b h These are parameters related to the battery hardware system and are determined by the system itself; The optimization objectives of the outer loop of the optimization matching include operating cost and acquisition cost. Therefore, the optimization matching problem is described as: minCost total (X)=min[Cost g (X)+Cost y (X)N num N year ]; In the formula, X is the configuration combination in the feasible domain; Cost g Cost y are the purchase cost and operation cost of the locomotive respectively. total Configure the total cost for the train; N num With N year They are the annual operating conditions and service life of the locomotive, and the optimal configuration result is selected according to the actual operation requirements of the line.

8. A fuel cell hybrid locomotive forward design and system integration method according to claim 7, characterized in that: In the energy management inner loop, assuming that the control system poles are p1, p2, and p3, the controller feedback gain obtained by configuring the feedback control system poles is: b h is the battery hardware system parameter; It can be seen from the above formula that k4 and k5 are linearly represented by k1 and k2, and k6=0, so only k1, k2, and k3 need to be optimized.

9. A fuel cell hybrid locomotive forward design and system integration method according to claim 7, characterized in that: In the inner loop of energy management, the objective function is established according to the power following cumulative error, fuel cell hydrogen consumption and lithium battery equivalent hydrogen consumption: In the formula, C fc (t) is the real-time hydrogen consumption of the fuel cell; C bat is the equivalent hydrogen consumption of lithium battery; β is the penalty factor, SOC t is the lithium battery SOC at the end time; ΔP and ΔSOC kt are the errors between the train power and the lithium battery SOC at time t; SOC k is the SOC reference value of lithium battery; SOCt is the actual SOC value of the lithium battery at time t. In order to ensure that the pole pi of the closed-loop system is always located in the left half plane, and the output power of the fuel cell and lithium battery and the SOC of the lithium battery must meet the variation range, the constraint expression of the system is: real(p i ) represents the real part of the pole pi; P fmin With P fmax are the upper and lower limits of fuel cell output power respectively; P bmin With P bmax They are the upper and lower limits of lithium battery output power; SOC min With SOC max They are the upper and lower limits of lithium battery SOC; SOC tmin With SOC tmax They are the upper and lower limits of the lithium battery SOC at the end of the operation.

10. A fuel cell hybrid locomotive forward design and system integration method according to claim 1, characterized in that: In step S300, the digital twin system is applied to the hydrogen locomotive system integration, and the digital twin system architecture includes a data acquisition module, a physical field calculation module and a visualization display module; (1) Data acquisition module: Use measurement equipment to systematically measure the physical hardware in the locomotive system, process and exchange the collected data through the processing interface using a standard consistent protocol and format, and output data in the same format; (2) Physical field calculation module: It is the core part of the digital twin system. It constructs and calculates the digital twin model based on the physical mechanism and formulas through the data imported from the data acquisition module, so as to calculate the corresponding results; (3) Visualization display module: After the digital twin calculation results are output to the data cloud, the visualization platform imports the data, and the display module reads the calculation results from the platform and rematches them.

Images