Quantum strategy learning energy management method of multi-stack fuel cell hybrid power system

Abstract

The invention belongs to the technical field of energy management of hybrid power systems, and discloses a quantum strategy learning energy management method and system for a multi-stack fuel cell hybrid power system, and the method comprises the steps: building a multi-stack fuel cell hybrid power system model which comprises a fuel cell power model, a lithium battery equivalent circuit model and power balance; constructing a quantum hybrid neural network, interacting with the established hybrid power system environment, distributing output power of different power sources, and constructing a state-action-reward set; the strategy is evaluated and updated on the basis of the collected state-action-reward data set, and then the energy management strategy is optimized. By fusing the variable component sub-circuit and the traditional neural network, efficient operation of the fuel cell system is ensured while the state of charge of the lithium battery is maintained in an ideal interval; and the gradient of the quantum circuit parameters is accurately calculated by using a parameter transfer rule, so that the optimization process of the quantum strategy learning energy management strategy is more accurate and efficient.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy management of hybrid power systems, and in particular relates to a quantum strategy learning energy management method and system for a multi-stack fuel cell hybrid power system. Background Art

[0002] As energy consumption and environmental pollution become increasingly serious, traditional means of transportation that rely on fossil fuels can no longer meet the needs of modern society. In this context, hybrid power systems, especially those that combine proton exchange membrane fuel cells and lithium batteries, have shown great potential. This system can not only effectively reduce carbon emissions, but also improve energy utilization, and is one of the key technologies for achieving sustainable transportation solutions.

[0003] Current energy management strategies mainly rely on traditional control theories and preset rules. Although these methods can achieve basic energy management and distribution to a certain extent, their performance is often unsatisfactory when faced with complex and dynamically changing driving environments. On the one hand, these strategies are difficult to accurately predict and respond to future energy needs and environmental changes in real time, resulting in low energy utilization efficiency and failure to maximize the potential of high energy density and high power density of hybrid power systems. On the other hand, these methods often lack sufficient flexibility and adaptability, making it difficult to maintain optimal energy distribution under different operating conditions and environmental conditions, affecting the overall performance and economic benefits of the system.

[0004] Especially for complex hybrid systems that include multiple fuel cells and lithium batteries, how to accurately control and optimize the coordination between different energy sources to cope with the ever-changing load demand has become the key to improving system performance. In existing technologies, the design of energy management strategies often fails to fully consider the system dynamics and uncertainty, resulting in inefficient energy distribution, which in turn affects the economy and reliability of the vehicle.

[0005] Therefore, developing a new strategy that can achieve efficient energy management while being highly adaptable and responsive is of great significance for improving the performance and environmental adaptability of heavy-duty hybrid trucks.

[0006] Through the above analysis, the problems and defects of the prior art are as follows:

[0007] (1) Existing technologies have low energy efficiency and lack sufficient flexibility and adaptability;

[0008] (2) The design of existing energy management strategies often fails to fully consider the system dynamics and uncertainty, resulting in inefficient energy distribution, which in turn affects the economy and reliability of the entire vehicle. Summary of the invention

[0009] In view of the problems existing in the prior art, the present invention provides a quantum strategy learning energy management method and system for a multi-stack fuel cell hybrid power system.

[0010] The present invention is implemented as follows: a quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system, the quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system comprising:

[0011] Step 1: Establish a multi-stack fuel cell hybrid power system model, including a fuel cell power model, a lithium battery equivalent circuit model, and power balance;

[0012] Step 2: construct a quantum hybrid neural network to interact with the established hybrid power system environment, allocate the output power of different power sources and construct a state-action-reward set;

[0013] In step three, the strategy is evaluated and updated based on the state-action-reward dataset collected in step two, thereby optimizing the energy management strategy.

[0014] Furthermore, the steps of establishing the hybrid power system model are as follows:

[0015] (1) Establishment of fuel cell model:

[0016]

[0017] Where V out and E represent the output voltage and electromotive force of the fuel cell respectively, V act 、V ohm and V conc are activation loss, ohmic loss and concentration loss respectively; T is the operating temperature of the fuel cell; F is the Faraday constant; C O2 is the concentration of oxygen in the fuel cell stack; i represents the current; R is the gas constant; P O2 and P H2 Represent the partial pressures of oxygen and hydrogen respectively; ξ1, ξ2, ξ3, ξ4, m, n are all empirical coefficients; R mem is the internal resistance of the fuel cell; j represents the current density; and e is an exponential function.

[0018] (2) Lithium battery equivalent circuit model:

[0019]

[0020] Where U ocv is the open circuit voltage of the lithium battery; R bat is the internal resistance of the lithium battery; U bat is the potential difference between the two ends of the battery; P bat is the output power of the lithium battery; Ibat is the output current of the lithium battery; Q max is the maximum capacity of the lithium battery; SOC and SOC0 are the state of charge of the lithium battery and the initial state of charge of the lithium battery respectively;

[0021] (3) Establishment of hybrid power system model:

[0022] P load =η fc ·P fc +η bat ·P bat (3)

[0023] Where P load is the load demand power; P fc is the fuel cell output power; P bat is the output power of the lithium battery; η fc is the efficiency of the fuel cell auxiliary equipment and converter after losses; η bat is the efficiency of the lithium battery through the converter;

[0024] (4) Energy management target modeling:

[0025] The energy management goal is to optimize the power output of fuel cells and lithium batteries while ensuring efficient operation of the system. It can be specifically expressed by the following optimization objective formula:

[0026]

[0027] Among them, η is the overall efficiency of the system, and the following constraints must also be met:

[0028] SOC min ≤SOC≤SOC max

[0029] P fc,min ≤P fc ≤P fc,max

[0030] P bat,min ≤P bat ≤P bat,max (5)

[0031] Among them, SOC min and SOC max Respectively represent the minimum and maximum charge states of lithium batteries; P fc,min , P fc,max

[0032] , P bat,min and P bat,max Represent the minimum and maximum output power limits of fuel cells and lithium batteries respectively.

[0033] Furthermore, the construction of the quantum hybrid neural network includes the following sub-steps:

[0034] (1) Constructing a variational quantum circuit (VQC) model:

[0035] The system state is encoded into the quantum state through angle encoding technology, the encoded quantum state is processed using VQC, and the expectation of the observed value is obtained through quantum measurement. This process can be expressed as:

[0036]

[0037] Among them, |ψ(θ)> is the quantum state after VQC processing, θ represents the variational parameter, is the observation operator, usually the Pauli-Z operator;

[0038] (2) Design subsequent neural network layers:

[0039] The expectation value of the quantum measurement is fed into a single-layer conventional neural network, which outputs a normalized representation of the power output value that matches the number of fuel cells, which can be approximated by the following formula:

[0040]

[0041] Among them, P out,norm =[P out,1,norm ,P out,2,norm ,...,P out,N,norm ] represents the normalized multi-stack fuel cell output power, N is the number of fuel cells, here it is 3, and P fc =P fc,max ·P out,norm , σ is the activation function, is the weight matrix, b∈R 1×N is the bias matrix, which maps the measurement expectation of the qubit <ψ> to the output action, n q is the number of qubits.

[0042] Furthermore, the energy management strategy comprises the following steps:

[0043] (1) Collecting load demand power P load ,The system monitors and records the energy demand of the vehicle in real time, ensuring that the energy allocation strategy can accurately match the actual usage;

[0044] (2) According to the current state of the multi-stack fuel cell hybrid system, including the load demand P load and the state of charge (SOC) of the lithium battery, using a quantum hybrid neural network to determine the optimal output power of each fuel cell;

[0045] (3) Multiple interactions between multi-stack fuel cell hybrid power system environment and quantum hybrid neural network decision-making to collect state-action-reward data sets.

[0046] Furthermore, the evaluation and update strategy constructs a traditional neural network as a strategy evaluator to achieve the evaluation of strategy performance and iterative update of parameters, evaluates the network input environment state and the output of the quantum hybrid neural network as input, outputs a predicted value to evaluate the current strategy, and then updates the parameters of the variational quantum circuit and the neural network, thereby optimizing the energy management strategy. The specific steps are as follows:

[0047] (1) Strategy evaluation network construction:

[0048] A traditional neural network is constructed as a strategy evaluator, which receives the current state of the hybrid system (lithium battery SOC and load power P load ) and the action information output by the quantum hybrid network as input, and the output value reflects the expected benefit of the current energy management strategy:

[0049] V=f eval (s,a;θ eval )(8)

[0050] Among them, V represents the value output, s represents the current state of the system, a represents the action information, and f eval represents the functional form of the evaluation network, θ eval Represents the parameters of the evaluation network;

[0051] (2) Quantum hybrid neural network update:

[0052] Based on the value of the evaluation network output, the gradient descent algorithm is used to update the parameters of the quantum hybrid network to maximize the expected benefits. The gradient of the quantum hybrid neural network is calculated by evaluating the output strategy of the quantum hybrid neural network through the value network, and then the traditional network of the hybrid network is back-propagated according to the gradient; the back-propagated gradient is transferred to the variational quantum circuit in the form of parameter transfer rules, and then the entire hybrid quantum neural network is updated to optimize the energy management strategy. The parameter update of the variational quantum circuit follows the parameter transfer rule, that is, the variational parameters of the quantum circuit are adjusted based on the evaluation results to optimize the circuit output:

[0053]

[0054] Among them, θ i Represents a tuning parameter in a quantum circuit, usually related to some operation of a qubit, such as the angle of a revolving door; Indicates that when θ i When the quantum circuit is offset by π / 2 and -π / 2, the observed operator is The expected value of the parameter transfer rule is calculated by measuring the quantum circuit twice. i The gradient of , and then gradient descent or other optimization algorithms can be applied to adjust the parameters of the quantum circuit.

[0055] Another object of the present invention is to provide a quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system, the quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system comprising:

[0056] System model building module, used to build a multi-stack fuel cell hybrid system model;

[0057] Neural network building blocks for building quantum hybrid neural networks;

[0058] The strategy update module is used to evaluate and update the strategy and optimize the energy management strategy.

[0059] Furthermore, in the quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system, the fuel cell is connected in parallel to the DC bus of a multi-stack fuel cell hybrid heavy truck through a unidirectional DC / DC converter, and the lithium battery is connected in parallel to the DC bus of the multi-stack fuel cell hybrid power heavy truck through a bidirectional DC / DC converter; the load demand power is provided by the multi-stack fuel cells, and the remaining power is absorbed or supplemented by the lithium battery.

[0060] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the quantum strategy learning energy management method for a multi-stack fuel cell hybrid system.

[0061] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the quantum strategy learning energy management method for a multi-stack fuel cell hybrid system.

[0062] Another object of the present invention is to provide an information data processing terminal, which is used to implement the quantum strategy learning energy management system for a multi-stack fuel cell hybrid system.

[0063] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0064] First, in view of the technical problems existing in the above-mentioned prior art, some creative technical effects are brought about after solving the problems. The specific description is as follows:

[0065] (1) The present invention integrates variational quantum circuits (VQC) and traditional neural networks to ensure that the state of charge (SOC) of the lithium battery is maintained in an ideal range while ensuring efficient operation of the fuel cell system;

[0066] (2) This scheme provides a fast and efficient energy management strategy through the high efficiency of quantum computing and the learning ability of neural networks. Compared with the finite state machine strategy, which is also a real-time control, it has the advantages mentioned in (1);

[0067] (3) By utilizing the quantum entanglement and superposition properties, complex energy management strategies can be learned using only a small number of network parameters, allowing quantum strategies to achieve more complex decisions with limited storage space.

[0068] Second, the present invention proposes a quantum strategy learning energy management system for multi-stack fuel cell hybrid systems, which realizes an innovative energy management strategy by combining variational quantum circuits (VQC) and traditional neural network technology. From the perspective of the product, this technical solution has the following significant technical effects and advantages:

[0069] 1. Optimize energy distribution: Through efficient quantum computing and the learning ability of neural networks, the present invention achieves dual optimization of the state of charge (SOC) of the lithium battery and the operating efficiency of the fuel cell system. This not only improves the overall performance of the fuel cell hybrid system, but also extends the service life of the fuel cell and lithium battery.

[0070] 2. Efficient energy management strategy: The energy management strategy provided by the present invention is more efficient and faster than the traditional finite state machine strategy. This is because the entanglement and superposition characteristics of quantum computing allow the use of fewer network parameters to learn and implement complex energy management strategies, thereby showing excellent performance in real-time control.

[0071] 3. Reduced storage requirements: By leveraging the characteristics of quantum computing, the present invention can achieve the learning and execution of complex decisions with limited storage space. This means that compared with traditional methods, the present invention can deploy more complex and efficient energy management strategies in a smaller physical space.

[0072] 4. Improve overall system efficiency: The present invention significantly improves the overall energy efficiency of the hybrid system by ensuring that the fuel cell system and lithium battery operate in optimal working conditions. This optimization not only reduces energy waste, but also helps reduce emissions and is environmentally friendly.

[0073] 5. Innovative technology integration: This invention combines the advantages of quantum computing technology and artificial intelligence technology (i.e. neural network) to present a new solution approach. This integration not only improves the efficiency and effectiveness of energy management, but also provides new ideas and possibilities for future technological innovation.

[0074] In summary, the present invention provides an efficient, energy-saving and forward-looking solution for multi-stack fuel cell hybrid power systems through its innovative quantum strategy learning energy management system, which has important practical value and broad development prospects.

[0075] Third, as auxiliary evidence of the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0076] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation are:

[0077] The technical solution of the present invention directly reduces energy consumption and operating costs by improving the energy efficiency of the fuel cell hybrid system and extending the life of the lithium battery. In addition, its high efficiency and energy-saving characteristics are in line with the global trend of energy conservation and emission reduction, bringing significant environmental and social benefits to enterprises. These advantages are expected to attract widespread market interest, provide enterprises with competitive advantages, and thus create significant commercial value.

[0078] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0079] The present invention combines variational quantum circuit (VQC) and neural network technology to innovatively propose a new energy management strategy, which is the first in the field of fuel cell technology at home and abroad. The implementation of this technology not only improves the efficiency of energy management, but also optimizes the overall performance of the system, filling the technical gap in this field.

[0080] (3) The technical solution of the present invention solves the technical problems that people have been eager to solve but have never been able to solve successfully:

[0081] How to effectively manage the energy distribution of fuel cells and lithium batteries in hybrid systems to achieve efficient energy utilization and extend battery life has long been a difficult problem that engineers and scientists have tried to solve. This invention successfully solves this technical challenge through its innovative quantum strategy learning method, meeting people's long-standing expectations.

[0082] (4) The technical solution of the present invention overcomes technical prejudice:

[0083] Before this invention, many experts may have reservations about the application of quantum computing in actual energy management systems, believing that it is too complex and difficult to implement. This invention not only realizes the effective integration of quantum computing and traditional neural networks, but also proves that quantum strategies can be actually applied in energy management systems, effectively overcoming the previous technical prejudice.

[0084] Fourth, the quantum strategy learning energy management method for multi-stack fuel cell hybrid power systems proposed in this invention has achieved remarkable results in solving problems in existing industrial applications and promoting technological progress. The following is a detailed description of the technical problems it has solved and the remarkable technological progress achieved in industrial applications:

[0085] Technical issues solved:

[0086] 1. Optimization of complex energy management strategies: Multi-stack fuel cell hybrid systems involve multiple energy sources and complex energy conversion processes. Traditional energy management methods often find it difficult to achieve global optimization. This invention optimizes complex energy management strategies by constructing a quantum hybrid neural network and combining the parallelism and efficiency of quantum computing.

[0087] 2. Real-time dynamic energy distribution: The energy demand of the hybrid system changes with time and working conditions, and the output power of different energy sources needs to be adjusted in real time. The present invention uses the method of quantum strategy learning to achieve real-time perception of the system state and dynamic energy distribution.

[0088] 3. Improve system efficiency and stability: By optimizing the power output of fuel cells and lithium batteries, the present invention not only ensures efficient operation of the system, but also improves the stability of the system and extends the service life of the energy source.

[0089] Significant technological advances in industrial applications:

[0090] 1. Promote the application of quantum computing in the field of energy management: This invention combines quantum computing with the energy management of fuel cell hybrid power systems, provides new ideas and directions for the application of quantum computing in the field of energy management, and promotes the practical application and industrialization of quantum computing technology.

[0091] 2. Improve the performance of hybrid power systems: Through the energy management method of quantum strategy learning, the performance of hybrid power systems has been significantly improved, including improving energy utilization efficiency, reducing energy consumption, reducing emissions, etc., which will help promote the development of new energy vehicles and clean energy technologies.

[0092] 3. Reduce operating and maintenance costs: The optimized energy management strategy makes the hybrid system run more smoothly and efficiently, reduces the possibility of failure, and thus reduces operating and maintenance costs.

[0093] 4. Enhance system adaptability: The quantum strategy learning method enables the energy management system to adaptively adjust strategies to adapt to different working conditions and needs, thereby improving the flexibility and adaptability of the system.

[0094] In summary, the present invention successfully solves the complex problems in energy management of multi-stack fuel cell hybrid systems by introducing an energy management method based on quantum strategy learning, and has achieved significant technological progress in industrial applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 It is a flow chart of a quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system provided by an embodiment of the present invention;

[0096] Figure 2 It is a diagram of the structure of a quantum hybrid neural network provided by an embodiment of the present invention;

[0097] Figure 3 It is a structure diagram of a quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system provided by an embodiment of the present invention;

[0098] Figure 4 is the New European Driving Cycle (NEDC) operating condition provided by an embodiment of the present invention;

[0099] Figure 5 It is a diagram of the quantum hybrid neural network update result provided by an embodiment of the present invention;

[0100] Figure 6 is a schematic diagram of power distribution of different power sources provided by an embodiment of the present invention;

[0101] Figure 7 is a curve diagram of the change of SOC of a lithium battery for power distribution provided by an embodiment of the present invention;

[0102] Figure 8 It is a training curve diagram of the traditional neural network and quantum energy management strategy provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0103] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0104] 1. Explanatory Examples In order to enable those skilled in the art to fully understand how to implement the present invention, this section provides an illustrative example that expands and describes the technical solution of the claims.

[0105] like Figure 1 As shown, the quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system provided by an embodiment of the present invention includes the following steps:

[0106] S101, establishing a multi-stack fuel cell hybrid power system model, including a fuel cell power model, a lithium battery equivalent circuit model and power balance;

[0107] S102, constructing a quantum hybrid neural network to interact with the established hybrid power system environment, allocating the output power of different power sources and constructing a state-action-reward set;

[0108] S103, evaluating and updating the strategy based on the state-action-reward data set collected in step 2, and then optimizing the energy management strategy.

[0109] The steps of establishing a hybrid power system model provided by the embodiment of the present invention are as follows:

[0110] (1) Establishment of fuel cell model:

[0111]

[0112] Where V out and E represent the output voltage and electromotive force of the fuel cell respectively, V act 、V ohm and V conc are activation loss, ohmic loss and concentration loss respectively; T is the operating temperature of the fuel cell; F is the Faraday constant; C O2 is the concentration of oxygen in the fuel cell stack; i represents the current; R is the gas constant; P O2 and P H2 represent the partial pressures of oxygen and hydrogen respectively; ξ 1～4 And m, n are empirical coefficients; R mem is the internal resistance of the fuel cell; j represents the current density; and e is an exponential function.

[0113] (2) Lithium battery equivalent circuit model:

[0114]

[0115] Where U ocv is the open circuit voltage of the lithium battery; R bat is the internal resistance of the lithium battery; U bat is the potential difference between the two ends of the battery; P bat is the output power of the lithium battery; I bat is the output current of the lithium battery; Q max is the maximum capacity of the lithium battery; SOC and SOC0 are the state of charge of the lithium battery and the initial state of charge of the lithium battery respectively;

[0116] (3) Establishment of hybrid power system model:

[0117] P load =η fc ·P fc +η bat ·P bat (3)

[0118] Where P load is the load demand power; P fc is the fuel cell output power; P bat is the output power of the lithium battery; η fc is the efficiency of the fuel cell auxiliary equipment and converter after losses; η bat is the efficiency of the lithium battery through the converter;

[0119] (4) Energy management target modeling:

[0120] The energy management goal is to optimize the power output of fuel cells and lithium batteries while ensuring efficient operation of the system. It can be specifically expressed by the following optimization objective formula:

[0121]

[0122] Among them, η is the overall efficiency of the system, and the following constraints must also be met:

[0123] SOC min ≤SOC≤SOC max

[0124] P fc,min ≤P fc ≤P fc,max

[0125] P bat,min ≤P bat ≤P bat,max (5)

[0126] Among them, SOC min and SOC max Respectively represent the minimum and maximum state of charge of lithium batteries; P fc,min , P fc,max , P bat,min and P bat,max Represent the minimum and maximum output power limits of fuel cells and lithium batteries respectively.

[0127] Through the establishment of the above model and the definition of optimization objectives, the energy management strategy aims to dynamically adjust the output power of fuel cells and lithium batteries to adapt to different load requirements, while ensuring that the system operates at the optimal efficiency point to achieve efficient use of energy.

[0128] The method for constructing a quantum hybrid neural network provided in an embodiment of the present invention includes the following sub-steps:

[0129] (1) Construct a variational quantum circuit (VQC) model: Encode the system state into the quantum state through angle encoding technology. Use VQC to process the encoded quantum state and obtain the expectation of the observed value through quantum measurement. This process can be expressed as:

[0130]

[0131] Among them, |ψ(θ)> is the quantum state after VQC processing, θ represents the variational parameter, is the observation operator, and the Pauli-Z operator is usually chosen.

[0132] (2) Design of subsequent neural network layers: The expected value of the quantum measurement is input into a single-layer conventional neural network, which outputs a normalized representation of the power output value that matches the number of fuel cells. This can be approximated by the following formula:

[0133]

[0134] Among them, P out,norm =[P out,1,norm ,P out,2,norm ,...,P out,N,norm ] represents the normalized multi-stack fuel cell output power, N is the number of fuel cells, here it is 3, and P fc =P fc,max ·P out,norm , σ is the activation function, is a weight matrix that maps the measurement expectation of the qubit <ψ> to the output action, n q is the number of qubits.

[0135] The structure of quantum hybrid neural network is shown in the figure Figure 2 shown.

[0136] The energy management strategy provided by the embodiment of the present invention includes the following steps:

[0137] (1) Collecting load demand power P load ,The system monitors and records the energy demand of the vehicle in real time, ensuring that the energy allocation strategy can accurately match the actual usage;

[0138] (2) According to the current state of the multi-stack fuel cell hybrid system, including the load demand P load and the state of charge (SOC) of the lithium battery, using a quantum hybrid neural network to determine the optimal output power of each fuel cell;

[0139] (3) Multiple interactions between multi-stack fuel cell hybrid power system environment and quantum hybrid neural network decision-making to collect state-action-reward data sets.

[0140] The evaluation and update strategy provided by the embodiment of the present invention constructs a traditional neural network as a strategy evaluator to achieve the evaluation of strategy performance and iterative update of parameters. The evaluation network inputs the environmental state and the output of the quantum hybrid neural network as input, outputs a predicted value to evaluate the current strategy, and then updates the parameters of the variational quantum circuit and the neural network, thereby optimizing the energy management strategy. The specific steps are as follows:

[0141] (1) Strategy Evaluation Network Construction

[0142] A traditional neural network is constructed as a strategy evaluator, which receives the current state of the hybrid system (lithium battery SOC and load power P load ) and the action information output by the quantum hybrid network are taken as input, and the output value reflects the expected benefit of the current energy management strategy.

[0143] V=f eval (s,a;θ eval ) (8)

[0144] Among them, V represents the value output, s represents the current state of the system, a represents the action information, and f eval represents the functional form of the evaluation network, θ eval Represents the parameters of the evaluation network.

[0145] (2) Quantum Hybrid Neural Network Update

[0146] Based on the value of evaluating the network output, the gradient descent algorithm is used to update the parameters of the quantum hybrid network to maximize the expected benefits. The gradient of the quantum hybrid neural network is calculated by evaluating the output strategy of the quantum hybrid neural network through the value network, and then the traditional network of the hybrid network is updated based on the gradient back-propagation; the back-propagated gradient is transferred to the variational quantum circuit in the form of parameter transfer rules, and then the entire hybrid quantum neural network is updated to optimize the energy management strategy.

[0147] The parameter update of the variational quantum circuit follows the parameter transfer rule, that is, the variational parameters of the quantum circuit are adjusted based on the evaluation results to optimize the circuit output:

[0148]

[0149] in, Indicates that the parameter θ i When the quantum circuit is offset by π / 2 and -π / 2, the observed operator is The parameter transfer rule calculates the parameter θ by measuring the quantum circuit twice. i The gradient of , and then gradient descent or other optimization algorithms can be applied to adjust the parameters of the quantum circuit.

[0150] like Figure 3 As shown, the quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system provided by an embodiment of the present invention includes:

[0151] System model building module, used to build a multi-stack fuel cell hybrid system model;

[0152] Neural network building blocks for building quantum hybrid neural networks;

[0153] The strategy update module is used to evaluate and update the strategy and optimize the energy management strategy.

[0154] The quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system provided by an embodiment of the present invention is that the fuel cell is connected in parallel to the DC bus of a multi-stack fuel cell hybrid heavy truck through a unidirectional DC / DC converter, and the lithium battery is connected in parallel to the DC bus of the multi-stack fuel cell hybrid power heavy truck through a bidirectional DC / DC converter; the load demand power is provided by the multi-stack fuel cells, and the remaining power is absorbed or supplemented by the lithium battery.

[0155] The quantum strategy learning energy management method for a multi-stack fuel cell hybrid system provided by an application embodiment of the present invention is applied to a computer device, wherein the computer device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the quantum strategy learning energy management method for a multi-stack fuel cell hybrid system.

[0156] The quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system provided by an application embodiment of the present invention is applied to an information data processing terminal, and the information data processing terminal is used to implement the quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system.

[0157] Choose Figure 4 The New European Driving Cycle (NEDC) operating conditions shown in the figure are simulated and verified; the number of quantum strong entanglement gate layers selected is 2, which can be adjusted according to different systems; the number of neurons in each layer of the value network is [3+3]-[3+3]-3-1; the output power of the fuel cell and lithium battery are limited to 100W to 33kW and -50kW to 50kW respectively.

[0158] After completing the above work, simulate the invention. Figure 5For the training curve of the present invention under the described environment, 6 random seeds were randomly selected and their training reward curves were averaged. It was found that the selected random seeds can converge. Figure 6 The power distribution results of the fuel cell and lithium battery are shown as follows: Figure 7 As shown in the figure, the change trend of the lithium battery SOC is shown. Through analysis, it can be seen that the present invention effectively realizes the optimized power configuration of the multi-stack fuel cell hybrid power system. Such a configuration enables the fuel cell to operate in the high-efficiency area most of the time, avoiding drastic power fluctuations. At the same time, the SOC of the lithium battery can always be maintained within a healthy level range under the guidance of the energy management strategy of the present invention.

[0159] 3. Evidence of the effects of the embodiments. The embodiments of the present invention have achieved some positive effects during the development or use process, and indeed have great advantages over the prior art. The following content is described in conjunction with the data, charts, etc. of the test process.

[0160] Through observation of the development and use of embodiments of the present invention, some obvious positive effects can be seen. These effects are clearly demonstrated through experimental data and charts, especially when comparing the quantum energy management strategy of the present invention with the performance of traditional neural networks.

[0161] like Figure 8 As shown in the figure, the convergence behavior of traditional neural networks during the learning process. In many cases, traditional neural networks tend to show a certain degree of learning improvement in the early stages of training, but as the training progresses, its performance improvement tends to be flat, and even in some cases, performance declines, indicating that it has not achieved the ideal convergence state. The final experiment also proved that no effective energy management strategy has been learned. In contrast, the quantum energy management strategy shows remarkable stability and better convergence performance. Figure 8 It can be seen that the performance of the quantum energy management strategy steadily improves over time and eventually reaches a stable high reward level, which shows the superiority of the quantum energy management strategy in learning and optimizing strategies.

[0162] The robustness exhibited by the quantum energy management strategy may be due to the inherent characteristics of quantum computing, such as quantum entanglement and quantum superposition, which makes it more efficient in dealing with complex optimization problems. In addition, the quantum energy management strategy can achieve more complex decision-making processes with fewer network parameters, improving the quality of the learning process and the effectiveness of the final strategy. Therefore, the quantum energy management strategy of the present invention can not only converge to a better solution, but also the stability of its learning process also provides a more reliable and efficient method for the energy management of hybrid power systems. This can be clearly confirmed by the comparison of the above experimental data.

[0163] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware portion can be implemented using dedicated logic; the software portion can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. A person of ordinary skill in the art will appreciate that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0164] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A quantum strategy learning energy management method for a multi-stack fuel cell hybrid system, characterized in that: The quantum strategy learning energy management method for a multi-stack fuel cell hybrid system includes: Step 1: Establish a multi-stack fuel cell hybrid power system model, including a fuel cell power model, a lithium battery equivalent circuit model, and power balance; Step 2: construct a quantum hybrid neural network to interact with the established hybrid power system environment, allocate the output power of different power sources and construct a state-action-reward set; In step three, the strategy is evaluated and updated based on the state-action-reward dataset collected in step two, thereby optimizing the energy management strategy.

2. The quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system according to claim 1 is characterized in that: The steps of establishing the hybrid power system model are as follows: (1) Establishment of fuel cell model: Where V out and E represent the output voltage and electromotive force of the fuel cell respectively, V act 、V ohm and V conc are activation loss, ohmic loss and concentration loss respectively; T is the operating temperature of the fuel cell; F is the Faraday constant; C O2 is the concentration of oxygen in the fuel cell stack; i represents the current; R is the gas constant; P O2 and P H2 represent the partial pressures of oxygen and hydrogen respectively; ξ 1～4 And m, n are empirical coefficients; R mem is the internal resistance of the fuel cell; j represents the current density; and e is an exponential function. (2) Lithium battery equivalent circuit model: Where U ocv is the open circuit voltage of the lithium battery; R bat is the internal resistance of the lithium battery; U bat is the potential difference between the two ends of the battery; P bat is the output power of the lithium battery; I bat is the output current of the lithium battery; Q max is the maximum capacity of the lithium battery; SOC and SOC0 are the state of charge of the lithium battery and the initial state of charge of the lithium battery respectively; (3) Establishment of hybrid power system model: P load =the fc ·P fc +n bat ·P bat (3) Where P load is the load demand power; P fc is the fuel cell output power; P bat is the output power of the lithium battery; η fc It is the efficiency after the loss of fuel cell auxiliary equipment and converter; η bat is the efficiency of the lithium battery through the converter; (4) Energy management target modeling: The energy management goal is to optimize the power output of fuel cells and lithium batteries while ensuring efficient operation of the system. It can be specifically expressed by the following optimization objective formula: Among them, η is the overall efficiency of the system, and the following constraints must also be met: SOC min ≤SOC≤SOC max P fc,min ≤P fc ≤P fc,max P bat,min ≤P bat ≤P bat,max (5) Among them, SOC min and SOC max Respectively represent the minimum and maximum charge states of lithium batteries; P fc,min , P fc,max , P bat,min and P bat,max Represent the minimum and maximum output power limits of fuel cells and lithium batteries respectively.

3. The quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system according to claim 1 is characterized in that: The construction of the quantum hybrid neural network comprises the following sub-steps: (1) Constructing a variational quantum circuit (VQC) model: The system state is encoded into the quantum state through angle coding technology, the encoded quantum state is processed using VQC, and the expectation of the observed value is obtained through quantum measurement. This process can be expressed as: Among them, |ψ(θ)> is the quantum state after VQC processing, θ represents the variational parameter, is the observation operator, usually the Pauli-Z operator; (2) Design subsequent neural network layers: The expectation value of the quantum measurement is fed into a single-layer conventional neural network, which outputs a normalized representation of the power output value that matches the number of fuel cells, which can be approximated by the following formula: Among them, P out,norm =[P out,1,norm ,P out,2,norm ,...,P out,N,norm ] represents the normalized multi-stack fuel cell output power, N is the number of fuel cells, here it is 3, and P fc =P fc,max ·P out,norm , σ is the activation function, is the weight matrix, b∈R 1×N is the bias matrix, which maps the measurement expectation of the qubit <ψ> to the output action, n q is the number of qubits.

4. The quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system according to claim 1 is characterized in that: The energy management strategy includes the following steps: (1) Collecting load demand power P load ,The system monitors and records the energy demand of the vehicle in real time, ensuring that the energy allocation strategy can accurately match the actual usage; (2) According to the current state of the multi-stack fuel cell hybrid system, including the load demand P load and the state of charge (SOC) of the lithium battery, using a quantum hybrid neural network to determine the optimal output power of each fuel cell; (3) Multiple interactions between multi-stack fuel cell hybrid power system environment and quantum hybrid neural network decision-making to collect state-action-reward data sets.

5. The quantum strategy learning energy management method for a multi-stack fuel cell hybrid power system according to claim 1 is characterized in that: The evaluation and update strategy constructs a traditional neural network as a strategy evaluator to achieve the evaluation of strategy performance and iterative update of parameters. The evaluation network input environment state and the output of the quantum hybrid neural network are used as inputs, and a predicted value is output to evaluate the current strategy, thereby updating the parameters of the variational quantum circuit and the neural network to optimize the energy management strategy. The specific steps are as follows: (1) Strategy evaluation network construction: A traditional neural network is constructed as a strategy evaluator, which receives the current state of the hybrid system, the lithium battery SOC and the load power P load The action information output by the quantum hybrid network is used as input, and the output value reflects the expected benefit of the current energy management strategy: V=f eval (s,a;θ eval (8) Among them, V represents the value output, s represents the current state of the system, a represents the action information, and f eval represents the functional form of the evaluation network, θ eval Represents the parameters of the evaluation network; (2) Quantum hybrid neural network update: Based on the value of the evaluation network output, the gradient descent algorithm is used to update the parameters of the quantum hybrid network to maximize the expected benefits. The gradient of the quantum hybrid neural network is calculated by evaluating the output strategy of the quantum hybrid neural network through the value network, and then the traditional network of the hybrid network is back-propagated according to the gradient; the back-propagated gradient is transferred to the variational quantum circuit in the form of parameter transfer rules, and then the entire hybrid quantum neural network is updated to optimize the energy management strategy. The parameter update of the variational quantum circuit follows the parameter transfer rule, that is, the variational parameters of the quantum circuit are adjusted based on the evaluation results to optimize the circuit output: Among them, θ i Represents a tuning parameter in a quantum circuit, usually related to some operation of a qubit, such as the angle of a revolving door; Indicates that when θ i When the quantum circuit is offset by π / 2 and -π / 2, the observed operator is The expected value of the parameter transfer rule is calculated by measuring the quantum circuit twice. i The gradient of , and then gradient descent or other optimization algorithms can be applied to adjust the parameters of the quantum circuit.

6. A quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system according to any one of claims 1 to 5, characterized in that: The quantum strategy learning energy management system for a multi-stack fuel cell hybrid system includes: System model building module, used to build a multi-stack fuel cell hybrid system model; Neural network building blocks for building quantum hybrid neural networks; The strategy update module is used to evaluate and update the strategy and optimize the energy management strategy.

7. The quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system according to claim 6 is characterized in that: In the system, the fuel cell is connected in parallel to the DC bus of a multi-stack fuel cell hybrid heavy truck through a unidirectional DC / DC converter, and the lithium battery is connected in parallel to the DC bus of a multi-stack fuel cell hybrid heavy truck through a bidirectional DC / DC converter; the load demand power is provided by the multi-stack fuel cell, and the remaining power is absorbed or supplemented by the lithium battery.

8. A computer device, characterized in that: The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the quantum strategy learning energy management method for a multi-stack fuel cell hybrid system as described in any one of claims 1-5.

9. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the quantum strategy learning energy management method for a multi-stack fuel cell hybrid system as described in any one of claims 1 to 5.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement a quantum strategy learning energy management system for a multi-stack fuel cell hybrid power system as described in any one of claims 6-7.

Images