Self-adaptive multi-objective optimization method and device for distributed energy scheduling system

Abstract

The invention relates to the technical field of energy scheduling, and discloses a self-adaptive multi-objective optimization method and device for a distributed energy scheduling system, and the method comprises the steps: S1, taking the scheduling cost, the energy utilization efficiency and the system stability as optimization objectives, and constructing a comprehensive objective function of multi-objective optimization of the distributed energy scheduling system; s2, acquiring operation data of the distributed energy scheduling system; s3, according to the operation data, using a deep neural network to adjust the parameter weight of the comprehensive objective function, and using a genetic algorithm and a particle swarm optimization algorithm to update the comprehensive objective function to generate a candidate scheduling scheme; and S4, iteratively executing the steps S2 to S3 until the maximum number of iterations is reached or the comprehensive objective function is converged, and generating a final scheduling scheme. According to the method, the optimization efficiency and precision of the distributed energy scheduling system can be effectively improved, and the robustness of the algorithm is enhanced, so that efficient and stable scheduling optimization in the complex and changeable distributed energy system is ensured.

Description

Technical Field

[0001] The present invention relates to the field of energy dispatching technology, and more specifically, to an adaptive multi-objective optimization method and device for a distributed energy dispatching system. Background Art

[0002] With the acceleration of the transformation of the global energy structure, the proportion of renewable energy in the energy system has increased significantly, and distributed energy systems have become an important part of the modern energy structure. Distributed energy systems are composed of various forms of energy (such as solar energy, wind energy, energy storage devices, etc.), and are distributed and dynamic. Such systems must not only meet user load requirements, but also coordinate the balance between operating costs, energy efficiency and system stability. However, due to the intermittent nature of renewable energy, the uncertainty of power load, and the dynamic changes in the operating status of equipment, distributed energy scheduling faces great complexity and uncertainty, and effective optimization scheduling methods are urgently needed to ensure the efficient and stable operation of the system.

[0003] At present, the methods of distributed energy scheduling are mainly based on single-objective optimization, such as minimizing operating costs or maximizing energy utilization efficiency. These methods have achieved certain results in the application scenario of a single optimization objective, but it is difficult to meet the comprehensive needs of multiple objectives in distributed energy systems. In recent years, some studies have attempted to use multi-objective optimization algorithms, such as weighted optimization methods based on weight distribution or Pareto optimal solution methods, to achieve a trade-off between multiple optimization objectives. However, these methods usually lack adaptive mechanisms and cannot flexibly adjust the optimization strategy according to the dynamic changes in the system state, and they show obvious deficiencies in dealing with dynamic load fluctuations, equipment failures or environmental changes. At the same time, the existing methods also have problems of low computational efficiency and insufficient robustness in large-scale data processing and real-time scheduling capabilities, which are difficult to meet the actual application needs. Therefore, the existing technology still has a lot of room for improvement in multi-objective comprehensive optimization and dealing with dynamic changes. Summary of the invention

[0004] In order to overcome the defects of the existing distributed energy scheduling technology, such as lack of adaptive mechanism, low computational efficiency and low robustness, the present invention proposes the following technical solutions:

[0005] In a first aspect, the present invention proposes an adaptive multi-objective optimization method for a distributed energy dispatching system, comprising:

[0006] S1: Taking dispatching cost, energy utilization efficiency and system stability as optimization targets respectively, a comprehensive objective function of multi-objective optimization of distributed energy dispatching system is constructed;

[0007] S2: Obtain the operating data of the distributed energy dispatching system;

[0008] S3: Based on the operation data, the parameter weights of the comprehensive objective function are adjusted using a deep neural network, and the comprehensive objective function is updated using a genetic algorithm and a particle swarm optimization algorithm to generate candidate scheduling solutions;

[0009] S4: Iterate S2 to S3 until the maximum number of iterations is reached or the comprehensive objective function converges, and generate the final scheduling plan.

[0010] As a preferred technical solution, a comprehensive objective function for multi-objective optimization of distributed energy dispatching system is constructed by weighted summation, and its expression is as follows:

[0011]

[0012] Where n is the total number of optimization objectives; f i (x) represents the function value of the i-th optimization objective, which represents the evaluation value of the current scheduling scheme on the optimization objective i; ω i is the weight of the i-th optimization objective.

[0013] As a preferred technical solution, the parameter weights of the comprehensive objective function are adjusted according to the following formula:

[0014]

[0015] Among them, ω i (t+1) represents the weight value of the i-th optimization objective in the t+1-th iteration; is the initial weight value of the i-th optimization objective; Δω i (t) is the dynamic adjustment amount; α is the adjustment factor; f i (t) represents the function value of the i-th optimization objective in the current iteration, and represents the evaluation result of the current scheduling scheme on this objective.

[0016] As a preferred technical solution, the function value f of each optimization objective in the current iteration i (t) include:

[0017] Scheduling cost objective function value:

[0018]

[0019] Among them, L(t) is the system load at the current moment, L max is the maximum load allowed by the system, C(t) is the electricity price at the current moment;

[0020] Energy efficiency objective function value:

[0021]

[0022] Among them, R(t) is the renewable energy power generation at the previous moment, R max is the maximum power generation of renewable energy, S(t) is the available capacity of the energy storage system at the current moment, S max is the maximum capacity of the energy storage system;

[0023] System stability objective function value:

[0024]

[0025] Where G(t) is the stability value of the power grid at the current moment, including frequency offset and / or voltage fluctuation range; G max is the maximum stability value allowed by the power grid; Q(t) is the number of equipment failures in the system at the current moment; Q max is the maximum number of failures that the system can tolerate.

[0026] As a preferred technical solution, the parameter weights of the comprehensive objective function are dynamically adjusted according to the following adjustment rules:

[0027] When the system load value exceeds the preset threshold, the target weights of scheduling cost and system stability are increased;

[0028] When renewable energy generation exceeds a preset threshold, the target weights for reducing dispatch costs and system stability;

[0029] When the grid stability parameter is lower than the preset threshold or the number of equipment failures exceeds the preset threshold, the target weights of system stability and dispatch cost are increased.

[0030] As a preferred technical solution, genetic algorithm and particle swarm optimization algorithm are used to update the comprehensive objective function and generate candidate scheduling solutions, including:

[0031] Initialize the population and randomly generate multiple candidate scheduling schemes. Each scheme consists of scheduling parameters in the distributed energy system, including distributed generation power, energy storage equipment charging and discharging power, and load distribution ratio.

[0032] The speed and position of each candidate scheduling scheme are updated by the particle swarm optimization algorithm, where the speed update formula is:

[0033] v i (t+1)=ω·v i (t)+c1·rand1·(pbest i -x i (t))+c2·rand2(gbest-x i (t))

[0034] The position update formula is:

[0035] xi (t+1)=x i (t)+v i (t+1)

[0036] Among them, x i (t) represents the current parameter value of the i-th candidate scheduling scheme, including distributed generation power, energy storage device charging and discharging power, and load distribution ratio; v i (t) represents the adjustment range of the parameters of the i-th candidate scheduling scheme; ω is the inertia weight; c1 and c2 are acceleration factors, rand1 and rand2 are random numbers in the range of [0,1], and pbest i is the historical optimal parameter of the i-th scheduling scheme, and gbest is the global optimal parameter;

[0037] According to the fitness of the candidate scheduling schemes, the parent scheme is selected from the candidate scheduling schemes, and its expression is as follows:

[0038]

[0039] Among them, f i (x) is the fitness value of the i-th candidate scheduling solution, which is calculated according to the i-th sub-objective function;

[0040] Randomly select a crossover point k, and perform a crossover operation on the selected parent solution to generate a child solution according to the following formula:

[0041] offspring1=(P1[1:k],P2[k+1:n])

[0042] offspring2=(P2[1:k],P1[k+1:n])

[0043] Among them, P1 and P2 are parent solutions, and offspring1 and offspring2 are child solutions;

[0044] Set the mutation probability, mutate the scheduling parameters of each child plan according to the following formula, and update the child plan:

[0045]

[0046] As a preferred technical solution, before S3, the method further includes constructing a deep neural network, including:

[0047] Setting an input layer of the deep neural network, the input layer is used to receive real-time input features including load demand, grid status, renewable energy generation and energy storage status;

[0048] Setting multiple hidden layers, which capture the complex relationship between input features and optimization target weights through nonlinear activation functions;

[0049] Set the output layer, which is used to output the dynamic weight adjustment value of each optimization objective.

[0050] As a preferred technical solution, after constructing the deep neural network, the method further includes training the deep neural network, including:

[0051] Select historical data containing different loads, grid states, renewable energy generation, and energy storage state conditions as training samples;

[0052] Based on the training samples, the back propagation algorithm and gradient descent algorithm are used to optimize the parameters of the deep neural network, and the network weights and bias values ​​are adjusted until the weight adjustment value output by the deep neural network meets the ideal target value.

[0053] As a preferred technical solution, when the comprehensive objective function satisfies the following formula, it is judged that the comprehensive objective function converges:

[0054] |F best (t)-F best (t-1)|<∈

[0055] Among them, F best (t) is the optimal solution of the current iteration, and ∈ is the preset convergence threshold.

[0056] In a second aspect, the present invention further proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the operations performed by the adaptive multi-objective optimization method for a distributed energy dispatching system as described in any one of the schemes in the first aspect are implemented.

[0057] The beneficial effects of the present invention include at least:

[0058] The present invention takes dispatching cost, energy utilization efficiency and system stability as optimization objectives, constructs a comprehensive objective function, and through a mechanism of dynamically adjusting the objective weights, enables the dispatching scheme to respond in real time to dynamic environments such as changes in system load, fluctuations in renewable energy power generation, and changes in power grid status. At the same time, the advantages of genetic algorithms and particle swarm optimization algorithms are combined. The genetic algorithm provides global search capabilities, and the particle swarm optimization algorithm improves local convergence accuracy. The synergistic effect of the two effectively improves the optimization efficiency and accuracy, and enhances the robustness of the algorithm, thereby ensuring efficient and stable dispatching optimization in complex and changeable distributed energy systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1A schematic flow chart of an adaptive multi-objective optimization method for a distributed energy scheduling system provided in this embodiment.

[0060] Figure 2 A schematic diagram of the convergence of the adaptive multi-objective iterative update optimal solution provided by an embodiment of the present invention.

[0061] Figure 3 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred technical solutions. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred technical solutions are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0063] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0064] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0065] Example 1

[0066] This embodiment proposes an adaptive multi-objective optimization method for a distributed energy scheduling system. Figure 1 As shown, Figure 1 A schematic diagram of a flow chart of an adaptive multi-objective optimization method for a distributed energy scheduling system provided in this embodiment, the method comprising the following steps:

[0067] S1: Taking dispatching cost, energy utilization efficiency and system stability as optimization targets respectively, a comprehensive objective function of multi-objective optimization of distributed energy dispatching system is constructed;

[0068] S2: Obtain the operating data of the distributed energy dispatching system;

[0069] S3: Based on the operation data, the parameter weights of the comprehensive objective function are adjusted using a deep neural network, and the comprehensive objective function is updated using a genetic algorithm and a particle swarm optimization algorithm to generate candidate scheduling solutions;

[0070] S4: Iterate S2 to S3 until the maximum number of iterations is reached or the comprehensive objective function converges, and generate the final scheduling plan.

[0071] It can be understood that the present invention constructs a comprehensive objective function by taking scheduling cost, energy utilization efficiency and system stability as optimization objectives, and through a mechanism of dynamically adjusting target weights, enables the scheduling scheme to respond in real time to dynamic environments such as changes in system load, fluctuations in renewable energy power generation, and changes in power grid status; at the same time, combining the advantages of genetic algorithms and particle swarm optimization algorithms, genetic algorithms provide global search capabilities, and particle swarm optimization algorithms improve local convergence accuracy. The synergistic effect of the two effectively improves the optimization efficiency and accuracy, and enhances the robustness of the algorithm, thereby ensuring efficient and stable scheduling optimization in complex and changeable distributed energy systems.

[0072] Example 2

[0073] This embodiment makes improvements based on the adaptive multi-objective optimization method for distributed energy scheduling system proposed in Embodiment 1.

[0074] In this embodiment, a comprehensive objective function of multi-objective optimization of a distributed energy dispatching system is constructed by weighted summation, and its expression is as follows:

[0075]

[0076] Where n is the total number of optimization objectives; f i (x) represents the function value of the i-th optimization objective, which represents the evaluation value of the current scheduling scheme on the optimization objective i; ω i is the weight of the i-th optimization objective.

[0077] In this embodiment, the parameter weights of the comprehensive objective function are adjusted according to the following formula:

[0078]

[0079] Among them, ω i (t+1) represents the weight value of the i-th optimization objective in the t+1-th iteration; is the initial weight value of the i-th optimization objective; Δω i (t) is the dynamic adjustment amount; α is the adjustment factor; f i (t) represents the function value of the i-th optimization objective in the current iteration, and represents the evaluation result of the current scheduling scheme on this objective.

[0080] In this embodiment, the function value f of each optimization objective in the current iteration i (t) include:

[0081] Scheduling cost objective function value:

[0082]

[0083] Among them, L(t) is the system load at the current moment, L max is the maximum load allowed by the system, C(t) is the electricity price at the current moment;

[0084] Energy efficiency objective function value:

[0085]

[0086] Among them, R(t) is the renewable energy power generation at the previous moment, R max is the maximum power generation of renewable energy, S(t) is the available capacity of the energy storage system at the current moment, S max is the maximum capacity of the energy storage system;

[0087] System stability objective function value:

[0088]

[0089] Where G(t) is the stability value of the power grid at the current moment, including frequency offset and / or voltage fluctuation range; G max is the maximum stability value allowed by the power grid; Q(t) is the number of equipment failures in the system at the current moment; Q max is the maximum number of failures that the system can tolerate.

[0090] In this embodiment, the parameter weights of the comprehensive objective function are dynamically adjusted according to the following adjustment rules:

[0091] When the system load value exceeds the preset threshold, the target weights of scheduling cost and system stability are increased;

[0092] When renewable energy generation exceeds a preset threshold, the target weights for reducing dispatch costs and system stability;

[0093] When the grid stability parameter is lower than the preset threshold or the number of equipment failures exceeds the preset threshold, the target weights of system stability and dispatch cost are increased.

[0094] In this embodiment, the comprehensive objective function is updated using a genetic algorithm and a particle swarm optimization algorithm to generate candidate scheduling schemes, including:

[0095] Initialize the population and randomly generate multiple candidate scheduling schemes. Each scheme consists of scheduling parameters in the distributed energy system, including distributed generation power, energy storage equipment charging and discharging power, and load distribution ratio.

[0096] The speed and position of each candidate scheduling scheme are updated by the particle swarm optimization algorithm, where the speed update formula is:

[0097] v i (t+1)=ω·v i (t)+c1·rand1·(pbest i -x i (t))+c2·rand2(gbest-x i (t))

[0098] The position update formula is:

[0099] x i (t+1)=x i (t)+v i (t+1)

[0100] Among them, x i (t) represents the current parameter value of the i-th candidate scheduling scheme, including distributed generation power, energy storage device charging and discharging power, and load distribution ratio; v i (t) represents the adjustment range of the parameters of the i-th candidate scheduling scheme; ω is the inertia weight; c1 and c2 are acceleration factors, rand1 and rand2 are random numbers in the range of [0,1], and pbest i is the historical optimal parameter of the i-th scheduling scheme, and gbest is the global optimal parameter;

[0101] According to the fitness of the candidate scheduling schemes, the parent scheme is selected from the candidate scheduling schemes, and its expression is as follows:

[0102]

[0103] Among them, i (x) is the fitness value of the i-th candidate scheduling solution, which is calculated according to the i-th sub-objective function;

[0104] Randomly select a crossover point k, and perform a crossover operation on the selected parent solution to generate a child solution according to the following formula:

[0105] offspring1=(P1[1:k],P2[k+1:n])

[0106] offspring2=(P2[1:k],P1[k+1:n])

[0107] Among them, P1 and P2 are parent solutions, and offspring1 and offspring2 are child solutions;

[0108] Set the mutation probability, mutate the scheduling parameters of each child plan according to the following formula, and update the child plan:

[0109]

[0110] In this embodiment, before S3, the method further includes constructing a deep neural network, including:

[0111] Setting an input layer of the deep neural network, the input layer is used to receive real-time input features including load demand, grid status, renewable energy generation and energy storage status;

[0112] Setting multiple hidden layers, which capture the complex relationship between input features and optimization target weights through nonlinear activation functions;

[0113] Set up an output layer, which is used to output the dynamic weight adjustment value of each optimization objective.

[0114] In this embodiment, after constructing the deep neural network, the method further includes training the deep neural network, including:

[0115] Select historical data containing different loads, grid states, renewable energy generation, and energy storage state conditions as training samples;

[0116] Based on the training samples, the back propagation algorithm and gradient descent algorithm are used to optimize the parameters of the deep neural network, and the network weights and bias values ​​are adjusted until the weight adjustment value output by the deep neural network meets the ideal target value.

[0117] As an exemplary explanation, first, a training data set is selected to extract the real-time status information of the system, such as load demand, grid status, renewable energy generation and energy storage status, as the input features of the neural network, and the adjustment value of the optimization target weight (such as operating cost, energy efficiency, system stability, etc.) is used as the output label of the neural network. Secondly, the training data is preprocessed, including standardization of the input features to ensure that the data is within the same scale range, and historical data of the system under different load and grid conditions are collected to cover various typical operating conditions to ensure the comprehensiveness and representativeness of the training data. Then, a neural network model is designed, in which the input layer is used to receive real-time status information such as load demand, grid stability, and renewable energy generation. The hidden layer captures the complex relationship between the system state and the target weight through a nonlinear activation function (such as ReLU), and the output layer is used to output the dynamic weight adjustment value of each optimization target. Next, the training process uses backpropagation and gradient descent to train the model, optimizes the network parameters by minimizing the loss function, and uses the mean square error (MSE) to measure the gap between the predicted output and the target value; the training set is separated from the validation set, the training set is used to optimize the model parameters, and the validation set is used to evaluate the model performance to ensure its generalization ability and robustness; at the same time, the K-fold cross-validation method is used to verify the model to further ensure the stability and reliability of the training.

[0118] During the training process, the validation set is used to evaluate the performance of the model. If the model has a large error on the validation set, the network architecture needs to be adjusted, including increasing the number of layers, changing the activation function, or performing regularization to prevent overfitting. In the hyperparameter adjustment stage, hyperparameters such as learning rate and batch size are adjusted through network search or random search methods to obtain the best training effect. After the neural network training is completed, the model will receive the real-time status information of the system and perform forward propagation, dynamically adjust the weight of the optimization target, and the adjusted weight will guide the iterative process of the particle swarm optimization algorithm (PSO) and the genetic algorithm (GA), further optimize the value of the comprehensive objective function and generate the final scheduling plan.

[0119] In this embodiment, when the comprehensive objective function satisfies the following formula, it is judged that the comprehensive objective function converges:

[0120] |F best (t)-F best (t-1)|<∈

[0121] Among them, F best (t) is the optimal solution of the current iteration, and ∈ is the preset convergence threshold.

[0122] like Figure 2 As shown, Figure 2A schematic diagram of the convergence of the adaptive multi-objective iterative update optimal solution provided by an embodiment of the present invention, Figure 2 It shows the changing trend of the objective function value during the convergence process. The horizontal axis represents the number of iterations, and the vertical axis represents the fitness value of the objective function. Figure 2 It can be seen that the objective function value drops rapidly in the initial stage (0 to 500 iterations), indicating that the optimization algorithm can quickly search for a solution space close to the global optimal solution in the early stage. As the number of iterations increases (500 to 1000 iterations), the fitness value decreases gradually, indicating that the algorithm mainly performs local optimization at this stage and further approaches the optimal solution.

[0123] After 1000 iterations, the fitness value is basically stable within a certain range, indicating that the optimization process has converged and found a set of optimal solutions that meet the multi-objective balance. The entire optimization process demonstrates the algorithm's good global search ability and local convergence ability, verifies the effectiveness of the combination of the adaptive weight adjustment mechanism and the optimization algorithm, and can achieve fast and efficient optimal solution search in multi-objective optimization problems.

[0124] Example 3

[0125] Based on the adaptive multi-objective optimization method of the distributed energy dispatching system proposed in Example 2, this embodiment designs three typical experimental scenarios to simulate different environmental conditions and demand scenarios, which respectively reflect the changes in load demand, renewable energy generation, energy storage system status, and grid stability. In these scenarios, the system performance before and after optimization can be compared and analyzed, including indicators such as operating cost, energy efficiency, and system stability.

[0126] Scenario 1 is a high-load, low renewable energy generation situation. The system load is 120MW, which exceeds the normal load of the system. At this time, photovoltaic power generation is 30MW, wind power generation is 20MW, and the total renewable energy generation is only 50MW, which is relatively low. The remaining power of the energy storage system is 40MWh, but it has not been fully charged. The grid is relatively stable and there are no faults.

[0127] Scenario 2 is a low-load, high-renewable energy generation scenario, with a system load of 60MW, which is relatively low. Photovoltaic power generation is 80MW, wind power generation is 40MW, and the total renewable energy generation reaches 120MW, which is relatively high. The remaining power of the energy storage system is 100MWh, the system is basically fully charged, and the grid is stable without any failure.

[0128] Scenario 3 is a situation of equipment failure and load fluctuation. The system load is 90MW, and the load fluctuation is large. Photovoltaic power generation is 50MW, wind power generation is 30MW, and the total renewable energy power generation is 80MW. The remaining power of the energy storage system is 60MWh, some equipment failures have occurred in the power grid, and some power grids are unstable.

[0129] Through the experimental data in these different scenarios, the optimization process and results can be analyzed, and the applicability of the optimization method in complex environments can be verified.

[0130] Table 1 Comparison of high load and low renewable energy generation before and after optimization

[0131]

[0132]

[0133] As shown in Table 1, in the scenario of high load and low renewable energy generation, the various indicators of the system have been significantly improved through the adaptive multi-objective optimization algorithm. Specifically, the operating cost index was 50% before optimization and was reduced to 65% after optimization, reflecting that under high load conditions, the optimization algorithm can effectively reduce the resource cost consumed during system operation and achieve better resource allocation through reasonable scheduling strategies. The energy utilization efficiency increased from 55% to 70%. Although the renewable energy generation was low, the overall energy utilization efficiency of the system was significantly enhanced through the optimized utilization of energy storage equipment and the refined adjustment of load distribution. The system stability increased from 75% to 85%. Under high load conditions, the optimization algorithm gave priority to the stability of the power grid operation. By dynamically adjusting the target weight, the system stability was effectively improved, ensuring the safe operation of the system.

[0134] Overall, the algorithm shows strong adaptability in scenarios with heavy loads. By dynamically adjusting the weights of operating costs and stability targets, it prioritizes the system's operating requirements while taking into account energy efficiency. The results show that the optimized system can achieve a multi-objective balance of operating costs, energy efficiency, and system stability under complex conditions, providing a reliable method support for the dispatch optimization of distributed energy systems.

[0135] Table 2 Comparison of low load and high renewable energy generation before and after optimization

[0136]

[0137]

[0138] As shown in Table 2, in the scenario of low load and high renewable energy generation, the system has been optimized in terms of operating cost, energy utilization efficiency and system stability through the adaptive multi-objective optimization algorithm. The operating cost index was 60% before optimization and increased to 75% after optimization, indicating that under low load conditions, the algorithm prioritizes the use of renewable energy through reasonable scheduling and reduces dependence on traditional energy, thereby reducing the overall operating cost. Energy utilization efficiency increased from 80% to 90%. By making full use of renewable energy generation and coordinated optimization of energy storage equipment, the system has achieved a higher level of resource utilization while reducing energy loss. System stability increased from 80% to 85%. Although the grid operation conditions are relatively stable, the optimization algorithm further enhances the robustness of the system by dynamically adjusting the target weights in real time, ensuring operational safety under high renewable energy penetration.

[0139] Overall, in this scenario, the optimization algorithm effectively utilizes the high power generation of renewable energy, and by dynamically adjusting the target weight, it takes into account the operating cost, energy efficiency and system stability, showing a good optimization effect. The results show that the optimized system maximizes resource utilization efficiency while maintaining low operating costs and stable operating status, providing an efficient and sustainable solution for distributed energy scheduling.

[0140] Table 3 Comparison of equipment failure and load fluctuation before and after optimization

[0141] index Before optimization After optimization Running costs 45％ 60％ Energy efficiency 55％ 65％ System stability 70％ 80％

[0142] In the scenario of equipment failure and load fluctuation, the system's operating performance has been significantly improved in many aspects through the adaptive multi-objective optimization algorithm. Before optimization, the operating cost index was 45%, and after optimization it increased to 60%. This shows that the algorithm can effectively reduce the additional costs caused by unstable operation by dynamically adjusting the scheduling strategy and resource allocation under the conditions of system failure and load fluctuation. The energy utilization efficiency increased from 55% to 65%. In the case of system fluctuation, the optimization algorithm reduces energy waste and improves the utilization of system resources by coordinating the use of renewable energy generation and energy storage equipment.

[0143] The system stability has increased from 70% to 80%, indicating that the optimization algorithm has significantly enhanced its dispatch decision-making ability under faults and load fluctuations. By real-time monitoring of equipment status and load changes, the optimization process prioritizes the system's stability goals, dynamically adjusts target weights, and rationally allocates power resources, significantly reducing the uncertainty and fault risks of system operation.

[0144] In summary, in the complex scenario of equipment failure and load fluctuation, the optimization algorithm demonstrates strong adaptability and scheduling efficiency, and can simultaneously improve the system's stability, energy utilization efficiency, and operating cost management capabilities, providing reliable technical support for distributed energy scheduling under complex working conditions, and ensuring the continuity and efficiency of system operation.

[0145] Example 4

[0146] Figure 3 The electronic device 100 provided in this embodiment is a schematic diagram of the structure of the electronic device 100. The electronic device 100 includes: a memory 101, a processor 102, and a computer program stored in the memory 101 and executable on the processor 102.

[0147] When the processor 102 executes the program, the image classification method based on multi-instance transfer learning provided in the above embodiment is implemented.

[0148] Furthermore, the electronic device 100 also includes: a communication interface 103 for communication between the memory 101 and the processor 102 .

[0149] The memory 101 may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0150] If the memory 101, the processor 102 and the communication interface 103 are implemented independently, the communication interface 103, the memory 101 and the processor 102 can be connected to each other through a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0151] Optionally, in a specific implementation, if the memory 101, the processor 102 and the communication interface 103 are integrated on a chip, the memory 101, the processor 102 and the communication interface 103 can communicate with each other through an internal interface.

[0152] The processor 102 may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention.

[0153] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0154] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0155] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.

[0156] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array, a field programmable gate array, etc.

[0157] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0158] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. An adaptive multi-objective optimization method for a distributed energy dispatching system, characterized in that: include: S1: Taking dispatching cost, energy utilization efficiency and system stability as optimization targets respectively, a comprehensive objective function of multi-objective optimization of distributed energy dispatching system is constructed; S2: Obtain the operating data of the distributed energy dispatching system; S3: Based on the operation data, the parameter weights of the comprehensive objective function are adjusted using a deep neural network, and the comprehensive objective function is updated using a genetic algorithm and a particle swarm optimization algorithm to generate candidate scheduling solutions; S4: Iterate S2 to S3 until the maximum number of iterations is reached or the comprehensive objective function converges, and generate the final scheduling plan.

2. The adaptive multi-objective optimization method for distributed energy scheduling system according to claim 1 is characterized in that: Through weighted summation, the comprehensive objective function of multi-objective optimization of distributed energy scheduling system is constructed, and its expression is as follows: Where n is the total number of optimization objectives; f i (x) represents the function value of the i-th optimization objective, which represents the evaluation value of the current scheduling scheme on the optimization objective i; ω i is the weight of the i-th optimization objective.

3. The adaptive multi-objective optimization method for distributed energy scheduling system according to claim 2 is characterized in that: According to the following formula, adjust the parameter weights of the comprehensive objective function: Among them, ω i (t+1) represents the weight value of the i-th optimization objective in the t+1-th iteration; is the initial weight value of the i-th optimization target; Δω i (t) is the dynamic adjustment amount; α is the adjustment factor; f i (t) represents the function value of the i-th optimization objective in the current iteration, and represents the evaluation result of the current scheduling scheme on this objective.

4. The adaptive multi-objective optimization method for distributed energy scheduling system according to claim 3 is characterized in that: The function value f of each optimization objective in the current iteration i (t) include: Scheduling cost objective function value: Among them, L(t) is the system load at the current moment, L max is the maximum load allowed by the system, C(t) is the electricity price at the current moment; Energy efficiency objective function value: Among them, R(t) is the renewable energy power generation at the previous moment, R max is the maximum power generation of renewable energy, S(t) is the available capacity of the energy storage system at the current moment, S max is the maximum capacity of the energy storage system; System stability objective function value: Where G(t) is the stability value of the power grid at the current moment, including frequency offset and / or voltage fluctuation range; G max is the maximum stability value allowed by the power grid; Q(t) is the number of equipment failures in the system at the current moment; Q max is the maximum number of failures that the system can tolerate.

5. The adaptive multi-objective optimization method for distributed energy scheduling system according to claim 4 is characterized in that: The parameter weights of the comprehensive objective function are dynamically adjusted according to the following adjustment rules: When the system load value exceeds the preset threshold, the target weights of scheduling cost and system stability are increased; When renewable energy generation exceeds a preset threshold, the target weights for reducing dispatch costs and system stability; When the grid stability parameter is lower than the preset threshold or the number of equipment failures exceeds the preset threshold, the target weights of system stability and dispatch cost are increased.

6. The adaptive multi-objective optimization method for distributed energy scheduling system according to claim 1, characterized in that: Genetic algorithm and particle swarm optimization algorithm are used to update the comprehensive objective function and generate candidate scheduling solutions, including: Initialize the population and randomly generate multiple candidate scheduling schemes. Each scheme consists of scheduling parameters in the distributed energy system, including distributed generation power, energy storage equipment charging and discharging power, and load distribution ratio. The speed and position of each candidate scheduling scheme are updated by the particle swarm optimization algorithm, where the speed update formula is: v i (t+1)=ω·v i (t)+c1·rand1·(pbest i -x i (t))+c2·rand2(gbest-x i (t)) The position update formula is: x i (t+1)=x i (t)+v i (t+1) Among them, x i (t) represents the current parameter value of the i-th candidate scheduling scheme, including distributed generation power, energy storage device charging and discharging power, and load distribution ratio; v i (t) represents the adjustment range of the parameters of the i-th candidate scheduling scheme; ω is the inertia weight; c1 and c2 are acceleration factors, rand1 and rand2 are random numbers in the range of [0,1], and pbest i is the historical optimal parameter of the i-th scheduling scheme, and gbest is the global optimal parameter; According to the fitness of the candidate scheduling schemes, the parent scheme is selected from the candidate scheduling schemes, and its expression is as follows: Among them, f i (x) is the fitness value of the i-th candidate scheduling solution, which is calculated according to the i-th sub-objective function; Randomly select a crossover point k, and perform a crossover operation on the selected parent solution to generate a child solution according to the following formula: offspring1=(P1[1:k],P2[k+1:n]) offspring2=(P2[1:k],P1[k+1:n]) Among them, P1 and P2 are parent solutions, and offspring1 and offspring2 are child solutions; Set the mutation probability, mutate the scheduling parameters of each child plan according to the following formula, and update the child plan:

7. The adaptive multi-objective optimization method for distributed energy scheduling system according to claim 1, characterized in that: Prior to S3, methods also included building deep neural networks, including: Setting an input layer of the deep neural network, the input layer is used to receive real-time input features including load demand, grid status, renewable energy generation and energy storage status; Setting multiple hidden layers, which capture the complex relationship between input features and optimization target weights through nonlinear activation functions; Set the output layer, which is used to output the dynamic weight adjustment value of each optimization objective.

8. The adaptive multi-objective optimization method for distributed energy dispatching system according to claim 7, characterized in that: After building the deep neural network, the method further includes training the deep neural network, including: Select historical data containing different loads, grid states, renewable energy generation, and energy storage state conditions as training samples; Based on the training samples, the back propagation algorithm and gradient descent algorithm are used to optimize the parameters of the deep neural network, and the network weights and bias values ​​are adjusted until the weight adjustment value output by the deep neural network meets the ideal target value.

9. The adaptive multi-objective optimization method for distributed energy dispatching system according to any one of claims 1 to 8, characterized in that: When the comprehensive objective function satisfies the following formula, it is judged that the comprehensive objective function converges: |F best (t)-F best (t-1)|<∈ Among them, F best (t) is the optimal solution of the current iteration, and ∈ is the preset convergence threshold.

10. An electronic device, characterized in that: The control device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the operations performed by the adaptive multi-objective optimization method for a distributed energy dispatching system as described in any one of claims 1 to 9 are implemented.

Images