A distributed power supply system and method based on smart energy management platform

By adopting a smart energy management platform in a distributed power supply system, combining neural networks and genetic algorithms, scheduling the output power of the generator set, the problem of difficult to take into account both the stability and cost factors of the power grid in the existing technology is achieved, and the stability of the system is achieved Improved sex and flexibility.

CN117767430BActive Publication Date: 2024-11-26STATE GRID JIBEI ELECTRIC POWER COMPANY LIMITED CHENGDE POWER SUPPLY
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202311724348.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-15
Publication Date
2024-11-26
Estimated Expiration
2043-12-15

AI Technical Summary

Technical Problem

The prior art is difficult to consider both grid stability and cost factors in distributed power supply systems, and the scheduling algorithm has limited accuracy and lacks flexibility, making it difficult to meet the needs of complex power grids.

Method used

The system based on the smart energy management platform is adopted, including power prediction module, weight allocation module, discrimination module and scheduling module. The output power of the generator set is scheduled through neural network algorithms and genetic algorithms, and the loss weight and power fluctuation loss weight are set to ensure the system Stability and flexibility.

Benefits of technology

Reasonable scheduling of the output power of different generator sets is achieved, the stability and scheduling flexibility of complex distributed power supply systems are ensured, and the calculation accuracy and system stability are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117767430B_ABST
    Figure CN117767430B_ABST
Patent Text Reader

Abstract

The present application proposes a distributed power supply system and method based on a smart energy management platform, the core solution of which includes at least two generator sets and an energy management platform, the energy management platform including: a global power prediction module for calculating the expected power usage in the first time period of the power supply area based on a linear regression equation; a unit power prediction module for calculating the maximum unit output power of the generator set in the first time period based on a first prediction model; a weight allocation module for setting the loss weight of the generator set; a discrimination module for selecting a first genetic algorithm as a global scheduling algorithm; a scheduling module for calculating the expected power of the generator set based on the first genetic algorithm. The distributed power supply system and method of the present application can reasonably schedule the output power of different generator sets, ensure the stability of a complex distributed power supply system, and improve the flexibility of system scheduling.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to a distributed power supply system, and in particular to a distributed power supply system and method based on a smart energy management platform. Background Art

[0002] With the advancement of renewable energy technology and the need for grid modernization, distributed power supply systems have become a key development trend in the power industry. Distributed power supply systems often include a variety of generators, such as wind power, thermal power, hydropower, and solar power. In a complex power supply system, an effective power distribution strategy is essential to ensure the stability of the power grid. At the same time, considering the cost of electricity, how to dispatch the output power of generators based on stability and cost factors has become an urgent problem to be solved.

[0003] In addition, since distributed power generation systems are relatively complex, how to improve the flexibility of system scheduling in such a complex system is also an issue worthy of attention.

[0004] Although existing technologies such as CN111668882A and CN114329317A have proposed scheduling solutions for hybrid generator sets, they have not yet comprehensively considered the stability and cost factors of the power grid. In addition, the scheduling algorithms provided have limited accuracy and lack flexibility, making it difficult to meet the needs of complex power grids. Summary of the invention

[0005] This application proposes a distributed power supply system and method based on a smart energy management platform, which can reasonably schedule the output power of different generator sets, ensure the stability of complex distributed power supply systems, and improve the flexibility of system scheduling.

[0006] On the one hand, the present application proposes a distributed power supply system based on a smart energy management platform, including at least two generator sets and an energy management platform, wherein the at least two generator sets include a first generator set and a second generator set, and the generator types of the at least two generator sets are different, and the energy management platform includes:

[0007] Power prediction module, including global power prediction module and unit power prediction module;

[0008] The above-mentioned global power prediction module is used to calculate the expected power usage in the first time period of the power supply area based on the linear regression equation ;

[0009] The above-mentioned unit power prediction module is used to calculate the maximum unit output power of the first generator set in the above-mentioned first time period based on the first prediction model , the maximum unit output power of the second generator set in the first time period ; The algorithm based on the above-mentioned first prediction model is a neural network algorithm;

[0010] A weight allocation module, used to set the loss weight w1 of the first generator set and the loss weight w2 of the second generator set;

[0011] A discrimination module, configured to select the first genetic algorithm as the global scheduling algorithm when it is identified that the at least two generator sets do not include a third generator set; the generator type of the third generator set is different from that of the first generator set and the second generator set;

[0012] A scheduling module is used to calculate the expected power of the first generator set in the first time period so that the fitness function based on the first genetic algorithm is at a minimum value according to the first genetic algorithm. and the expected power of the second generator set in the first time period ;

[0013] The fitness function based on the first genetic algorithm mentioned above is , the above , are the loss functions of the first generator set and the second generator set respectively, , is the output power of the first generator set and the second generator set; The power fluctuation loss weight is used to adjust the loss caused by power fluctuation to the system;

[0014] The above fitness function The constraints are: , .

[0015] The distributed power supply system has the following advantages: The discrimination module determines genetic algorithms of different complexity based on the properties of the generator set, which can avoid poor calculation results or host system crashes due to overly complex parameters, ensuring the stability of scheduling. Different loss weights are set for different generator sets, and the output power of different generator sets can be adjusted according to the requirements of decision makers by setting weights, ensuring the flexibility of scheduling. The fitness function includes the loss function of the generator set and the power fluctuation loss , the loss function represents the loss generated by the generator set during the power generation process, and the power fluctuation loss is used to evaluate the loss caused by the imbalance of power supply and demand. The fitness function can fully consider the power generation loss and the loss caused by the imbalance of power supply and demand in the system scheduling process, ensuring the stability and flexibility of scheduling. The maximum unit output power represents the maximum power generation power of the generator set. By predicting the maximum unit output power, the expected power can be limited to the maximum power generation power, thereby improving the accuracy and stability of scheduling. The global prediction module uses a linear regression equation for calculation because the power demand changes in the power supply area are easier to calculate from historical data. Using a simpler linear regression equation can save system computing power. The above scheme fully realizes the reasonable scheduling of the output power of different generator sets, ensures the stability of the complex distributed power supply system, and improves the flexibility of system scheduling.

[0016] In one embodiment, the generator type of the first generator set is a thermal generator, the generator type of the second generator set is a wind generator, the generator type of the third generator set is an energy storage generator or a hydroelectric generator, and the loss function of the first generator set is , the loss function of the second generator set is , the above a1, a2, a3, b1, b2, b3 are the loss constants of the generator.

[0017] The chromosome matrix used by the first genetic algorithm is =[ , ], among which, the above They are , Vectorized representation of ;

[0018] The crossover operator based on the first genetic algorithm mentioned above is - ; where t is the number of iterations, is the number of iterations in which The chromosome matrix with the smallest value, where c1 and c2 are random vectors.

[0019] In a possible implementation, the mutation operator based on the first genetic algorithm is ,in, is a random vector, , .

[0020] This application uses improved crossover operators and mutation operators to calculate the scheduling plan, which can greatly improve the calculation accuracy and ensure the stability of system scheduling.

[0021] In one embodiment, the first prediction model includes a first neural network algorithm and a second neural network algorithm, and the unit power prediction module is specifically used for:

[0022] The historical data belonging to the same day are input into the first neural network to obtain the first predicted maximum output power. , , input the historical data of the first time period belonging to different days into the second neural network to obtain the second predicted maximum output power , ;

[0023] The first predicted maximum output power and the second predicted maximum output power are weighted to obtain the above , .

[0024] This application adopts a combination of horizontal calculation and vertical calculation. The maximum output power of the generator set is calculated through historical data of the same day and historical data of the same time period on different days, which can greatly improve the calculation accuracy and ensure the stability of system scheduling.

[0025] The loss weight w1 of the first generator set is greater than the loss weight w2 of the second generator set.

[0026] Because the first generator set is a thermal generator set and the second generator set is a wind generator set, considering environmental factors, the loss weight w1 of the first generator set can be set greater than the loss weight w2 of the second generator set, and more power generation power can be allocated to the wind generator set.

[0027] In one embodiment, when it is identified that the at least two generator sets include a third generator set, a second genetic algorithm is selected as the global scheduling algorithm; wherein the complexity of the second genetic algorithm is less than the complexity of the first genetic algorithm, and the complexity is used to represent the time consumed by the operation.

[0028] When it is identified that the system has multiple generator sets, a less complex global scheduling algorithm is promptly adopted to avoid system crashes or long scheduling times.

[0029] In one embodiment, the weight allocation module includes: a monitoring module for monitoring weight setting instructions; a verification module for reading instruction verification information when obtaining the weight setting instruction, and verifying the weight setting instruction based on the instruction verification information; and a parsing module for parsing the weight setting instruction when the weight setting instruction is verified, to obtain the loss weight w1 of the first generator group and the loss weight w2 of the second generator group.

[0030] By setting up a monitoring module, this application can obtain weight setting instructions in a timely manner and adjust the scheduling strategy in a timely manner. Because weight allocation is the highest authority decision-making project, if it is improperly invaded, it will seriously undermine the stability of the scheduling system. For this reason, this application specifically verifies the weight setting instructions to ensure the security of scheduling.

[0031] The distributed power supply system provided in the present application can reasonably dispatch the output power of different generator sets, ensure the stability of the complex distributed power supply system, and improve the flexibility of system scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the drawings required for use in the embodiments of the present application or the background technology will be described below.

[0033] Figure 1 This is a distributed power supply system based on a smart energy management platform proposed in this application;

[0034] Figure 2 This application proposes a distributed power supply method based on a smart energy management platform;

[0035] Figure 3 This application proposes a distributed power supply device based on a smart energy management platform. DETAILED DESCRIPTION

[0036] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods or devices.

[0037] Figure 1 This application proposes a distributed power supply system based on a smart energy management platform.

[0038] The above-mentioned distributed power supply system includes at least two generator sets 100 and an energy management platform; the above-mentioned at least two generator sets include a first generator set 110 and a second generator set 120, and the generator types of the above-mentioned at least two generator sets are different; the above-mentioned energy management platform includes a power prediction module 210, a weight allocation module 220, a judgment module 230, and a scheduling module 240, and the above-mentioned power prediction module includes a global power prediction module 211 and a unit power prediction module 212.

[0039] The generator types of at least two generator sets may include thermal generators, wind generators, solar generators, hydroelectric generators, energy storage generators, etc., and the generator types of at least two generator sets are different.

[0040] The global power prediction module 211 is used to calculate the expected power usage in the first time period of the power supply area based on the linear regression equation. The linear regression equation may be a univariate linear regression model; the input parameter of the linear regression equation is the power usage of the power supply area in the same time period on different dates. The power supply area is the target power supply area of ​​the distributed power supply system.

[0041] In a distributed power supply system, a certain power supply area is often powered by multiple power supply units of different categories, and the power supply area is relatively small. The aforementioned power supply area is the target power supply area of ​​the distributed power supply system. The first time period is the time period to be predicted, which is often the peak power consumption period. During this time period, the output power of the generator set usually needs to be enhanced.

[0042] The unit power prediction module 212 is used to calculate the maximum unit output power of the first generator set in the first time period based on the first prediction model. , the maximum unit output power of the second generator set in the first time period The algorithm based on the first prediction model is a neural network algorithm. The maximum unit output power is related to the power generation factor. For a wind turbine, the maximum unit output power is related to the wind force in the first time period; for a thermal power generator, the maximum unit output power is related to the fuel input in the first time period; for a hydropower generator, the maximum unit output power is related to the water flow in the first time period.

[0043] Since the new energy power generation elements are difficult to predict, a neural network algorithm is specially set up to predict the maximum unit output power to improve accuracy. Furthermore, in order to further improve the accuracy of the prediction, the unit output power of the generator can be calculated by combining horizontal calculation and vertical calculation.

[0044] Specifically, the first prediction model includes a first neural network algorithm and a second neural network algorithm. The horizontal calculation refers to: inputting historical data of different time periods belonging to the same day into the first neural network to obtain the first predicted maximum output power , The longitudinal calculation refers to: inputting the historical data of the first time period belonging to different days into the second neural network to obtain the second predicted maximum output power , Finally, the first predicted maximum output power and the second predicted maximum output power are weighted to obtain the above , The weight of the weighted operation may be allocated based on the accuracy of the horizontal calculation result and the vertical calculation result determined by the test result. For example, if the maximum output power calculated by the horizontal calculation result is determined to be more consistent with the actual situation through testing, the horizontal calculation result will be allocated a higher weight.

[0045] It should be noted that the purpose of calculating the maximum unit output power is to determine the constraint conditions of the fitness function, that is, the power variable of the fitness function cannot exceed the maximum unit output power.

[0046] The weight allocation module 220 is used to set the loss weight w1 of the first generator set and the loss weight w2 of the second generator set. The above loss weights are used to construct the fitness function of the genetic algorithm, wherein, if the loss weight is increased, the output power that the generator set can ultimately allocate will decrease. In actual application scenarios, if the first generator set is a thermal generator and the second generator set is a wind generator, based on environmental considerations, w1>w2 can be set so that the output power of the thermal generator is lower and the output power of the wind generator is higher.

[0047] The discrimination module 230 is used to select the first genetic algorithm as the global scheduling algorithm when it is identified that the at least two generator sets do not include the third generator set; the generator type of the third generator set is different from the first generator set and the second generator set. Among them, the generator type of the first generator set is a thermal generator, the generator type of the second generator set is a wind generator, and the generator type of the third generator set is a storage generator or a hydroelectric generator. Because the first genetic algorithm is a scheduling algorithm with higher complexity, its calculation accuracy is relatively high, but it is not suitable for systems with too complex parameters. When the number of generator sets is large, especially including hydroelectric generators with more complex parameters, a genetic algorithm with lower complexity should be adjusted. Therefore, the discrimination module 230 is also used to: when it is identified that the at least two generator sets include the third generator set, select the second genetic algorithm as the global scheduling algorithm; wherein the complexity of the second genetic algorithm is less than the complexity of the first genetic algorithm, and the complexity is used to represent the time consumed by the operation.

[0048] The scheduling module 240 is used to calculate the expected power of the first generator set in the first time period according to the first genetic algorithm so that the fitness function based on the first genetic algorithm is at the minimum value. and the expected power of the second generator set in the first time period The aforementioned expected power is the planned output power of the generator set in the first time period.

[0049] The fitness function based on the first genetic algorithm mentioned above is , the above , are the loss functions of the first generator set and the second generator set respectively, , is the output power of the first generator set and the second generator set; It is the power fluctuation loss weight, which is used to adjust the weight of the loss caused by power fluctuation to the system.

[0050] As mentioned above, the fitness function includes two parts: the loss of the generator set and the power fluctuation loss. The loss of the generator set includes the sum of the losses of the generators of each unit. The absolute value of is the power fluctuation loss weight. The larger w0 is, the greater the impact of power fluctuation factors on the calculation results. Conversely, the smaller the impact of the generator set loss is, the calculation results will make the power fluctuation factor The absolute value of is smaller.

[0051] Above , are the loss functions of the first and second generator sets respectively, among which the thermal generator set , the loss function of the above wind turbine generator set is , a1, a2, a3, b1, b2, b3 are the loss constants of the generator. The loss constant is related to the properties of the unit equipment itself and often needs to be obtained by testing the unit equipment.

[0052] If the system also includes a third generator set (such as an energy storage generator set or a hydroelectric generator set), the fitness function can also include the loss function of the third generator set. Among them, the loss function of the energy storage generator can be defined as a linear function, or it can be defined as a function that is positively correlated with the source of the energy storage generator's electricity (such as hydropower, thermal power, and wind power). The loss function of a hydroelectric generator is usually related to factors such as equipment investment cost, operation and maintenance cost, water source cost, environmental and regulatory cost, and is more complex than other generators. One of the expressions is: = , where c1, c2, and c3 are loss constants. To make the calculation more accurate, the loss function often needs to be measured through field investigation, comprehensive consideration of equipment and environmental conditions, and the use of complex exponential, logarithmic and other expressions.

[0053] This application uses a genetic algorithm as a global scheduling algorithm, in which the basic steps of the genetic algorithm include: population initialization, fitness evaluation, chromosome selection, chromosome crossover, chromosome mutation, termination condition check and other basic steps. There are already a large number of expressions related to each basic step in the art, which are not listed here one by one. When initializing the chromosome, the variables need to be encoded into a chromosome matrix form, for example, =[ ], among which, the above They are , A vectorized or matrix representation of .

[0054] In order to improve the calculation accuracy, this application specifically improves the crossover operator in the chromosome crossover step and the mutation operator in the chromosome mutation step. - . Among them, the above t is the number of iterations, the above is the number of iterations in which The above c1 and c2 are random vectors. In one embodiment, each element can be randomly selected between 0 and 1, and the length and width are The crossover operator incorporates the global optimal value into the crossover calculation formula, which can greatly improve the calculation accuracy.

[0055] In one embodiment, the mutation operator is ,in, is a random vector, , The mutation operator combines the results of the first two operations to further improve the calculation accuracy.

[0056] The above is only a preferred calculation method. In fact, other formulas in the prior art can be selected, but the calculation accuracy may be different. For example, the chromosome selection step can use conventional formulas such as roulette and tournament selection, and the method for encoding the chromosome can be a conventional binary encoding method.

[0057] In one possible implementation, the weight allocation module 220 includes: a monitoring module 221 for monitoring weight setting instructions; a verification module 222 for reading instruction verification information when obtaining the weight setting instruction, and verifying the weight setting instruction based on the instruction verification information; and a parsing module 223 for parsing the weight setting instruction when the weight setting instruction is verified, to obtain the loss weight w1 of the first generator group and the loss weight w2 of the second generator group.

[0058] In fact, the loss weight of the generator and the power fluctuation loss weight can largely determine the power distribution of the generator set and often represent the will of the decision maker. Therefore, this application specifically limits the monitoring and verification of the weight setting instructions to fully ensure the real-time and stability of the power distribution of the power supply system.

[0059] In a possible implementation, the scheduling module 240 is further used to: send a scheduling instruction to the at least two generator sets, wherein the scheduling instruction is used to set the output power of the first generator set to , set the output power of the second generator set to After the expected power of each generator set is calculated, the expected power needs to be sent to each generator set in a timely manner so that each generator set sets the output power of the generator set to the aforementioned expected power within the first time period.

[0060] The present application has the following beneficial effects: by determining genetic algorithms of different complexities based on the properties of the generator set, it is possible to avoid poor calculation results or host system crashes due to overly complex parameters, thereby ensuring the stability of scheduling. Different loss weights are set for different generator sets, and the output power of different generator sets can be adjusted according to the requirements of decision makers by setting weights to ensure the flexibility of scheduling. The fitness function includes the loss function and power fluctuation loss of the generator set, which can fully consider the power generation loss and the loss caused by the imbalance between power supply and demand during the system scheduling process, ensuring the stability and flexibility of scheduling. By predicting the maximum unit output power, the expected power can be limited to the maximum power generation power, thereby improving the accuracy and stability of scheduling. The present application realizes the reasonable scheduling of the output power of different generator sets, ensuring the stability of the complex distributed power supply system, while improving the flexibility of system scheduling.

[0061] like Figure 2 As shown, the present application also provides a distributed power supply method based on a smart energy management platform, which is used in a distributed power supply system. The distributed power supply system includes at least two generator sets and an energy management platform; the at least two generator sets include a first generator set and a second generator set, and the generator types of the at least two generator sets are different; the energy management platform includes a power prediction module, a weight allocation module, a discrimination module, and a scheduling module, and the power prediction module includes a global power prediction module and a unit power prediction module, characterized in that the method includes:

[0062] 101. Calculate the expected power usage in the first time period of the power supply area based on the linear regression equation using the global power prediction module ;

[0063] 102. Calculate the maximum unit output power of the first generator set in the first time period based on the first prediction model by the unit power prediction module. , the maximum unit output power of the second generator set in the first time period ; The algorithm based on the above-mentioned first prediction model is a neural network algorithm;

[0064] 103. Setting the loss weight w1 of the first generator set and the loss weight w2 of the second generator set through a weight allocation module;

[0065] 104. When the discrimination module identifies that the at least two generator sets do not include a third generator set, the first genetic algorithm is selected as the global scheduling algorithm; the generator type of the third generator set is different from that of the first generator set and the second generator set;

[0066] 105. Calculate, by means of a scheduling module, the expected power of the first generator set in the first time period so that the fitness function based on the first genetic algorithm is at a minimum value according to the first genetic algorithm. and the expected power of the second generator set in the first time period ;

[0067] The fitness function based on the first genetic algorithm mentioned above is , the above , are the loss functions of the first generator set and the second generator set respectively, , is the output power of the first generator set and the second generator set; The power fluctuation loss weight is used to adjust the weight of the loss caused by power fluctuation to the system;

[0068] The above fitness function The constraints are: , .

[0069] For the extended part and details of the aforementioned distributed power supply method based on the smart energy management platform, please refer to the implementation example of the distributed power supply system based on the smart energy management platform, which will not be repeated here.

[0070] like Figure 3The figure shows a schematic diagram of the structure of a distributed power supply device based on a smart energy management platform provided by the present application. The device includes: at least one processor 310, such as a central processing unit (CPU), at least one memory 320, and at least one bus 330. The memory 320 may store program instructions, and the processor 310 may be used to call the program instructions to execute a distributed power supply method based on a smart energy management platform.

Claims

1. A distributed power supply system based on a smart energy management platform, comprising at least two generator sets and an energy management platform, characterized in that: The at least two generator sets include a first generator set and a second generator set, the generator types of the at least two generator sets are different, and the energy management platform includes: Power prediction module, including global power prediction module and unit power prediction module; The global power prediction module is used to calculate the expected power usage in the first time period of the power supply area based on the linear regression equation ; The unit power prediction module is used to calculate the maximum unit output power of the first generator set in the first time period based on the first prediction model , the maximum unit output power of the second generator set in the first time period ; The algorithm based on the first prediction model is a neural network algorithm; A weight allocation module, used to set a loss weight w1 of the first generator set and a loss weight w2 of the second generator set; A discrimination module, configured to select the first genetic algorithm as the global scheduling algorithm when it is identified that the at least two generator sets do not include a third generator set; the generator type of the third generator set is different from that of the first generator set and the second generator set; a scheduling module, configured to calculate, based on the first genetic algorithm, the expected power of the first generator set in the first time period so that the fitness function based on the first genetic algorithm is at a minimum value; and the expected power of the second generator set in the first time period ; The fitness function of the first genetic algorithm is: , , are the loss functions of the first generator set and the second generator set respectively, , are the output powers of the first generator set and the second generator set respectively; The power fluctuation loss weight is used to adjust the weight of the loss caused by power fluctuation to the system; The fitness function The constraints are: , ; The generator type of the first generator set is a thermal generator, the generator type of the second generator set is a wind generator, the generator type of the third generator set is an energy storage generator or a hydroelectric generator, and the loss function of the first generator set , the loss function of the second generator set is , a1, a2, a3, b1, b2, b3 are the loss constants of the generator; The chromosome matrix used by the first genetic algorithm is =[ ], wherein said They are , Vectorized representation of ; The crossover operator based on the first genetic algorithm is ; Wherein, t is the number of iterations, is the number of iterations in which The chromosome matrix with the smallest value, wherein c1 and c2 are random vectors; The mutation operator based on the first genetic algorithm is ,in, is a random vector, , .

2. The system according to claim 1, characterized in that: The first prediction model includes a first neural network algorithm and a second neural network algorithm, and the unit power prediction module is specifically used for: Inputting historical data of different time periods belonging to the same day into the first neural network to obtain a first predicted maximum output power , , inputting the historical data of the first time period belonging to different days into the second neural network to obtain the second predicted maximum output power , ; The first predicted maximum output power and the second predicted maximum output power are weighted to obtain the , .

3. The system according to claim 2, characterized in that: The loss weight w1 of the first generator set is greater than the loss weight w2 of the second generator set.

4. The system according to claim 1, characterized in that: The discrimination module is also used for: When it is identified that the at least two generator sets include a third generator set, a second genetic algorithm is selected as a global scheduling algorithm; wherein the complexity of the second genetic algorithm is less than the complexity of the first genetic algorithm, and the complexity is used to represent the time consumed by the operation.

5. The system according to claim 4, characterized in that: The weight distribution module comprises: A monitoring module is used to monitor weight setting instructions; A verification module, configured to read instruction verification information when obtaining a weight setting instruction, and verify the weight setting instruction based on the instruction verification information; The parsing module is used to parse the weight setting instruction when the weight setting instruction is verified, and obtain the loss weight w1 of the first generator set and the loss weight w2 of the second generator set.

6. The system according to claim 2, characterized in that: The scheduling module is also used for: Sending a scheduling instruction to the at least two generator sets, wherein the scheduling instruction is used to set the output power of the first generator set to , set the output power of the second generator set to .

7. A distributed power supply method based on a smart energy management platform, used in a distributed power supply system, wherein the distributed power supply system comprises at least two generator sets and an energy management platform; the at least two generator sets comprise a first generator set and a second generator set, and the generator types of the at least two generator sets are different; the energy management platform comprises a power prediction module, a weight allocation module, a discrimination module, and a scheduling module, and the power prediction module comprises a global power prediction module and a unit power prediction module, characterized in that: The method comprises: The expected power usage of the power supply area in the first time period is calculated by the global power prediction module based on the linear regression equation ; The maximum unit output power of the first generator set in the first time period is calculated by the unit power prediction module based on the first prediction model. , the maximum unit output power of the second generator set in the first time period ; The algorithm based on the first prediction model is a neural network algorithm; Setting the loss weight w1 of the first generator set and the loss weight w2 of the second generator set by a weight allocation module; When the discrimination module identifies that the at least two generator sets do not include a third generator set, the first genetic algorithm is selected as the global scheduling algorithm; the generator type of the third generator set is different from that of the first generator set and the second generator set; The scheduling module calculates the expected power of the first generator set in the first time period so that the fitness function based on the first genetic algorithm is at a minimum value according to the first genetic algorithm. and the expected power of the second generator set in the first time period ; The fitness function of the first genetic algorithm is: , , are the loss functions of the first generator set and the second generator set respectively, , are the output powers of the first generator set and the second generator set respectively; The power fluctuation loss weight is used to adjust the weight of the loss caused by power fluctuation to the system; The fitness function The constraints are: , ; The generator type of the first generator set is a thermal generator, the generator type of the second generator set is a wind generator, the generator type of the third generator set is an energy storage generator or a hydroelectric generator, and the loss function of the first generator set , the loss function of the second generator set is , a1, a2, a3, b1, b2, b3 are the loss constants of the generator; The chromosome matrix used by the first genetic algorithm is =[ ], wherein said They are , Vectorized representation of ; The crossover operator based on the first genetic algorithm is ; Wherein, t is the number of iterations, is the number of iterations in which The chromosome matrix with the smallest value, wherein c1 and c2 are random vectors; The mutation operator based on the first genetic algorithm is ,in, is a random vector, , .

Citation Information

Patent Citations

  • Micro power supply output optimization method and device in smart energy looped network

    CN111668882A

  • Output power scheduling method and device based on hybrid generator set, and terminal

    CN114329317A

  • Reliability optimization method of distributed power supply system

    CN102509175A

  • Method and device for planning distributed type power supply system

    CN104218578A

  • Distributing type energy access system

    CN205544325U