CN118249408A - Grid-connected hybrid renewable energy system based on combination optimization and machine learning algorithm - Google Patents
Grid-connected hybrid renewable energy system based on combination optimization and machine learning algorithm Download PDFInfo
- Publication number
- CN118249408A CN118249408A CN202410674710.6A CN202410674710A CN118249408A CN 118249408 A CN118249408 A CN 118249408A CN 202410674710 A CN202410674710 A CN 202410674710A CN 118249408 A CN118249408 A CN 118249408A
- Authority
- CN
- China
- Prior art keywords
- particle
- grid
- power generation
- optimal
- energy system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 31
- 238000010801 machine learning Methods 0.000 title claims abstract description 27
- 238000010248 power generation Methods 0.000 claims abstract description 57
- 238000013480 data collection Methods 0.000 claims abstract description 4
- 239000002245 particle Substances 0.000 claims description 101
- 230000006870 function Effects 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000000034 method Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 6
- 230000035772 mutation Effects 0.000 claims description 5
- 230000001133 acceleration Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 230000006978 adaptation Effects 0.000 claims description 2
- 238000012549 training Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 3
- 230000005611 electricity Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000002068 genetic effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Abstract
The application discloses a grid-connected hybrid renewable energy system based on a combination optimization and machine learning algorithm, which comprises the following components: a data collection module; photovoltaic power generation module: converting sunlight into mathematical representation of electric energy to obtain total photovoltaic power generation; converting wind power into mathematical representation of electric energy to obtain total wind power generation capacity; PSO-GA-LSTM mixing module: obtaining optimal system configuration combination according to the maximized power grid use coefficient and outputting the optimal system configuration combination; and (3) implementing an energy management module: according to the optimal system configuration combination, the corresponding energy system management strategy is implemented, and each data index is continuously monitored to evaluate the actual sustainability and reliability of the grid-connected hybrid renewable energy system based on the combination optimization and machine learning algorithm.
Description
Technical Field
The application relates to the technical field of data processing, in particular to a grid-connected hybrid renewable energy system based on a combination optimization and machine learning algorithm.
Background
In recent years, in order to cope with global climate problems caused by conventional energy sources, generation of electricity using HRES is gradually increasing and gradually replacing conventional generation of electricity. As a new state of the energy industry, hybrid renewable energy systems are systems that contain wind energy, solar energy, hybrid solar energy and wind energy to generate electricity and to store energy. However, hybrid renewable energy systems are susceptible to external climates and geographical factors, and suffer from instability and volatility. The development of the intelligent power grid is greatly promoted under the promotion of the Internet of things technology, and the integration of the hybrid renewable energy system into the intelligent power grid can effectively improve the reliability and the sustainability of power supply, and is an important technical means for promoting the development of clean and sustainable energy.
Constructing a sustainable and reliable energy system in the smart grid, optimizing HRES configuration is a crucial process. The current HRES optimization method is mainly a meta heuristic method, such as a genetic algorithm, such as Bilil H (morocco, 2014), a particle swarm optimization algorithm, such as Sharafi M (canada, 2015), and an evolutionary algorithm, such as Wang Rui (national defense university of science and technology, 2015), which can optimize the constituent elements of the hybrid renewable energy system to obtain an optimal configuration combination. The conventional particle swarm optimization algorithm obtains the position and the speed of the particles through random initialization, and the problem that the particles are easy to fall into local optimum cannot be avoided.
Disclosure of Invention
The application provides a grid-connected hybrid renewable energy system based on a combination optimization and machine learning algorithm to solve the problems in the background technology.
The technical scheme adopted for solving the technical problems is as follows: a grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm, comprising:
And a data collection module: acquiring data affecting photovoltaic power generation and data affecting wind power generation;
Photovoltaic power generation module: establishing a photovoltaic power generation module based on data influencing photovoltaic power generation, and converting sunlight into mathematical representation of electric energy to obtain the total power generation of the photovoltaic;
Wind power generation module: establishing a wind power generation module based on data influencing wind power generation, and converting wind power into mathematical expression of electric energy to obtain total wind power generation capacity;
PSO-GA-LSTM mixing module: initializing PSO, GA, LSTM algorithm parameters, wherein the algorithm parameters comprise particle speed, particle position and maximum iteration times, the number of photovoltaic panels, total photovoltaic power generation amount, the number of wind driven generators, total wind power generation amount and the number and capacity of system battery storage devices are used as parameters to be input into PSO, GA, LSTM algorithm, a function of calculated power grid use coefficients is used as fitness function, the total capacity actually output by the system in a certain time period is predicted, the power grid use coefficients are the ratio of the total capacity actually output by the system to the total rated capacity of the system, the maximum power grid use coefficient in a certain time period is calculated according to the predicted total capacity actually output by the system in the certain time period, the optimal system configuration combination is obtained and output according to the maximum power grid use coefficients, and the optimal system configuration combination comprises the optimal photovoltaic panel number, the photovoltaic total power generation amount, the total wind driven generator number, the total wind power generation amount and the number and capacity of the system battery storage devices;
And (3) implementing an energy management module: and implementing a corresponding energy system management strategy according to the optimal system configuration combination, and continuously monitoring each data index to evaluate the actual sustainability and reliability of the grid-connected hybrid renewable energy system based on the combination optimization and the machine learning algorithm.
The grid-connected hybrid renewable energy system based on the combined optimization and machine learning algorithm mixes PSO and GA, has the characteristics of global convergence and genetic characteristics of GA, has the characteristics of high convergence speed and high memory of the PSO algorithm, is used as a time prediction network based on machine learning, has strong time sequence modeling and complex data processing capacity, can accurately predict the generated energy of the hybrid energy system based on the acquired historical data and assist the smart grid to reasonably configure and optimize HRES, and combines the PSO and GA optimization algorithm and the LSTM-based prediction algorithm to construct a hybrid PSO-GA-LSTM optimization and prediction model, so that the grid use coefficient is effectively improved, and considerable system reliability, environmental sustainability and economic feasibility are also reflected.
Preferably, the PSO-GA-LSTM hybrid module computes the maximum grid usage factor including,
Step A: calculating the fitness y of the current position of the particle by using a fitness function, and combining the fitness y of the current position of the particle with the fitness of the optimal position of the particle historyComparing, if the particle current position fitness y is greater than the particle historical optimal position fitness/>The current position of the particle is taken as the optimal position of the particle history and the adaptation degree of the optimal position of the particle history is updated；
And (B) step (B): selecting the position of each particle 2/3 to perform the variation operation of the particle speed, and finishing the update of the particle speed;
step C: each particle after the mutation operation is respectively and continuously subjected to cross operation with the corresponding individual extremum particle and the global extremum particle, so that the updating of the particle position is completed;
Step D: adding the latest speed information and the latest position information obtained according to the mutation operation and the crossover operation to finish updating the particles;
Step E: and D, repeating the steps A to D, and judging whether the power grid use coefficient is better than the value of the previous generation to obtain the maximized power grid use coefficient.
Preferably, the corresponding parameters are updated according to the maximized power grid use coefficient, the optimal parameter configuration combination is obtained, the learning training of the LSTM network is carried out on the basis of the preset LSTM network parameters and the learning mode and the optimal parameter configuration combination until the output condition is met, the time series data of the energy system are predicted to obtain the predicted output electric energy of the hybrid energy system, and the predicted output electric energy of the hybrid energy system is used for calculating the power grid use coefficient for evaluating the reliability and the sustainability of the system.
Preferably, the variation operation of the particle velocity is as follows:， Where k represents the number of iterations at particle 2/3; v i is the particle velocity of individual i; v min、Vmax is the minimum and maximum value of particle velocity, respectively; a 1、a2 is a random number between [0,1 ]; g max is the maximum number of evolutions; f (k) represents the coefficient of variation in speed at the number of iterations of k.
Preferably, the crossover operation is as follows: Wherein X i (k) and X j (k) represent the position information of the paired individual i and individual j at the kth iteration, X i (k+1) and X j (k+1) paired individual i and individual j after the completion of the crossover operation, a 3 is a random number between [0,1], and k represents the number of iterations.
Preferably, the latest particle velocity calculation formula is: W is an inertia factor, c 1、c2 is an acceleration factor, r 1、r2 is a random number, and the value range is (0, 1)/> Representing the optimal position searched for by particle i,/>Representing the global optimal position in the space searched by the particle swarm,/>Indicating the velocity of particle i at time t,/>The position of particle i at time t is indicated.
Preferably, the latest particle position calculation formula includes:， wherein the position information of particle i at time t is represented,/> The velocity information of the particle i at time t+1 is shown.
Preferably, the mathematical relationship of the forgetting gate, input gate and output gate state quantities of the LSTM at time t is as follows:，/>， Wherein/> Representing a sigmoid function; x t is the unit data of the input layer at the moment of the model t; h t-1 is the hidden state from the previous time step; w f and b f represent the weight matrix and bias of the forgetting gate, respectively; w i and b i represent the weight matrix and bias of the input gates, respectively; w o and b o represent the weight matrix and bias of the output gates, respectively.
Preferably, the calculation formula of the predicted output electric energy of the LSTM output hybrid energy system is as follows: Wherein o t represents the output gate of LSTM; c t represents the memory cell state; tanh represents an activation operation.
Preferably, the predicted output electric energy of the hybrid energy system is used for calculating the grid use coefficient, and an optimal system configuration combination is obtained and output, wherein the optimal system configuration combination comprises the optimal photovoltaic panel number, the photovoltaic total power generation amount, the wind driven generator number, the wind power total power generation amount and the number and the capacity of the system battery storage equipment for evaluating the reliability and the sustainability of the system.
The application has the following substantial effects:
1. The grid-connected hybrid renewable energy system based on the combined optimization and machine learning algorithm uses PSO and GA in a mixed mode, has global convergence and genetic characteristics of GA, and has the characteristics of high convergence speed and high memory of the PSO algorithm;
2. In the grid-connected hybrid renewable energy system based on the combined optimization and machine learning algorithm, the LSTM is used as a time prediction network based on machine learning, has strong time sequence modeling and complex data processing capacity, can accurately predict the generated energy of the hybrid energy system based on the existing optimization result and the collected historical data, and assists the smart grid to reasonably configure and optimize HRES;
3. The grid-connected hybrid renewable energy system is designed based on a combination optimization and machine learning algorithm, and the most effective combination of the number of photovoltaic power generation plates and the generated energy, the number of fan turbines and the generated energy and the number and capacity of energy storage devices is determined.
Drawings
FIG. 1 is a data processing flow diagram of a PSO-GA-LSTM mixing module in accordance with a first embodiment of the present application;
FIG. 2 is a 24-hour energy distribution diagram of a photovoltaic power module according to an embodiment of the present application;
Fig. 3 is a basic unit structure diagram of an LSTM according to a first embodiment of the present application.
Detailed Description
The technical scheme of the application is further specifically described by the following specific examples.
Example 1
A grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm, comprising:
and a data collection module: the method comprises the steps of obtaining data affecting photovoltaic power generation and data affecting wind power generation, including illumination intensity, ambient temperature, wind speed and the like, and carrying out structured data processing processes, including data cleaning, preprocessing and organization, so as to ensure the consistency and quality of the data.
Photovoltaic power generation module: the photovoltaic power generation module is established based on data affecting photovoltaic power generation, sunlight is converted into mathematical representation of electric energy, and total photovoltaic power generation is obtained, and as shown in fig. 2, the electric energy output power calculation formula of a single photovoltaic power generation plate is as follows: Wherein/> Representing the electrical energy output of a single photovoltaic panel at time t,/>Representing the reduction coefficient of dust accumulation,/>Representing the conversion efficiency of a photovoltaic panel,/>Representing the area of the photovoltaic panel,/>Representing the temperature coefficient,/>Indicating the current battery temperature;
The calculation of the total photovoltaic power generation is as follows: Wherein, the method comprises the steps of, wherein, Representing the total power generated by a group of photovoltaic panels at time t,/>Represents the total number of photovoltaic panels,Representing the output power of an individual photovoltaic panel at time t.
Wind power generation module: the method comprises the steps of establishing a wind power generation module based on data influencing wind power generation, converting wind power into mathematical expression of electric energy, and obtaining total wind power generation capacity, wherein the calculation formula of the output power of a single wind power generator is as follows: Wherein/> Representing the output power of the wind turbine at time t,/>Represents air density, A represents rotor swept area of the wind turbine,/>Representing the power coefficient of the wind turbine, V representing the wind speed;
The total amount of wind power generation is calculated as follows: Wherein/> Representing the total power output by a group of wind turbines at time t,/>Representing the number of wind turbines to be used,Representing the output power of an individual wind turbine at time t.
PSO-GA-LSTM mixing module: initializing PSO, GA, LSTM algorithm parameters, wherein the algorithm parameters comprise particle speed, particle position and maximum iteration times, the number of photovoltaic panels, total photovoltaic power generation amount, the number of wind driven generators, total wind power generation amount and the number and capacity of system battery storage devices are used as parameters to be input into PSO, GA, LSTM algorithm, a function of calculated power grid use coefficients is used as fitness function, the total capacity actually output by the system in a certain time period is predicted, the power grid use coefficients are the ratio of the total capacity actually output by the system to the total rated capacity of the system, the maximum power grid use coefficient in a certain time period is calculated according to the predicted total capacity actually output by the system in the certain time period, the optimal system configuration combination is obtained and output according to the maximum power grid use coefficients, and the optimal system configuration combination comprises the optimal photovoltaic panel number, the photovoltaic total power generation amount, the total wind driven generator number, the total wind power generation amount and the number and capacity of the system battery storage devices;
as shown in fig. 1, the PSO-GA-LSTM hybrid module calculating the maximized grid usage coefficient includes:
Step A1: the method comprises the steps of inputting the number of photovoltaic panels, the total photovoltaic power generation amount, the number of wind driven generators, the total wind power generation amount and the number and capacity of storage equipment of a system battery into a PSO-GA-LSTM hybrid module as parameters, and selecting a function for calculating a power grid use coefficient as a fitness function.
Step A2: calculating the fitness y of the current position of the particle by using a fitness function, and combining the fitness y of the current position of the particle with the fitness of the optimal position of the particle historyComparing, if the particle current position fitness y is greater than the particle historical optimal position fitness/>Then the current position of the particle is taken as the optimal position of the particle history and the fitness/>, of the optimal position of the particle history, is updated。
Step A3: in order to balance global search and local search, reduce search space and avoid over randomization, selecting 2/3 position of each particle to perform variation operation of particle speed, and finishing update of particle speed;
the variation operation process of the particle velocity is as follows: ， Where k represents the number of iterations at particle 2/3; v i is the particle velocity of individual i; v min、Vmax is the minimum and maximum value of particle velocity, respectively; a 1、a2 is a random number between [0,1 ]; g max is the maximum number of evolutions; f (k) represents the coefficient of variation in speed at the number of iterations of k.
Step A4: and (3) continuing to perform cross operation on each particle subjected to mutation operation and corresponding individual extremum particles and global extremum particles to finish updating of particle positions, wherein the individual extremum particles are usually stored in a data structure of each particle, and in each iteration, the particles compare the fitness value of the current position and the individual optimal position, and if the current position is better, the individual optimal position is updated. The global extremum particles are the optimal positions which are historically achieved by all particles in the whole particle swarm, namely the optimal particles, in each iteration, the whole particle swarm compares the individual optimal positions of each particle, the global optimal positions are found and recorded as the global extremum particles, then the positions of 2/3 of the total dimension of each particle are randomly selected to be crossed with the optimal particles, and the crossing operation process is as follows: Wherein X i (k) and X j (k) represent the position information of the paired individual i and individual j at the kth iteration, X i (k+1) and X j (k+1) paired individual i and individual j after the completion of the crossover operation, a 3 is a random number between [0,1], and k represents the number of iterations.
Step A5: adding the latest speed information and the latest position information obtained according to the mutation operation and the crossover operation to finish updating the particles;
the latest particle velocity calculation formula is: W is an inertia factor, c 1、c2 is an acceleration factor, r 1、r2 is a random number, and the value range is (0, 1)/> Representing the optimal position searched for by particle i,/>Representing the global optimal position in the space searched by the particle swarm,/>The velocity of the particle i at time t is indicated,Indicating the position of particle i at time t;
the latest particle position calculation formula includes: ，/> wherein the position information of particle i at time t is represented,/> The velocity information of the particle i at time t+1 is shown.
Step A6: repeating the steps A2 to A5, judging whether the power grid use coefficient is better than the previous generation value, obtaining the maximized power grid use coefficient, updating the corresponding parameter according to the maximized power grid use coefficient, obtaining the optimal parameter configuration combination, carrying out the learning training of the LSTM network based on the preset LSTM network parameter and the learning mode and adopting the optimal parameter configuration combination until the output condition is met, predicting the time series data of the energy system to obtain the predicted output electric energy of the hybrid energy system,
As shown in fig. 3, the mathematical relationship of the forget gate, input gate and output gate state quantities of LSTM at time t is as follows:，/>， Wherein/> Representing a sigmoid function; x t is the unit data of the input layer at the moment of the model t; h t-1 is the hidden state from the previous time step; w f and b f represent the weight matrix and bias of the forgetting gate, respectively; w i and b i represent the weight matrix and bias of the input gates, respectively; w o and b o represent the weight matrix and bias of the output gates, respectively;
The calculation formula of the predicted output electric energy of the LSTM output hybrid energy system is as follows: Wherein o t represents the output gate of LSTM; c t represents the memory cell state; tanh represents an activation operation.
And calculating a power grid use coefficient by using the predicted output electric energy of the hybrid energy system, and obtaining and outputting an optimal system configuration combination, wherein the optimal system configuration combination comprises the optimal photovoltaic panel number, the optimal total power generation amount, the optimal wind power generator number, the optimal total power generation amount, and the optimal number and capacity of system battery storage devices for evaluating the reliability and the sustainability of the system.
And implementing a corresponding energy system management strategy according to the optimal system configuration combination, and continuously monitoring each data index to evaluate the actual sustainability and reliability of the grid-connected hybrid renewable energy system based on the combination optimization and the machine learning algorithm.
The above-described embodiment is only a preferred embodiment of the present application, and is not limited in any way, and other variations and modifications may be made without departing from the technical aspects set forth in the claims.
Claims (10)
1. A grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm, comprising:
And a data collection module: acquiring data affecting photovoltaic power generation and data affecting wind power generation;
Photovoltaic power generation module: establishing a photovoltaic power generation module based on data influencing photovoltaic power generation, and converting sunlight into mathematical representation of electric energy to obtain the total power generation of the photovoltaic;
Wind power generation module: establishing a wind power generation module based on data influencing wind power generation, and converting wind power into mathematical expression of electric energy to obtain total wind power generation capacity;
PSO-GA-LSTM mixing module: initializing PSO, GA, LSTM algorithm parameters, wherein the algorithm parameters comprise particle speed, particle position and maximum iteration times, the number of photovoltaic panels, total photovoltaic power generation amount, the number of wind driven generators, total wind power generation amount and the number and capacity of system battery storage devices are used as parameters to be input into PSO, GA, LSTM algorithm, a function of calculated power grid use coefficients is used as fitness function, the total capacity actually output by the system in a certain time period is predicted, the power grid use coefficients are the ratio of the total capacity actually output by the system to the total rated capacity of the system, the maximum power grid use coefficient in a certain time period is calculated according to the predicted total capacity actually output by the system in the certain time period, the optimal system configuration combination is obtained and output according to the maximum power grid use coefficients, and the optimal system configuration combination comprises the optimal photovoltaic panel number, the photovoltaic total power generation amount, the total wind driven generator number, the total wind power generation amount and the number and capacity of the system battery storage devices;
And (3) implementing an energy management module: and implementing a corresponding energy system management strategy according to the optimal system configuration combination, and continuously monitoring each data index to evaluate the actual sustainability and reliability of the grid-connected hybrid renewable energy system based on the combination optimization and the machine learning algorithm.
2. The grid-tie hybrid renewable energy system based on a combined optimization and machine learning algorithm of claim 1 wherein the PSO-GA-LSTM hybrid module calculating the maximized grid usage factor comprises,
Step A: calculating the fitness y of the current position of the particle by using a fitness function, and combining the fitness y of the current position of the particle with the fitness of the optimal position of the particle historyComparing, if the fitness y of the current position of the particle is greater than the fitness y of the historical optimal position of the particleThe current position of the particle is taken as the optimal position of the particle history and the adaptation degree of the optimal position of the particle history is updated；
And (B) step (B): selecting the position of each particle 2/3 to perform the variation operation of the particle speed, and finishing the update of the particle speed;
step C: each particle after the mutation operation is respectively and continuously subjected to cross operation with the corresponding individual extremum particle and the global extremum particle, so that the updating of the particle position is completed;
Step D: adding the latest speed information and the latest position information obtained according to the mutation operation and the crossover operation to finish updating the particles;
Step E: and D, repeating the steps A to D, and judging whether the power grid use coefficient is better than the value of the previous generation to obtain the maximized power grid use coefficient.
3. The grid-connected hybrid renewable energy system based on the combination optimization and machine learning algorithm according to claim 2, wherein the corresponding parameters are updated according to the maximized grid use coefficient to obtain the optimal parameter configuration combination, the learning training of the LSTM network is performed by adopting the optimal parameter configuration combination based on the preset LSTM network parameters and the learning mode until the output condition is met, the time series data of the energy system is predicted to obtain the predicted output electric energy of the hybrid energy system, and the grid use coefficient is calculated by using the predicted output electric energy of the hybrid energy system for evaluating the reliability and the sustainability of the system.
4. The grid-connected hybrid renewable energy system based on a combination optimization and machine learning algorithm of claim 2, wherein the variation operation process of the particle velocity is as follows:， Where k represents the number of iterations at particle 2/3; v i is the particle velocity of individual i; v min、Vmax is the minimum and maximum value of particle velocity, respectively; a 1、a2 is a random number between [0,1 ]; g max is the maximum number of evolutions; f (k) represents the coefficient of variation in speed at the number of iterations of k.
5. The grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm of claim 4, wherein the crossover operation process is as follows: Wherein X i (k) and X j (k) represent the position information of the paired individual i and individual j at the kth iteration, X i (k+1) and X j (k+1) paired individual i and individual j after the completion of the crossover operation, a 3 is a random number between [0,1], and k represents the number of iterations.
6. The grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm of claim 5, wherein the most recent particle velocity calculation formula is: W is an inertia factor, c 1、c2 is an acceleration factor, r 1、r2 is a random number, and the value range is (0, 1)/> Representing the optimal position searched for by particle i,/>Representing the global optimal position in the space searched by the particle swarm,/>Indicating the velocity of particle i at time t,/>The position of particle i at time t is indicated.
7. The grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm of claim 6, wherein the most recent particle location calculation formula comprises:，/> wherein the position information of particle i at time t is represented,/> The velocity information of the particle i at time t+1 is shown.
8. The grid-tie hybrid renewable energy system based on a combination optimization and machine learning algorithm according to claim 3, wherein the mathematical relationships of the forget gate, input gate and output gate state quantities of LSTM at time t are as follows:，/>， Wherein/> Representing a sigmoid function; x t is the unit data of the input layer at the moment of the model t; h t-1 is the hidden state from the previous time step; w f and b f represent the weight matrix and bias of the forgetting gate, respectively; w i and b i represent the weight matrix and bias of the input gates, respectively; w o and b o represent the weight matrix and bias of the output gates, respectively.
9. The grid-connected hybrid renewable energy system based on a combination optimization and machine learning algorithm of claim 8, wherein the predicted output power calculation formula of the LSTM output hybrid energy system is as follows: Wherein o t represents the output gate of LSTM; c t represents the memory cell state; tanh represents an activation operation.
10. The grid-connected hybrid renewable energy system based on a combination optimization and machine learning algorithm according to claim 9, wherein the predicted output power of the hybrid energy system is used to calculate the grid usage factor and obtain an optimal system configuration combination and output, the optimal system configuration combination including an optimal number of photovoltaic panels, a photovoltaic total power generation amount, a wind power generator number, a wind total power generation amount, and the number and capacity of system battery storage devices is used to evaluate the reliability and sustainability of the system.
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118249408A true CN118249408A (en) | 2024-06-25 |
Family
ID=
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Data-driven configuration optimization of an off-grid wind/PV/hydrogen system based on modified NSGA-II and CRITIC-TOPSIS | |
| Ghorbani et al. | Optimizing a hybrid wind-PV-battery system using GA-PSO and MOPSO for reducing cost and increasing reliability | |
| CN112583017B (en) | Hybrid micro-grid energy distribution method and system considering energy storage operation constraint | |
| CN110188915A (en) | Energy-storage system Optimal Configuration Method and system in virtual plant based on scene collection | |
| CN109103929A (en) | Based on the power distribution network economic optimization dispatching method for improving dynamic gram Li Sijin model | |
| CN113887858A (en) | Charging station micro-grid system optimal scheduling method based on CNN-LSTM load prediction | |
| Rouhani et al. | A comprehensive method for optimum sizing of hybrid energy systems using intelligence evolutionary algorithms | |
| CN116667325B (en) | Micro-grid-connected operation optimization scheduling method based on improved cuckoo algorithm | |
| CN111917111B (en) | Method, system, equipment and storage medium for online evaluation of distributed photovoltaic power supply acceptance capacity of power distribution network | |
| CN115526401A (en) | Novel power supply optimal planning method for electric power system based on digital twinning | |
| CN115169916A (en) | Electric heating comprehensive energy control method based on safety economy | |
| CN116402210A (en) | Multi-objective optimization method, system, equipment and medium for comprehensive energy system | |
| CN115423153A (en) | Photovoltaic energy storage system energy management method based on probability prediction | |
| Murugesan et al. | Cost optimization of PV-diesel systems in nanogrid using LJ Cuckoo search and its application in mobile towers | |
| Benatiallah et al. | Modelling and optimisation of wind energy systems | |
| Zhang et al. | Joint forecasting of regional wind and solar power based on attention neural network | |
| Yin et al. | Improved genetic algorithm-based optimization approach for energy management of microgrid | |
| Jemaa et al. | Optimum sizing of hybrid PV/Wind/battery installation using a fuzzy PSO | |
| CN107104429B (en) | power system load scheduling method considering distributed energy storage system | |
| Rouhani et al. | A teaching learning based optimization for optimal design of a hybrid energy system | |
| CN118249408A (en) | Grid-connected hybrid renewable energy system based on combination optimization and machine learning algorithm | |
| Ming et al. | Multi-objective optimization of hybrid renewable energy system with load forecasting | |
| Guru et al. | Wind generator and storage system scheduling for customer benefit and battery life | |
| Fang et al. | Energy scheduling and decision learning of combined cooling, heating and power microgrid based on deep deterministic policy gradient | |
| CN112270440B (en) | Power distribution network load prediction method based on capsule neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication |