CN115618753A - Hybrid energy storage system joint optimization method for frequency-adjustable pulse working condition - Google Patents
Hybrid energy storage system joint optimization method for frequency-adjustable pulse working condition Download PDFInfo
- Publication number
- CN115618753A CN115618753A CN202211629642.9A CN202211629642A CN115618753A CN 115618753 A CN115618753 A CN 115618753A CN 202211629642 A CN202211629642 A CN 202211629642A CN 115618753 A CN115618753 A CN 115618753A
- Authority
- CN
- China
- Prior art keywords
- working condition
- frequency
- energy storage
- storage system
- energy management
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/28—Arrangements for balancing of the load in a network by storage of energy
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
- H02J3/48—Controlling the sharing of the in-phase component
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/04—Power grid distribution networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/10—Power transmission or distribution systems management focussing at grid-level, e.g. load flow analysis, node profile computation, meshed network optimisation, active network management or spinning reserve management
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Abstract
The invention discloses a hybrid energy storage system joint optimization method for a frequency-adjustable pulse working condition, which comprises the following steps of: generating a plurality of capacity allocation schemes, and determining corresponding optimal energy management rules to form a training data set; training the neural network model based on the training data set; aiming at the high-power pulse working condition with adjustable frequency, outputting a global optimal capacity configuration scheme and an energy management rule under a given working condition based on a neural network model under a combined optimization framework; aiming at the high-power pulse working condition with specific frequency, the optimal energy management rule is obtained in real time based on the neural network model, and the energy management rule is updated in real time when the pulse frequency is adjusted. The method is applied to the field of hybrid energy storage, the influence of capacity configuration and energy management on the final performance of the system is considered in a combined mode, the global optimal solution of the capacity configuration optimization problem and the energy management optimization problem which are highly coupled can be found, and the optimization speed is greatly improved on the premise of ensuring the optimality of the solution.
Description
Technical Field
The invention relates to the technical field of hybrid energy storage, in particular to a hybrid energy storage system joint optimization method for frequency-adjustable pulse working conditions.
Background
The research on energy storage devices or energy storage systems having high power density characteristics and high energy density characteristics is a constant subject of the development of energy storage technology. At present, the requirements of a variable-frequency high-power pulse load on high energy density and high power density are difficult to meet simultaneously only by adopting a single type of energy storage device, so that the hybrid energy storage system integrating the energy type energy storage device and the power type energy storage device has a larger application value and a good prospect in the application scene.
Currently, research on hybrid energy storage systems mainly includes both capacity configuration and energy management of the hybrid energy storage systems. The capacity configuration scheme of the hybrid energy storage system specifically comprises the contents of the models, the number, the series-parallel connection mode and the like of the two types of devices, and determines the upper limit of the performance of the hybrid energy storage system. That is, if the hybrid energy storage system adopts a capacity allocation scheme that does not meet the requirement of the load operation condition, even if the hardware performance can be fully exerted, the hybrid energy storage system cannot meet the actual operation requirement of the load. Therefore, in the case that the capacity configuration can be designed in advance, it is necessary to optimize the capacity configuration for the hybrid energy storage system based on the load operation condition. The energy management of the hybrid energy storage system depends on the actual capacity allocation of the two types of devices in the system, and the purpose of the energy management is to enable the hybrid energy storage system to achieve the optimal performance under the given device capacity allocation scheme. The main means is to fully play the respective advantages of the energy type device and the power type device by reasonably distributing the power between the energy type device and the power type device, thereby not only providing higher power output for a load, but also providing higher continuous energy output for the load.
Current research often performs capacity configuration optimization and energy management optimization in two steps. The method comprises the steps of firstly, evaluating the performances of different capacity configuration schemes by adopting a universal energy management rule and optimizing, and secondly, optimizing energy management aiming at capacity configuration optimizing results and obtaining the corresponding optimal energy management strategy. When the capacity configuration optimization of the hybrid energy storage system is carried out, each capacity configuration scheme can exert the optimal performance of the scheme only by adopting the corresponding optimal energy management rule, so that the global optimality of the optimal capacity configuration scheme found by adopting a step-by-step method cannot be ensured.
In order to solve the problem that the step-by-step method cannot ensure global optimality, a small amount of research considers the joint optimization of capacity configuration and energy management at present. Some of the researches are limited by the traditional optimization method adopted by the research, and the capacity configuration value is often artificially over-discretized or the value range is limited in a narrower interval according to experience; the other part of the existing combined optimization research adopts an intelligent optimization method, but the energy management optimization of the combined optimization research either simply adopts parameter optimization of a piecewise linear function constructed based on experience and loses optimality, or the global optimization methods such as dynamic planning and the like are directly nested under an intelligent optimization framework, so that the calculation cost is overlarge. In addition, the current research related research of the hybrid energy storage system mainly aims at the working conditions with strong random characteristics, such as running of electric vehicles, power grid peak regulation and the like, the research on the periodic high-power pulse working conditions is less, and the research related to the hybrid energy storage system facing the high-power pulse working conditions with adjustable frequency is lacked.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a hybrid energy storage system joint optimization method facing a frequency-adjustable pulse working condition, which can find a global optimal solution of a highly-coupled capacity configuration optimization problem and an energy management optimization problem by jointly considering the influence of a capacity configuration scheme and an energy management rule on the final performance of a system facing a high-power pulse working condition with adjustable frequency. In addition, on the premise of ensuring the optimality of the solution, the optimization speed is greatly improved by combining a neural network method, and the calculation complexity in the whole solving process can be flexibly adjusted according to the actual requirement on the scheme optimality.
In order to achieve the purpose, the invention provides a hybrid energy storage system joint optimization method facing to a frequency adjustable pulse working condition, which comprises the following steps:
generating a plurality of capacity allocation schemes in the application constraint range of the hybrid energy storage system based on the alternative models of the energy type devices and the power type devices and device parameters, and determining the corresponding optimal energy management rule of each capacity allocation scheme under the periodic pulse working condition of a specific frequency to form a training data set;
training a neural network model based on a training data set, so that the neural network model can output a corresponding optimal energy management rule after a capacity configuration scheme and a pulse frequency are input;
aiming at the high-power pulse working condition with adjustable frequency, optimizing based on the mapping relation of capacity allocation-energy management rules established by a neural network model under a combined optimization framework, and outputting a global optimal capacity allocation scheme and energy management rules under a given working condition;
in practical application, aiming at a high-power pulse working condition with a specific frequency, an optimal energy management rule is obtained in real time based on a neural network model, and the energy management rule is updated in real time when the pulse frequency is adjusted, so that the optimal performance of the hybrid energy storage system is ensured.
In one embodiment, the construction process of the training data set specifically includes:
based on the alternative models of the energy type device and the power type device and device parameters provided by a supply manufacturer, generating a plurality of capacity allocation schemes by adopting a sampling method within the voltage grade constraint, the power performance constraint and the volume and mass constraint ranges of the hybrid energy storage system to obtain a capacity allocation scheme set;
based on the variable range of the pulse frequency in the working condition data, extracting a plurality of pulse frequencies by adopting a sampling method, and forming a working condition data set comprising a plurality of groups of periodic pulse working conditions with different frequencies with fixed parameters in the working condition data;
and optimizing the power distribution of the hybrid energy storage system under the working condition of each frequency in the working condition data set for each capacity configuration scheme to obtain an optimal power distribution curve corresponding to each capacity configuration scheme under the periodic pulse working condition of a plurality of frequencies, fitting the optimal power distribution curve, and extracting fitting function parameters as numerical representation of the energy management rule.
In one embodiment, the sampling method is a monte carlo method or a random sampling method.
In one embodiment, a dynamic programming method, a Pontryagin minimum theorem method, a particle swarm algorithm or a genetic algorithm is adopted to optimize the power distribution of the hybrid energy storage system under the working condition of each frequency in the working condition data set.
In one embodiment, the joint optimization framework is composed of a neural network model and an optimization algorithm.
In one embodiment, the optimization algorithm is an evolutionary algorithm, and the optimization is performed based on a mapping relationship between a capacity configuration and an energy management rule established by a neural network model under a joint optimization framework, specifically:
inputting basic data required for optimization including related data of the alternative devices and load working conditions into an evolutionary algorithm, completing power and voltage current constraint of the hybrid energy storage system and constraint expression of the maximum serial-parallel number of the battery and the capacitor monomers, and then extracting feasible solutions in a feasible domain defined by the constraints to form an initial population, namely a set of alternative capacity configuration schemes;
the method comprises the steps of performing simulation calculation on an objective function value of each individual in an initial population based on optimal energy management rules under different pulse frequencies output by a neural network model, performing non-dominated sorting or cross variation operation on a previous generation population according to the quality of the objective function value to generate a next generation population, continuously and circularly iterating to a preset termination algebra in an evolutionary algorithm frame, and outputting a pareto frontier formed by a capacity configuration scheme obtained by the last iteration.
The invention has the following beneficial technical effects:
1. the method can solve the problem of capacity configuration and energy management joint optimization of the hybrid energy storage system and is easy to apply practically;
2. compared with the conventional method for optimizing capacity allocation and optimizing energy management in two steps, the method disclosed by the invention can be used for carrying out combined optimization on the highly-coupled capacity allocation optimization problem and the energy management optimization problem, so that a global optimal scheme which cannot be found by the traditional method can be found;
3. compared with the existing joint optimization method, the optimization speed is greatly improved by combining the neural network method on the premise of ensuring the optimality of the solution, and the calculation complexity in the whole solving process can be flexibly adjusted according to the actual requirement on the optimality of the scheme;
4. in practical application, the method is based on the offline training neural network model, can obtain the optimal energy management scheme under different pulse frequencies in real time, and has strong practical operability.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the structures shown in the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a hybrid energy storage system joint optimization method for a frequency-tunable pulse working condition according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of the offline training of the neural network model according to the embodiment of the present invention;
FIG. 3 is a schematic flow chart of an offline joint optimization based on an evolutionary algorithm in an embodiment of the present invention;
fig. 4 is a schematic flow chart of online energy management of a hybrid energy storage system based on a neural network in an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In addition, the technical solutions in the embodiments of the present invention may be combined with each other, but it must be based on the realization of the technical solutions by those skilled in the art, and when the technical solutions are contradictory to each other or cannot be realized, such a combination of the technical solutions should not be considered to exist, and is not within the protection scope of the present invention.
The embodiment discloses a hybrid energy storage system joint optimization method facing a frequency-adjustable pulse working condition, and mainly comprises three parts, namely neural network model offline training, offline joint optimization based on an evolutionary algorithm and online energy management of a hybrid energy storage system based on a neural network, with reference to fig. 1. Specifically, the neural network model off-line training is to establish mapping from a capacity configuration scheme and pulse working condition data to an optimal energy management rule through a neural network method, wherein label data required by the training are from a large number of capacity configuration schemes, pulse working condition data and the corresponding optimal energy management rule which are obtained by pre-calculation based on a dynamic planning method. The offline joint optimization based on the evolutionary algorithm is that under the framework of the evolutionary algorithm, aiming at given load working conditions and operation environments, the optimal energy management rule output by a neural network which is trained offline is adopted to simulate and calculate the objective function value of each capacity allocation scheme, and the optimal capacity allocation scheme under the given load working conditions is finally output through algorithm operations such as non-dominated sorting, cross variation, cyclic iteration and the like. The online energy management of the hybrid energy storage system based on the neural network is an online application method of a joint optimization scheme, and for the hybrid energy storage system obtained by integration according to an optimal capacity configuration scheme, an optimal energy management rule corresponding to the current pulse frequency can be obtained in real time through the neural network trained in the early stage in the working process of the hybrid energy storage system, and power distribution between a power type device and an energy type device is carried out according to the rule.
In a specific implementation process, the hybrid energy storage system joint optimization method for the frequency-adjustable pulse working condition in the embodiment specifically includes the following steps:
generating a plurality of capacity allocation schemes in the application constraint range of the hybrid energy storage system based on the alternative models of the energy type devices and the power type devices and device parameters, and determining the corresponding optimal energy management rule of each capacity allocation scheme under the periodic pulse working condition of a specific frequency to form a training data set;
training a neural network model based on a training data set, so that the neural network model can output a corresponding optimal energy management rule after a capacity configuration scheme and a pulse frequency are input;
aiming at the high-power pulse working condition with adjustable frequency, optimizing based on the mapping relation of capacity allocation-energy management rules established by a neural network model under a combined optimization framework, and outputting a global optimal capacity allocation scheme and energy management rules under a given working condition;
in practical application, aiming at a high-power pulse working condition with a specific frequency, an optimal energy management rule is obtained in real time based on a neural network model, and the energy management rule is updated in real time when the pulse frequency is adjusted, so that the optimal performance of the hybrid energy storage system is ensured.
In the specific implementation process, the specific operation steps of the offline training of the neural network model are as shown in fig. 2. Because the research results in the aspect of neural network models are abundant at present, the accuracy, convergence speed and other aspects of various neural network models can be balanced easily according to the data characteristics, input and output forms and practical application requirements of the method, and a proper neural network structure can be selected, for example, a feedforward neural network or a feedback neural network can be selected in the embodiment. Therefore, the generation of the training data set required by training is the main content in the offline training operation step of the neural network model, and the training data set required by training is composed of a capacity configuration scheme (the models, the number, the serial-parallel connection mode and the like of energy type devices and power type devices), working condition data (including amplitude, frequency and the like) and a corresponding optimal energy management rule (a fitting function parameter of a curve of total required power and optimal output power of the devices).
In this embodiment, the process of offline training of the neural network model is divided into three steps, which are respectively:
the method comprises the following steps that a large number of capacity configuration schemes are generated by adopting a Monte Carlo sampling method in the voltage grade constraint, the power performance constraint and the volume quality constraint range of a hybrid energy storage system based on the alternative models of the energy type device and the power type device and data such as device parameters, dynamic response characteristics, aging characteristics and the like provided by a supplier, namely a capacity configuration scheme set is generated; meanwhile, based on the variable range of the pulse frequency in the working condition data, a plurality of pulse frequencies are extracted by adopting a Monte Carlo method, and a working condition data set comprising a plurality of groups of periodic pulse working conditions with different frequencies is formed by the pulse frequencies and fixed parameters (such as amplitude values and the like) in the working condition data. It should be noted that the monte carlo method is adopted to ensure the uniformity of sample distribution when the number of samples is small, and in the specific application process, the embodiment does not limit the use of the monte carlo method for sampling, and can also adopt direct random sampling and other sampling methods;
and secondly, performing backward optimization on the power distribution of the hybrid energy storage system under the working condition of each frequency in the working condition data set by adopting a dynamic programming method for each capacity allocation scheme to obtain an optimal power distribution curve corresponding to each capacity scheme under the periodic pulse working condition of a plurality of frequencies. It should be noted that in this embodiment, it is not limited to perform power distribution optimization only by using a dynamic programming method (DP), and optimization methods such as random dynamic programming (SDP), pointryagin minimum value principle (PMP), particle Swarm Optimization (PSO), genetic Algorithm (GA), etc. may also be used;
the third step is to process data, fit the power distribution curve obtained previously, and extract the fitting function parameter as the numerical representation of the energy management rule, and the concrete implementation process is as follows:
firstly, drawing an optimal power distribution curve (a function of total required power and output power of a super capacitor);
secondly, segmenting the drawn curve, dividing the curve into three segments by taking the inflection point of the curve as a segmentation point, and recording x-axis coordinates x1 and x2 of the segmentation point;
then, performing linear fitting on each section of the three-section curve, and recording the slopes k1, k2 and k3 and the intercepts b1, b2 and b3 of the three-section fitting function;
finally, x1, x2, k1, k2, k3, b1, b2, b3 are numerical representations of the energy management rules.
And finally, arranging all the data obtained in the three steps into a training data set required by training, wherein each piece of data comprises the models, the number, the serial-parallel connection mode, the pulse amplitude, the pulse frequency, the fitting function parameters of an optimal power distribution curve and the like of the energy type device and the power type device.
After the training data set is obtained, on the basis of balancing precision, cost and data characteristics and richness, a proper neural network model can be selected in the last step, off-line neural network training is carried out, a trained neural network is finally obtained, and mapping from a capacity configuration scheme to optimal energy management rules respectively corresponding to different pulse frequencies is achieved.
In this embodiment, specific operation steps of the offline joint optimization based on the evolutionary algorithm are shown in fig. 3. The key point of the joint optimization is that the influence of a capacity configuration scheme and an energy management rule on the final performance of the system is considered simultaneously in the optimization process so as to obtain a global optimal solution of a highly-coupled capacity configuration optimization problem and an energy management optimization problem; because the neural network model which is trained off line in the prior art can quickly output the corresponding optimal energy management rule after inputting the capacity configuration scheme and the pulse frequency, the objective function value is calculated and iterative optimization is carried out based on the optimal rule generated by the neural network in the evolutionary algorithm framework, and the global optimal solution of the joint optimization problem can be obtained. It should be noted that the introduction of the specific optimization method, namely the evolutionary algorithm, to perform the joint optimization step is to clarify elements involved in the optimization process, but the embodiment does not limit that only the Evolutionary Algorithm (EA) framework is used for performing the joint optimization, and various intelligent optimization or traditional optimization methods such as Particle Swarm Optimization (PSO), simulated Annealing (SA), genetic Algorithm (GA), pointryagin minimum value principle (PMP), and the like may also be used.
The process of performing the offline combined optimization of the hybrid energy storage system under the evolutionary algorithm framework specifically comprises the following steps:
firstly, inputting basic data required by optimization of energy type devices and power type devices, load working condition data and the like into an evolutionary algorithm to initialize algorithm parameters, and completing constraint expression of power, voltage and current constraints of a hybrid energy storage system, maximum serial-parallel numbers of batteries and capacitor monomers and the like;
secondly, extracting feasible solutions in a feasible domain defined by the constraints to form an initial population, namely a set of alternative capacity allocation schemes;
then, simulation calculation of an objective function value of each individual (namely, a capacity allocation scheme) in the initial population is performed based on the optimal energy management rule under different pulse frequencies output by the neural network, wherein the objective function value refers to an optimized target value of the hybrid energy storage system in the aspects of cost, service life, weight and the like through offline joint optimization, and is obtained through model simulation under the optimal energy management rule.
After the objective function value is obtained, the previous generation population is subjected to non-dominated sorting according to the quality of the objective function value, and operations such as cross variation and the like are performed to generate a next generation population.
Finally, iteration is carried out in a loop in an evolutionary algorithm frame to a preset termination algebra, a pareto frontier formed by the capacity allocation schemes obtained by the last iteration is output, and all the capacity allocation schemes on the pareto frontier have no good or bad difference and can be considered to be optimal, so that a user using the optimization method of the embodiment can select the allocation schemes used for actual hardware integration according to respective preference after obtaining all the schemes on the pareto frontier.
It should be noted that the intelligent algorithms such as the evolutionary algorithm balance the calculation cost and the optimality of the solution, and cannot guarantee to stably find the optimal solution, and if the calculation cost is not cared and the optimality of the optimization result is emphasized, the joint optimization can be performed by using the aforementioned traditional optimization framework such as the pointryagin minimum value principle (PMP).
In this embodiment, the online energy management step of the hybrid energy storage system based on the neural network is shown in fig. 4, and a globally optimal capacity configuration and energy management scheme can be obtained through the above joint optimization, so that device integration can be directly performed based on the optimal capacity configuration scheme to obtain the hardware of the hybrid energy storage system. Because the pulse discharge frequency during the load work is adjustable, each frequency corresponds to an optimal energy management rule, and in order to obtain the optimality on energy management, an Energy Management System (EMS) of the hybrid energy storage system can obtain the optimal energy management rule corresponding to the current capacity configuration scheme and the working frequency in real time through a trained neural network model.
If the stability and the response speed of the EMS work are emphasized, the optimal energy management rule under the possible working frequency can be calculated and stored in advance by using the neural network finished by offline training, and then the EMS obtains the optimal energy management rule corresponding to the current pulse discharge frequency through a table look-up method when the hybrid energy storage system runs.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention, and all modifications and equivalents of the present invention, which are made by the contents of the present specification and the accompanying drawings, or directly/indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (6)
1. A hybrid energy storage system joint optimization method for frequency-adjustable pulse working conditions is characterized by comprising the following steps:
generating a plurality of capacity allocation schemes in an application constraint range of the hybrid energy storage system based on the alternative models of the energy type devices and the power type devices and device parameters, and determining the optimal energy management rule corresponding to each capacity allocation scheme under the periodic pulse working condition of a specific frequency to form a training data set;
training a neural network model based on a training data set, so that the neural network model can output a corresponding optimal energy management rule after inputting a capacity configuration scheme and a pulse frequency;
aiming at the high-power pulse working condition with adjustable frequency, optimizing based on the mapping relation of capacity allocation-energy management rules established by a neural network model under a combined optimization framework, and outputting a global optimal capacity allocation scheme and energy management rules under a given working condition;
in practical application, aiming at a high-power pulse working condition with a specific frequency, an optimal energy management rule is obtained in real time based on a neural network model, and the energy management rule is updated in real time when the pulse frequency is adjusted, so that the optimal performance of the hybrid energy storage system is ensured.
2. The hybrid energy storage system joint optimization method for the frequency-tunable pulse working condition according to claim 1, wherein the construction process of the training data set specifically comprises:
based on the alternative models of the energy type device and the power type device and device parameters provided by a supply manufacturer, generating a plurality of capacity allocation schemes by adopting a sampling method within the voltage grade constraint, the power performance constraint and the volume and mass constraint ranges of the hybrid energy storage system to obtain a capacity allocation scheme set;
based on the variable range of the pulse frequency in the working condition data, a plurality of pulse frequencies are extracted by adopting a sampling method, and a working condition data set comprising a plurality of groups of periodic pulse working conditions with different frequencies is formed by the sampling method and fixed parameters in the working condition data;
and for each capacity allocation scheme, optimizing the power allocation of the hybrid energy storage system under the working condition of each frequency in the working condition data set to obtain an optimal power allocation curve corresponding to each capacity allocation scheme under the periodic pulse working condition of a plurality of frequencies, fitting the optimal power allocation curve, and extracting fitting function parameters as numerical representation of the energy management rule.
3. The hybrid energy storage system joint optimization method for the frequency tunable pulse conditions as claimed in claim 2, wherein the sampling method is a monte carlo method or a random sampling method.
4. The hybrid energy storage system joint optimization method for the frequency-adjustable pulse working condition according to claim 2, characterized in that a dynamic programming method, a Pontryagin minimum theorem method, a particle swarm optimization or a genetic algorithm is adopted to optimize the power distribution of the hybrid energy storage system under the working condition of each frequency in the working condition data set.
5. The hybrid energy storage system joint optimization method for the frequency-tunable pulse working condition according to any one of claims 1 to 4, wherein the joint optimization framework is composed of a neural network model and an optimization algorithm.
6. The hybrid energy storage system joint optimization method for the frequency-adjustable pulse working condition according to claim 5, wherein the optimization algorithm is an evolutionary algorithm, and optimization is performed based on a mapping relation between a capacity configuration and an energy management rule established by a neural network model under a joint optimization framework, specifically:
inputting basic data required for optimization including related data of the alternative devices and load working conditions into an evolutionary algorithm, completing power and voltage current constraint of the hybrid energy storage system and constraint expression of the maximum serial-parallel number of the battery and the capacitor monomers, and then extracting feasible solutions in a feasible domain defined by the constraints to form an initial population, namely a set of alternative capacity configuration schemes;
the method comprises the steps of performing simulation calculation on an objective function value of each individual in an initial population based on optimal energy management rules under different pulse frequencies output by a neural network model, performing non-dominated sorting or cross variation operation on a previous generation population according to the quality of the objective function value to generate a next generation population, continuously and circularly iterating to a preset termination algebra in an evolutionary algorithm frame, and outputting a pareto frontier formed by a capacity configuration scheme obtained by the last iteration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211629642.9A CN115618753B (en) | 2022-12-19 | 2022-12-19 | Hybrid energy storage system joint optimization method for frequency-adjustable pulse working condition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211629642.9A CN115618753B (en) | 2022-12-19 | 2022-12-19 | Hybrid energy storage system joint optimization method for frequency-adjustable pulse working condition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115618753A true CN115618753A (en) | 2023-01-17 |
| CN115618753B CN115618753B (en) | 2023-04-07 |
Family
ID=84880867
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211629642.9A Active CN115618753B (en) | 2022-12-19 | 2022-12-19 | Hybrid energy storage system joint optimization method for frequency-adjustable pulse working condition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115618753B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116436059A (en) * | 2023-02-07 | 2023-07-14 | 深圳先进储能材料国家工程研究中心有限公司 | Hybrid energy storage system and scheduling method thereof |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101882813A (en) * | 2010-06-02 | 2010-11-10 | 中国科学院电工研究所 | Mixed energy storage system |
| CN103366314A (en) * | 2013-07-18 | 2013-10-23 | 山东大学 | Wind power plant composite energy storage capacity planning method based on goal decomposition and complementation and leveling of goal decomposition |
| US20140368041A1 (en) * | 2013-06-17 | 2014-12-18 | Mcmaster Universtiy | Reconfigurable hybrid energy storage system for electrified vehicles |
| CN105680474A (en) * | 2016-02-22 | 2016-06-15 | 中国电力科学研究院 | Control method for restraining rapid power change of photovoltaic station based on energy storage system |
| WO2022221891A1 (en) * | 2021-04-20 | 2022-10-27 | Aviloo Gmbh | Ascertaining the state of health of a vehicle battery |
-
2022
- 2022-12-19 CN CN202211629642.9A patent/CN115618753B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101882813A (en) * | 2010-06-02 | 2010-11-10 | 中国科学院电工研究所 | Mixed energy storage system |
| US20140368041A1 (en) * | 2013-06-17 | 2014-12-18 | Mcmaster Universtiy | Reconfigurable hybrid energy storage system for electrified vehicles |
| CN103366314A (en) * | 2013-07-18 | 2013-10-23 | 山东大学 | Wind power plant composite energy storage capacity planning method based on goal decomposition and complementation and leveling of goal decomposition |
| CN105680474A (en) * | 2016-02-22 | 2016-06-15 | 中国电力科学研究院 | Control method for restraining rapid power change of photovoltaic station based on energy storage system |
| WO2022221891A1 (en) * | 2021-04-20 | 2022-10-27 | Aviloo Gmbh | Ascertaining the state of health of a vehicle battery |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116436059A (en) * | 2023-02-07 | 2023-07-14 | 深圳先进储能材料国家工程研究中心有限公司 | Hybrid energy storage system and scheduling method thereof |
| CN116436059B (en) * | 2023-02-07 | 2023-12-12 | 深圳先进储能材料国家工程研究中心有限公司 | Hybrid energy storage system and scheduling method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115618753B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104794541B (en) | A kind of power distribution network running optimizatin method based on simulated annealing and cone optimization | |
| CN108462184B (en) | Power system line series compensation optimization configuration method | |
| CN107067344B (en) | Distributed power supply multi-target access planning method considering power distribution network harmonic management | |
| CN105449675A (en) | Power network reconfiguration method for optimizing distributed energy access point and access proportion | |
| CN109839599B (en) | Lithium ion battery SOC estimation method based on second-order EKF algorithm | |
| CN104104081B (en) | A kind of uncertain tidal current analysis method of non-iterative based on optimization method | |
| CN115618753B (en) | Hybrid energy storage system joint optimization method for frequency-adjustable pulse working condition | |
| CN104578121A (en) | Method and system for distributing power of hybrid energy storage systems | |
| CN113300380B (en) | Load curve segmentation-based power distribution network reactive power optimization compensation method | |
| CN105938578A (en) | Large-scale photovoltaic power station equivalent modeling method based on clustering analysis | |
| CN113541146B (en) | Power flow calculation optimization method of power system considering distributed power supply | |
| CN109066728B (en) | Online damping coordination control method for multiple interval oscillation modes of extra-high voltage power grid | |
| CN109390953A (en) | Low-voltage network reactive voltage control method for coordinating and system containing distributed generation resource and electric car | |
| CN110474353A (en) | Layer-stepping energy-storage system and its power grid frequency modulation control method for coordinating of participation | |
| CN109962485B (en) | Source network charge-friendly interaction-oriented composite energy storage device site selection and volume fixing method | |
| CN108830451A (en) | A kind of the convergence potential evaluation method and system of user side distributed energy storage | |
| CN106451504A (en) | Control method and device for configuration cost of hybrid energy storage system | |
| Beus et al. | A model predictive control approach to operation optimization of an ultracapacitor bank for frequency control | |
| CN108429271A (en) | The micro- energy optimization method of mixed energy storage system and system adjusted for frequency | |
| CN104993503A (en) | Island microgrid frequency control method | |
| Barukčić et al. | The evolutionary optimization approach for voltage profile estimation in a radial distribution network with a decreased number of measurements | |
| Zhao et al. | Reactive power optimization considering dynamic reactive power reserves | |
| CN116316589A (en) | Distribution network distributed photovoltaic bearing capacity assessment method considering source load uncertainty | |
| CN114172149B (en) | Distributed generation cluster autonomous control method and system | |
| CN112787331B (en) | Deep reinforcement learning-based automatic power flow convergence adjusting method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |