Energy storage configuration optimization method and system for independent micro-grid

Abstract

The application discloses an energy storage configuration optimization method of an independent micro-grid, comprising determining independent variables describing weather; obtaining a cumulative density function of the weather by using kernel density estimation based on historical data of a target independent micro-grid; performing Latin hypercube sampling on the cumulative density function; performing inverse transformation on the cumulative density function to obtain an inverse cumulative density function; mapping the inverse cumulative density function to generate corresponding weather scenarios; performing reduction on the obtained weather scenarios to obtain a weather scenario set; and performing energy storage configuration optimization of the target independent micro-grid by using a particle swarm optimization algorithm according to the weather scenario set. The application further discloses a system for implementing the energy storage configuration optimization method of the independent micro-grid. The application can not only complete energy storage configuration optimization of the independent micro-grid, but also consider the influence of extreme weather scenarios, so that the reliability is higher and the accuracy is better.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of electrical automation, and particularly relates to an energy storage configuration optimization method and system for an independent micro-grid. BACKGROUND

[0002] With the development of economy and technology and the improvement of people's living standards, electric energy has become an essential secondary energy in people's production and life, bringing endless convenience to people's production and life. Therefore, guaranteeing the stable and reliable supply of electric energy has become one of the most important tasks of the power system.

[0003] At present, environmental problems are becoming more and more serious, and extreme weather events are occurring frequently. The stability and safety of the power system, especially the independent micro-grid system, are facing unprecedented challenges. Energy storage technology, as an effective energy regulation means, has been widely used in power systems and independent micro-grid systems, and has brought great guarantee for the safe and reliable operation of power systems and independent micro-grid systems.

[0004] In the case of extreme weather, the steady-state operating boundary of the independent micro-grid system is broken. In order to ensure the safe and stable operation of the independent micro-grid system in the case of extreme weather, it is necessary to optimize the energy storage configuration of the independent micro-grid. However, the existing energy storage configuration optimization scheme of the independent micro-grid is generally based on conventional power demand fluctuations and equipment failures, and does not consider the adverse effects of extreme weather. Therefore, the existing optimization scheme is no longer suitable for the current operation requirements of the independent micro-grid. SUMMARY

[0005] One of the purposes of the present application is to provide an energy storage configuration optimization method for an independent micro-grid with high reliability and good accuracy.

[0006] The second purpose of the present application is to provide a system for implementing the energy storage configuration optimization method of the independent micro-grid.

[0007] The energy storage configuration optimization method for an independent micro-grid provided by the present application comprises the following steps:

[0008] S1. Determine an independent variable describing the weather;

[0009] S2. Based on the historical data of the target independent micro-grid, the cumulative density function of the weather is obtained by using kernel density estimation according to the independent variable determined in step S1;

[0010] S3. Latin hypercube sampling is performed on the cumulative density function obtained in step S2;

[0011] S4. The inverse cumulative density function is obtained by performing inverse transformation on the cumulative density function obtained in step S3;

[0012] S5. mapping the inverse cumulative density function obtained in step S4 to generate a corresponding weather scenario;

[0013] S6. pruning the weather scenario obtained in step S5 to obtain a set of weather scenarios;

[0014] S7. using a particle swarm optimization algorithm to optimize the energy storage configuration of the target independent micro-grid according to the set of weather scenarios obtained in step S6.

[0015] The determination of the independent variable describing the weather in step S1 specifically includes the following steps:

[0016] The independent variable describing the weather includes temperature, humidity, wind speed, precipitation, and radiation intensity.

[0017] The step S2 of obtaining the cumulative density function of the weather based on the historical data of the target independent micro-grid according to the independent variable determined in step S1 specifically includes the following steps:

[0018] According to the independent variable determined in step S1, the corresponding historical data is obtained.

[0019] According to the obtained historical data, the following formula is used to estimate the corresponding probability density function:

[0020]

[0021] where f(t) is the probability density estimate at t; n is the total number of samples; h is the bandwidth; K() is the kernel function; T i is the i-th sample data.

[0022] The probability density function is normalized to obtain the cumulative density function of the independent variable of the weather where f(t) is the probability density estimate at t, and Δt is the sampling interval.

[0023] The step S3 of performing Latin hypercube sampling on the cumulative density function obtained in step S2 specifically includes the following steps:

[0024] Determine the number of factors and the number of samples for sampling;

[0025] Divide each cumulative density function into several equal-probability intervals;

[0026] Randomly select a sample point from each interval;

[0027] Arrange and combine each number of samples to ensure that each number of samples covers all intervals;

[0028] Shuffle all the samples to increase the randomness of the samples;

[0029] Finally, the Latin hypercube sampling is completed.

[0030] The inverse cumulative density function is obtained by inversely transforming the cumulative density function obtained in step S3, and specifically includes the following steps:

[0031] The cumulative density function is transformed to map the uniformly distributed random variable to the target probability distribution, so as to obtain the random sample conforming to the target probability distribution;

[0032] A sample value corresponding to the uniform random number is obtained, denoted as Where ICDF is the inverse cumulative density function, is the inverse function of the cumulative density function, and u is the cumulative probability density value.

[0033] The inverse cumulative density function obtained in step S4 is mapped to generate a corresponding weather scene, and specifically includes the following steps:

[0034] A random number is obtained, and the obtained random number is mapped to a specific scene according to the inverse cumulative density function obtained in step S4, so as to obtain the corresponding weather scene.

[0035] The weather scene obtained in step S5 is reduced to obtain a weather scene set, and specifically includes the following steps:

[0036] For each weather scene s i , the minimum distance D i,min between s i and other scenes s j is calculated using the following formula:

[0037] D i,min =minρ i d(s i ,s j )

[0038] In the formula, ρ i is the occurrence probability of the weather scene s i ; d(s i ,s j ) is the distance between s i and s j ;

[0039] The Latin hypercube sampling is performed on the probability density function to randomly generate scenes; within the Latin hypercube sampling interval, the scene with the lowest probability is deleted;

[0040] The probability of the deleted scene is increased to the nearest scene.

[0041] The above steps are repeated until the number of scenes in the 0-10% sampling interval or the 90-100% sampling interval is 1;

[0042] Finally, a weather scene set is obtained;

[0043] The scene probability in the 0-10% sampling interval or the 90-100% sampling interval is taken as a risk factor.

[0044] The weather scene set obtained according to step S6 is subjected to energy storage configuration optimization of the target independent micro-grid by using a particle swarm optimization algorithm, and the specific steps include the following steps:

[0045] The weather scene set obtained according to step S6 and the obtained risk factor are subjected to energy storage configuration optimization of the target independent micro-grid by using a particle swarm optimization algorithm;

[0046] In the process of optimization by using the particle swarm optimization algorithm, the following formula is used to calculate the local learning factor c1 and the global learning factor c2:

[0047]

[0048] In the formula, c1max is a local learning factor termination value; c1min is a local learning factor starting value; c2max is a global learning factor termination value; c2min is a global learning factor starting value; v is an intermediate variable and v = 0; N is a total number of samples, x is a current local optimal solution, x is a current global optimal solution.

[0049] ​This invention also provides a system for implementing the energy storage configuration optimization method of the aforementioned independent microgrid, comprising a variable determination module, a function determination module, a sampling module, an inverse transformation module, a mapping module, a reduction module, and a configuration optimization module; the variable determination module, function determination module, sampling module, inverse transformation module, mapping module, reduction module, and configuration optimization module are connected in series; the variable determination module is used to determine independent variables describing the weather and upload the data information to the function determination module; the function determination module is used to obtain the cumulative density function of the weather based on the received data information, historical data of the target independent microgrid, and the independent variables determined in step S1 using kernel density estimation, and upload the data information to the sampling module; the sampling module is used to obtain the cumulative density function of the weather based on the received data information. The received cumulative density function is subjected to Latin hypercube sampling, and the data is uploaded to the inverse transformation module. The inverse transformation module performs an inverse transformation on the received cumulative density function to obtain the inverse cumulative density function, and uploads the data to the mapping module. The mapping module maps the obtained inverse cumulative density function to generate corresponding weather scenarios based on the received data, and uploads the data to the reduction module. The reduction module reduces the obtained weather scenarios based on the received data to obtain a set of weather scenarios, and uploads the data to the configuration optimization module. The configuration optimization module optimizes the energy storage configuration of the target independent microgrid using a particle swarm optimization algorithm based on the received data and the obtained set of weather scenarios.

[0050] The energy storage configuration optimization method and system for independent microgrids provided by this invention generates extreme weather scenarios through historical meteorological data and Latin hypercube sampling, and models the extreme weather scenarios by combining kernel density estimation and inverse cumulative density function, and finally optimizes the energy storage configuration using particle swarm optimization algorithm; therefore, this invention can not only complete the energy storage configuration optimization of independent microgrids, but also take into account the impact of extreme weather scenarios, and has higher reliability and better accuracy. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0052] Figure 2 This is a schematic diagram of the functional modules of the system of the present invention. Detailed Implementation

[0053] like Figure 1 The diagram shown is a flowchart of the method of the present invention: The energy storage configuration optimization method for independent microgrids disclosed in this invention includes the following steps:

[0054] S1. Identify the independent variables describing the weather; this includes the following steps:

[0055] determining independent variables describing the weather, including temperature, humidity, wind speed, precipitation, and radiation intensity;

[0056] S2. Based on the historical data of the target independent micro-grid, the cumulative density function of the weather is obtained according to the independent variables determined in step S1 by using kernel density estimation; specifically including the following steps:

[0057] According to the independent variables determined in step S1, the corresponding historical data is obtained;

[0058] According to the obtained historical data, the corresponding probability density function is estimated by using the following formula:

[0059]

[0060] In the formula, f(t) is the probability density estimation at t; n is the total number of samples; h is the bandwidth; K() is the kernel function; T i is the i th sample data;

[0061] The probability density function is normalized to obtain the cumulative density function of the independent variable of the weather Where f(t) is the probability density estimation at t, and Δt is the sampling interval;

[0062] S3. Latin hypercube sampling is performed on the cumulative density function obtained in step S2; specifically including the following steps:

[0063] Determine the number of factors (the number of factors is the number of dimensions) and the number of samples for sampling;

[0064] Divide each cumulative density function into several equal probability intervals;

[0065] Randomly select a sample point from each interval;

[0066] Arrange and combine each factor number of samples to ensure that each factor number of samples covers all intervals;

[0067] Shuffle the order of all samples to increase the randomness of the samples;

[0068] Finally, the Latin hypercube sampling is completed;

[0069] S4. Inverse transformation is performed on the cumulative density function obtained in step S3 to obtain the inverse cumulative density function; specifically including the following steps:

[0070] Transform the cumulative density function to map the uniformly distributed random variable to the target probability distribution, thereby obtaining a random sample conforming to the target probability distribution;

[0071] Obtain the sample value corresponding to the uniform random number, denoted as Wherein ICDF is the inverse cumulative density function, is the inverse function of the cumulative density function, and u is the cumulative probability density value;

[0072] S5. The inverse cumulative density function obtained in step S4 is mapped to generate a corresponding weather scene; specifically including the following steps:

[0073] A random number is obtained, and the obtained random number is mapped to a specific scene according to the inverse cumulative density function obtained in step S4, so as to obtain a corresponding weather scene;

[0074] S6. The weather scene obtained in step S5 is pruned to obtain a weather scene set; specifically including the following steps:

[0075] For each weather scene s i , the minimum distance D i between s j and other scenes s i,min is calculated using the following formula:

[0076] D i,min = minρ i d(s i ,s j )

[0077] Wherein ρ i is the occurrence probability of the weather scene s i ; and d(s i ,s j ) is the distance between s i and s j ;

[0078] For the probability density function, Latin hypercube sampling is performed to randomly generate scenes; within the Latin hypercube sampling interval, the scene with the lowest probability is deleted; in specific implementation, the probability density function describes the probability of occurrence of different scenes, and when these scenes are applied to reality, the scenes need to be generated based on the probability density function, so sampling is needed to randomly generate scenes to ensure the objectivity and universality of the generated scenes; common sampling methods include Monte Carlo sampling, Latin hypercube sampling, etc.; since the probability of occurrence of extreme scenes is low, Monte Carlo sampling may cause the problem of no scene occurrence in the low-probability region or too large sample space leading to dramatic increase in calculation amount, Latin hypercube sampling uniformly divides the cumulative density function into layers to ensure the occurrence of scenes in the low-probability region, and does not require too many sample quantities, so Latin hypercube sampling is selected; "within the Latin hypercube sampling interval" means within the layer interval; the probability is the probability of occurrence of the scene; in order to make the probability distance between the probability distribution of the retained scene and the probability of the initial scene set shortest, the probability of the scene with small probability is added to the scene with the closest probability distance D i,min, for calculating such a probability distance;

[0079] The probability of the scene to be deleted is increased to the nearest scene;

[0080] The above steps are repeated until the number of scenes in the 0-10% sampling interval or the 90-100% sampling interval is 1;

[0081] Finally, a set of weather scenes is obtained;

[0082] The probability of the scene in the 0-10% sampling interval or the 90-100% sampling interval is taken as a risk factor;

[0083] S7. According to the set of weather scenes obtained in step S6, a particle swarm optimization algorithm is used to optimize the energy storage configuration of the target independent microgrid; specifically including the following steps:

[0084] According to the set of weather scenes obtained in step S6 and the obtained risk factor, a particle swarm optimization algorithm is used to optimize the energy storage configuration of the target independent microgrid;

[0085] In the process of optimization using the particle swarm optimization algorithm, the following formula is used to calculate the local learning factor c1 and the global learning factor c2:

[0086]

[0087] In the formula, c1max is the terminal value of the local learning factor; c1min is the starting value of the local learning factor; c2max is the terminal value of the global learning factor; c2min is the starting value of the global learning factor; is an intermediate variable and N is the total number of samples, is the current local optimal solution, is the current global optimal solution.

[0088] The application integrates extreme weather risk factors into the optimization process of the energy storage system, generates extreme weather scenarios by analyzing historical meteorological data using the Latin hypercube sampling method, and models weather risks by combining kernel density estimation and inverse cumulative density function, so that the configuration of the energy storage system can more comprehensively consider the influence of extreme weather on system operation, and the system's resilience and reliability are improved; the application generates extreme weather risk factors by using the Latin hypercube sampling method, breaks through the limitations of random sampling in traditional methods, can more uniformly and efficiently cover the entire risk space, avoids the problem of extreme weather sample distribution biasing the small probability interval, optimizes the representativeness and rationality of extreme weather scenarios by combining scenario reduction and shortest distance calculation strategy, improves the accuracy and reliability of scenario sampling, and provides more scientific risk input for subsequent energy storage system optimization; the application dynamically adjusts the learning factor of particles in the particle swarm optimization algorithm, realizes fast exploration in the early stage and accurate convergence in the later stage, improves the attractiveness of the global optimal solution, ensures that the particle swarm can accurately find the global optimal solution, avoids missing the best configuration due to premature convergence, and effectively deals with the problems of constraint overrun and multi-objective optimization in the process of independent micro-grid energy storage configuration optimization under the influence of extreme weather risk factors.

[0089] As Figure 2 shown in the functional module schematic diagram of the system of the application: the system for implementing the energy storage configuration optimization method of the independent micro-grid disclosed by the application comprises a variable determination module, a function determination module, a sampling module, an inverse transformation module, a mapping module, a reduction module and a configuration optimization module; the variable determination module, the function determination module, the sampling module, the inverse transformation module, the mapping module, the reduction module and the configuration optimization module are connected in series; the variable determination module is used to determine independent variables describing weather, and upload data information to the function determination module; the function determination module is used to obtain the cumulative density function of weather by using kernel density estimation based on the independent variables determined in step S1 according to the received data information based on the historical data of the target independent micro-grid, and upload data information to the sampling module; the sampling module is used to perform Latin hypercube sampling on the obtained cumulative density function according to the received data information, and upload data information to the inverse transformation module; the inverse transformation module is used to perform inverse transformation on the obtained cumulative density function according to the received data information, to obtain an inverse cumulative density function, and upload data information to the mapping module; the mapping module is used to map the obtained inverse cumulative density function to generate corresponding weather scenarios according to the received data information, and upload data information to the reduction module; the reduction module is used to reduce the obtained weather scenarios to obtain a weather scenario set according to the received data information, and upload data information to the configuration optimization module; the configuration optimization module is used to perform energy storage configuration optimization of the target independent micro-grid by using a particle swarm optimization algorithm according to the obtained weather scenario set according to the received data information.

Claims

1. A method for optimizing energy storage configuration of an independent microgrid, comprising the following steps: S1. determining independent variables describing weather; S2. based on historical data of a target independent microgrid, obtaining a cumulative density function of weather according to the independent variables determined in step S1 by using kernel density estimation; S3. performing Latin hypercube sampling on the cumulative density function obtained in step S2; S4. performing inverse transformation on the cumulative density function obtained in step S3 to obtain an inverse cumulative density function; S5. mapping the inverse cumulative density function obtained in step S4 to generate corresponding weather scenarios; specifically comprising the following steps: obtaining a random number, and mapping the obtained random number to a specific scenario according to the inverse cumulative density function obtained in step S4 to obtain a corresponding weather scenario; S6. reducing the weather scenarios obtained in step S5 to obtain a set of weather scenarios; specifically comprising the following steps: For each weather scenario s i , the minimum distance D i,min to other scenarios s j is calculated using the following formula: i ​ D i,min = min p i d(s i ,s j ) In the formula ρ i For weather scenes i The probability of occurrence of d(s) i ,s j ) for s i With s j The distance; performing Latin hypercube sampling on the probability density function to randomly generate scenarios; deleting the scenario with the lowest probability within the Latin hypercube sampling interval; increasing the probability of the deleted scenario to the nearest scenario; repeating the above steps until the number of scenarios within 0-10% sampling interval or 90-100% sampling interval is 1; finally, obtaining a set of weather scenarios; taking the scenario probability within 0-10% sampling interval or 90-100% sampling interval as a risk factor; S7. using a particle swarm optimization algorithm to optimize the energy storage configuration of the target independent microgrid according to the set of weather scenarios obtained in step S6.

2. The method of claim 1, wherein The determination of the independent variables describing weather in step S1 specifically comprises the following steps: determining independent variables describing weather, including temperature, humidity, wind speed, precipitation, and radiation intensity.

3. The method of claim 2, wherein The step S2 of obtaining a cumulative density function of weather according to the independent variables determined in step S1 by using kernel density estimation based on historical data of a target independent microgrid specifically comprises the following steps: obtaining corresponding historical data according to the independent variables determined in step S1; estimating the corresponding probability density function using the following formula based on the obtained historical data: where f(t) is the probability density estimate at t; n is the total number of samples; h is the bandwidth; K() is the kernel function; T i is the ith sample data; The probability density function is normalized to obtain the cumulative density function of the independent variable of the weather as where f(t) is the probability density estimate at t, and Δt is the sampling interval.

4. The method of claim 3, wherein The step S3 of performing Latin hypercube sampling on the cumulative density function obtained in step S2 specifically comprises the following steps: determining the number of factors and the number of samples for sampling; dividing each cumulative density function into several equal-probability intervals; randomly selecting a sample point from each interval; performing permutation and combination on each factor number of samples to ensure that each factor number of samples covers all intervals; shuffling the order of all samples to increase the randomness of the samples; finally, completing the Latin hypercube sampling.

5. The method of claim 4, wherein The step S4 of performing inverse transformation on the cumulative density function obtained in step S3 to obtain an inverse cumulative density function specifically comprises the following steps: performing transformation on the cumulative density function to map uniformly distributed random variables to the target probability distribution, thereby obtaining random samples conforming to the target probability distribution; A sample value corresponding to the uniform random number is obtained, denoted as where ICDF is the inverse cumulative density function, is the inverse function of the cumulative density function, and u is the cumulative probability density value.

6. The method of claim 5, wherein The step S7 of using a particle swarm optimization algorithm to optimize the energy storage configuration of the target independent microgrid according to the set of weather scenarios obtained in step S6 specifically comprises the following steps: According to the weather scene set obtained from step S6 and the obtained risk factor, a particle swarm optimization algorithm is used to optimize the energy storage configuration of the target independent micro-grid; In the process of optimization using the particle swarm optimization algorithm, the following formula is used to calculate the local learning factor c1 and the overall learning factor c2: wherein is a local learning factor termination value; is a local learning factor start value; is a global learning factor termination value; is a global learning factor start value; is an intermediate variable and N is the total number of samples, is the current local optimum solution, is the current global optimum solution.

7. A system for implementing the energy storage configuration optimization method of any one of claims 1-6, characterized in that The system comprises a variable determination module, a function determination module, a sampling module, an inverse transformation module, a mapping module, a reduction module and a configuration optimization module; the variable determination module, the function determination module, the sampling module, the inverse transformation module, the mapping module, the reduction module and the configuration optimization module are connected in series; the variable determination module is used to determine the independent variable describing the weather, and upload the data information to the function determination module; the function determination module is used to obtain the cumulative density function of the weather based on the historical data of the target independent micro-grid according to the received data information, the independent variable determined in step S1 and the kernel density estimation, and upload the data information to the sampling module; The sampling module is used to perform Latin hypercube sampling on the obtained cumulative density function according to the received data information, and upload the data information to the inverse transformation module; the inverse transformation module is used to perform inverse transformation on the obtained cumulative density function according to the received data information, to obtain the inverse cumulative density function, and upload the data information to the mapping module; The mapping module is used to map the obtained inverse cumulative density function according to the received data information, to generate the corresponding weather scene, and upload the data information to the reduction module; The reduction module is used to reduce the obtained weather scene according to the received data information, to obtain a weather scene set, and upload the data information to the configuration optimization module; the configuration optimization module is used to optimize the energy storage configuration of the target independent micro-grid according to the obtained weather scene set and the particle swarm optimization algorithm according to the received data information.

Images