Optimized micro-grid scheduling method and system based on random response surface method scheduling strategy

Abstract

The invention relates to the technical field of micro-grid dispatching, in particular to an optimized micro-grid dispatching method and system based on a random response surface method dispatching strategy, which are used for modeling and analyzing the power generation uncertainty of renewable energy sources such as wind energy and solar energy in a micro-grid by introducing a random response surface method. Through historical data analysis, a random response surface model of microgrid operation is constructed, and influences of various factors on system operation are captured. And based on the model, a corresponding scheduling strategy is formulated, and dynamic optimization scheduling of the micro-grid is realized. By introducing intelligent optimization technologies such as a genetic algorithm, economy, environmental friendliness and system stability are comprehensively considered, an optimal scheduling scheme is sought, energy waste is reduced, and the resource utilization rate is improved. Compared with a traditional scheduling method, the method has high adaptability and accuracy, and can quickly make a response in the face of emergencies and load fluctuation to ensure safe and stable operation of the micro-grid.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid scheduling, and particularly to an optimized microgrid scheduling method and system based on a scheduling strategy of the stochastic response surface method. Background Art

[0002] Under the background of continuous growth in energy demand and high emphasis on environmental protection today, the microgrid, as a new type of energy supply system, has attracted much attention. A microgrid consists of distributed power sources, energy storage devices, loads, and an energy management system, and can achieve efficient utilization and autonomous control of energy.

[0003] Traditional microgrid scheduling methods usually adopt deterministic models. In this mode, the influence of many uncertain factors is often ignored. For example, the output power fluctuations of distributed power sources and the changes in load demands are often overlooked in traditional methods. This leads to the problem that the traditional scheduling method may produce inaccurate scheduling results in practical applications, and even has an adverse impact on the safe and stable operation of the microgrid. To address the deficiencies of traditional methods, stochastic optimization methods have been widely applied in microgrid scheduling in recent years.

[0004] Among them, the stochastic response surface method is a relatively effective stochastic optimization method. It approximates the objective function and constraints by constructing a response surface model, and then transforms the stochastic optimization problem into a deterministic optimization problem. This method has the advantages of high computational efficiency and good accuracy, and has been widely applied in the engineering field.

[0005] However, the current microgrid scheduling methods based on the stochastic response surface method are not perfect. On the one hand, the construction of the response surface model requires a large amount of sample data, which may be restricted in practical applications; on the other hand, the existing methods are often not flexible enough in dealing with multi-objective optimization problems and are difficult to meet the specific needs of different users.

[0006] Therefore, in order to overcome the defects of the existing technology, it is urgent to propose an optimized microgrid scheduling method based on a scheduling strategy of the stochastic response surface method. This method can effectively handle the uncertain factors in the microgrid, significantly improve the accuracy and reliability of the scheduling results. At the same time, it also has high computational efficiency and flexibility, and can better meet the needs of different users. Summary of the Invention

[0007] In view of the problems existing in the above-mentioned prior art, the present invention is proposed.

[0008] Therefore, the present invention provides an optimized microgrid scheduling method based on a random response surface method scheduling strategy, which mainly solves the key technical problems of microgrids in the context of random fluctuations in renewable energy supply and load demand. It addresses the volatility of renewable energy generation and the randomness of load demand in the microgrid, making the power output stable and facilitating accurate prediction and control. At the same time, on the premise of meeting the operating constraints of the microgrid, it realizes multi-objective optimization, including minimizing economic costs, maximizing energy efficiency, minimizing environmental impacts, etc., and coordinates the operation of various distributed power sources and loads. In addition, the present invention can also improve the reliability and stability of scheduling, respond to emergencies, reduce power fluctuations to improve power quality, and ensure stable power supply for the microgrid. It is used to enhance the operating economy, stability, and low-carbon performance of the microgrid under random fluctuations in renewable energy supply and load demand.

[0009] To solve the above technical problems, the present invention provides the following technical solutions. The optimized microgrid scheduling method based on a random response surface method scheduling strategy includes: standardizing random variables;

[0010] Constructing the coefficient matrix of the Hermite chaos polynomial;

[0011] Conducting source-load type classification and constructing a short-term prediction model;

[0012] Constructing a scheduling model;

[0013] The system constructs a multi-objective function.

[0014] As a preferred embodiment of the optimized microgrid scheduling method based on a random response surface method scheduling strategy of the present invention, wherein: the standardizing random variables use the Nataf transformation to convert the correlated variables of the random distribution function into independent standard normal distribution variables.

[0015] As a preferred embodiment of the optimized microgrid scheduling method based on a random response surface method scheduling strategy of the present invention, wherein: the constructing the coefficient matrix of the Hermite chaos polynomial includes that the mapping relationship between the random response output Y and the random response output obeying an independent standard normal random distribution can be constructed into a chaos polynomial according to the Hersemite orthogonal polynomial.

[0016] As a preferred embodiment of the optimized microgrid scheduling method based on a random response surface method scheduling strategy of the present invention, wherein: the conducting source-load type classification is classified according to the ability of flexible adjustment.

[0017] As a preferred solution of the optimized microgrid scheduling method based on the random response surface method scheduling strategy of the present invention, wherein: the construction of the short-term prediction model includes constructing a short-term prediction model of WP&L based on K-MEANS clustering and long short-term memory neural network. The WP&L historical data is used for the training and verification of the long short-term memory neural network model to obtain the Gaussian distribution function and parameters of the model prediction error.

[0018] As a preferred solution of the optimized microgrid scheduling method based on the random response surface method scheduling strategy of the present invention, wherein: the construction of the scheduling model includes the FAS microgrid scheduling model and the NAS microgrid scheduling model.

[0019] As a preferred solution of the optimized microgrid scheduling method based on the random response surface method scheduling strategy of the present invention, wherein: the system constructs a multi-objective function expressed as

[0020]

[0021] In the formula,

[0022] f1 represents the expected value of the microgrid operation cost, which is used to measure the average level of the microgrid operation cost during the entire scheduling period. Among them, is the mathematical expectation of the microgrid operation cost Y Cop(t) at time t, which reflects the average trend of the operation cost in this period. Since the expected value of the random response output of the microgrid operation cost is equal to the first element of the corresponding Hermite chaos polynomial coefficient matrix, so

[0023] f2 represents the variance of the microgrid operation cost, which is used to measure the fluctuation degree of the operation cost during the scheduling period. The smaller the variance, the more stable the operation cost. Among them, D[Y Cop(t) is the variance of the microgrid operation cost Y Cop(t) at time t, which reflects the degree of dispersion of the operation cost around the expected value. In the random response surface method, since the variance D[Y Cop(t) of the random response output of the microgrid operation cost is equal to the sum of the squares of the elements other than the first element in its chaos polynomial coefficient A Cop(t) matrix, and the standardized SRSM sample number N Cop(t) of the second-order three-dimensional SRSM model in this paper is 10, so when summing the variances of the operation costs of each period during the scheduling period, so a

[0024] f3 represents the expected value of the microgrid carbon emission cost, which is used to evaluate the average situation of the microgrid carbon emission cost during the entire scheduling period to reflect the low-carbon performance of the microgrid operation. Among them,​ is the carbon emission cost of the microgrid at time period t The mathematical expectation of, reflecting the average level of carbon emission cost during this period. Since the expected value of the random response output of the microgrid operating cost is equal to the first element of the corresponding Hermite chaos polynomial coefficient matrix, so

[0025] contains the Hermite chaos polynomial coefficients related to the operating cost (when K = OP) or carbon emission cost (when K = C) of the microgrid at time period t. H(ξ) is the Hermite coefficient matrix, determined by the standardized SRSM samples, which is used to construct the relationship between the random response and the standardized samples in the SRSM model. H(ξ) -1 is its inverse matrix, used to solve the chaos polynomial coefficients. is a cost-related variable. When K = OP, it represents the operating cost of the microgrid including random response during the time period; when K = C, it represents the carbon emission cost of the microgrid including random response during the time period.

[0026] As a preferred solution of the optimized microgrid scheduling system based on the random response surface method scheduling strategy described in the present invention, it includes: a variable standardization module, a coefficient matrix construction module, a source-load classification prediction module, a scheduling model construction module, and a multi-objective optimization module.

[0027] A computer device includes a memory and a processor. The memory stores a computer program. The characteristic is that when the processor executes the computer program, it implements the steps of any one of the methods in the optimized microgrid scheduling method based on the random response surface method scheduling strategy.

[0028] A computer-readable storage medium stores a computer program. The characteristic is that when the computer program is executed by the processor, it implements the steps of any one of the methods in the optimized microgrid scheduling method based on the random response surface method scheduling strategy.

[0029] Advantages of the present invention: The present invention can effectively address multiple challenges in the operation of microgrids, showing distinct advantages compared to existing methods. When dealing with uncertain factors, the present invention ingeniously handles the complex correlation of WP&L by means of Nataf transformation, constructs a stochastic response surface model to accurately analyze uncertainties, greatly improves the scheduling accuracy and reliability, and effectively stabilizes the foundation of power grid operation. In the process of multi-objective optimization, the present invention constructs a multi-objective function, comprehensively considers various factors, accurately formulates strategies by dividing NAS and FAS, realizes the coordinated progress of multi-objective optimization, and promotes the efficient operation and green upgrade of microgrids in all aspects. In terms of computational efficiency and flexibility, although sample data is required to construct the model, its computational efficiency is still prominent and its flexibility is extraordinary. According to different operation scenarios and requirements, various power source and load scheduling models can be flexibly embedded in the stochastic response model and parameters can be adjusted freely, fully ensuring the stable, efficient, economic and environmentally friendly operation of microgrids under complex and changeable working conditions, reducing both costs and emissions, and the adaptability and optimization effect far exceeding traditional methods, laying a solid path for the sustainable development of microgrids. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other accompanying drawings without creative efforts based on these drawings.

[0031] Figure 1 It is a flowchart of an optimized microgrid scheduling method based on a stochastic response surface method scheduling strategy provided by an embodiment of the present invention.

[0032] Figure 2 It is the stochastic response of comparing costs through Monte Carlo simulation for an experimental case of an optimized microgrid scheduling method based on a stochastic response surface method scheduling strategy provided by an embodiment of the present invention.

[0033] Figure 3 It is a comparison of the all-day cost PDF curves for an experimental case of an optimized microgrid scheduling method based on a stochastic response surface method scheduling strategy provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0035] Embodiment 1

[0036] Referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides an optimized microgrid scheduling method based on a stochastic response surface method scheduling strategy, as Figure 1 shown, including:

[0037] S1: Standardize the random variables.

[0038] It should be noted that the standardized random variables are obtained by transforming the relevant variables of the random distribution function into variables subject to an independent standard normal distribution through calculation.

[0039] Furthermore, in the embodiment of the present application, the calculation method uses the Nataf transformation to transform the relevant variables of the random distribution function into variables subject to an independent standard normal distribution.

[0040] It should be noted that SRSM is only applicable to random input variables subject to an independent standard normal distribution. However, the WP&L in the microgrid are correlated and subject to other forms of distribution functions. Therefore, the Nataf transformation is used to transform the relevant variables of the random distribution function into variables subject to an independent standard normal distribution.

[0041] Let the correlation coefficient matrix of the random variables be:

[0042]

[0043] where ρ ij is the correlation coefficient of the random variables.

[0044] Let U = [U1, U2,..., U n T be subject to the standard normal distribution, and its cumulative distribution function (CDF) can be expressed as φU(U i ). The mathematical expectation of U i is 1, and the variance is 0. The correlation matrix of U can be expressed as:

[0045]

[0046] where ρ′ ij is the correlation coefficient of the random variables U i and U j .

[0047] According to the principle of equiprobable transformation, the transformation between Xi and Ui can be taken as:

[0048] X i = F -1 (φ U (U i )) i = 1, 2,..., n

[0049] ρ​ij It can be expressed as:

[0050]

[0051] Obviously, C U is a symmetric matrix, and C U can be decomposed by Cholesky decomposition C U = LL T The relationship between U that follows the correlated standard normal distribution and ξ that follows the independent standard normal distribution can be expressed as ξ = L -1 U. Combining the above formula, the relationship between the random variable X and ξ can be constructed. This process of transforming X into ξ is called the Nataf forward transformation, and the Nataf inverse transformation can transform ξ into X through U = Lξ and the above formula. Therefore, ρ ij and ρ′ ij have the following relationship:

[0052]

[0053] S2: Construct the coefficient matrix of the Hermite chaos polynomial. Establishing a coefficient matrix of the Hermite chaos polynomial, the random response process of the microgrid adjustment can be described by the Hermite chaos polynomial.

[0054] It should be noted that the mapping relationship between the random response output Y and the random response output that follows the independent standard normal random distribution can be constructed into a chaos polynomial according to the Hersemite orthogonal polynomial.

[0055] It should be noted that considering the computational burden and accuracy of engineering applications, in the embodiments of the present invention, a second-order chaos polynomial is selected:

[0056]

[0057] In the formula are the undetermined coefficients of the Hermite chaos polynomial, and n is the dimension of the random variable, representing the number of random variables.

[0058] For the second-order chaos polynomial, the third-order Hermite orthogonal polynomial H3 = x 3 - 3x's roots 0, are the collocation points of the chaos polynomial ξ. When the dimension of the chaos polynomial is n, the corresponding number of collocation points is M = 3n. Therefore, the above formula can be rewritten in the following matrix form:

[0059] Y = HB

[0060]

[0061] H is the Hermite coefficient matrix.

[0062] The collocation point selection based on the principle of linear independence can significantly improve the calculation efficiency. Using Gaussian elimination to remove the linearly dependent rows of H(ξ), when the number of rows of H is M = Na and H is a full-rank matrix, the collocation point set appears. These collocation points are the standardized SRSM samples, taken as ξ = [ξ1, ξ2, …, ξ Na T , where ξ i is an n-dimensional vector. Then the random response function is

[0063]

[0064] where

[0065] S3: Considering various power sources and load types comprehensively, classify them according to whether they can be flexibly adjusted; specifically, to ensure the economic and low-carbon operation of the microgrid under the random fluctuations of wind power (WT), photovoltaic (PV) and load (L), classify them according to whether they can be flexibly adjusted, divided into non-adjustable source loads (NAS) and flexibly adjustable source loads (FAS). The former participates in the day-ahead scheduling, and the latter participates in the intra-day regulation, and its power value, cost and carbon emissions are incorporated into the random response output modeling.

[0066] It should be noted that in the embodiment of the present application, the source-load type classification adopts three methods: WT, PV and ESS;

[0067] Furthermore, in the grid-connected operation mode, the grid interaction power is usually considered as a balanced source. For island operation, MT and FC are used as the basic power sources, and a high-speed diesel engine (DE) is used as a flexible adjustable power source. The demand response proposed by the present invention includes unselectable loads (UL), shiftable loads (SL) and reducible loads (RL).

[0068] The flexibly adjustable sources or loads are divided into flexibly adjustable source loads (FAS), which participate in the intra-day regulation, that is, FAS = {GRID, DE, WT, PV, RL}. The power value, operating cost and carbon emissions of FAS are modeled as the output of the microgrid adjustment random response.

[0069] By dividing NAS and FAS, a day-ahead microgrid scheduling SO strategy based on SRSM is constructed. This strategy mainly consists of a microgrid scheduling model, a random response model based on SRSM, an extreme power shortage constraint and an SO multi-objective. In the microgrid scheduling SO model, the decision variables are the scheduling plans of NAS and FAS in each scheduling period, expressed as: P k (t), (k ∈ {NAS, FAS}, t ∈ T).

[0070] ​For the stochastic response model, for the stochastic response under the stochastic fluctuations of WP&L, only the adjusted power of FAS needs to be estimated. According to the standardized SRSM sample X i , (i ∈ {1, 2, …, N a}), the response output of the adjusted power value of FAS is expressed as where the variable symbol with a mark. represents the stochastic response output (or stochastic variable input), and the subscript i represents the i-th standardized SRSM sample. The scheduling scheme of FAS is selected as its expected value. The expected value of the stochastic response output of the adjusted power is directly equal to the first element of the Hermite chaos polynomial coefficient matrix:

[0071]

[0072] S4: Construct a short-term prediction model. WP&L standardized SRSM samples; specifically, by using K-MEANS clustering and long short-term memory (LSTM) neural network technology, construct a short-term prediction model of WP&L, and obtain stochastic variables through the prediction model. These stochastic variables will respond to the FAS output and operating cost of the microgrid scheduling stochastic response model.

[0073] It should be noted that constructing the short-term prediction model includes WP&L standardized SRSM samples. The present invention constructs a short-term prediction model of WP&L based on K-MEANS clustering and long short-term memory (LSTM) neural network. WP&L historical data is used for the training and verification of the LSTM model. Then, the Gaussian distribution function and parameters of the model prediction error are obtained. In the microgrid optimal scheduling, the stochastic variable is the uncertain prediction error of WP&L, that is, X i ={X PV,i , X WT,i , X RL,i}, where it should follow a third-order mixed Gaussian distribution, as follows:

[0074]

[0075] In the formula, x is the prediction error; μ, σ, and α represent the expected value, standard deviation, and weight of each group of Gaussian distributions.

[0076] Therefore, the present invention selects a second-order three-dimensional SRSM model. The number of standardized SRSM samples N a = 10. According to the configured linear independence principle, the standardized SRSM sample ξ i = [ξ i,1 , ξ i,2 , ξ i,3 T (i ∈ {1, 2, …, N a ​}) Thus, the Hermite coefficient matrix H(ξ) of the second-order three-dimensional SRSM model is determined.

[0077] Then, through the Nataf inverse transformation, the standardized SRSM sample ξ i (i ∈ {1, 2, …, N a}) is transformed into X i (i ∈ {1, 2, …, N a}) to obtain the uncertain prediction errors of the 10 groups of random variables corresponding to WP&L Respond to the FAS output and operating cost of the microgrid scheduling stochastic response model. Therefore, the stochastic response model based on SRSM can be mapped to a second-order three-dimensional SRSM model composed of 10 groups of deterministic standardized SRSM samples X i .

[0078] S5: Construct the scheduling model.

[0079] It should be noted that the construction of the NAS microgrid scheduling models MT, FC, SL, and ESS is not applicable to the intraday adjustment of the random fluctuations of WP&L, but only participates in the day-ahead scheduling. Their scheduling models are mainly composed of deterministic constraint equations to ensure a stable role in the operation of the microgrid. Specifically as follows:

[0080] MT, FC, SL, and ESS are not applicable to the intraday adjustment of the random fluctuations of WP&L, but only participate in the day-ahead scheduling. Therefore, they are only composed of deterministic constraint equations. The constraint conditions of MT and FC are as follows:

[0081]

[0082] The constraint conditions of SL are as follows:

[0083]

[0084] ESS mainly participates in the day-ahead scheduling and plays the role of peak shaving and valley filling. There are many forms of ESS. At present, the battery ESS is the most commonly used form in the microgrid. Therefore, the present invention takes the battery ESS as the object for mathematical modeling.

[0085] Considering the self-discharge rate (δ ESS ) and the charge and discharge efficiency (η ch ) and the charge and discharge efficiency (η dis ), and considering the limitation of the daily charge and discharge times, the ESS constraint conditions are as follows:

[0086]

[0087] Furthermore, to construct the FAS microgrid scheduling model, for GRID, DE, WT, PV, and RL, their scheduling models need to embed the stochastic response model based on SRSM, which is specifically as follows:

[0088] FAS includes GRID, DE, WT, PV, and RL, and their scheduling models need to embed the stochastic response model based on SRSM.

[0089] In the interconnected mode, GRID can act as a balancing node to counter random fluctuations. However, since the interactive power is one of the main decision variables in the day-ahead scheduling plan, it is not advisable to deviate significantly from the scheduled value during intraday adjustment, otherwise it will cause inconvenience to the distribution network scheduling. Therefore, under the constraint of the stochastic response model, a regulation range γ GRID ×P GRID,max . In the planned island operation mode, the interactive power is zero during the island period. Therefore, the GIRD stochastic response model composed of Na standardized SRSM samples (N a = 10) includes the interconnected and island operation modes:

[0090]

[0091] For fast DE, considering the constraint of the economic load ratio, DE will automatically shut down when the power load is less than the minimum load ratio. The DE fuel consumption model is a quadratic function and requires further convex cone relaxation. The stochastic response model of DE is:

[0092]

[0093] Generally, WT and PV should track their maximum output power. If the microgrid cannot consume the maximum power of WT or PV, the shedding of WT or PV should be considered. The stochastic response models of WT and PV are:

[0094]

[0095] The schedule of RL is similar to that of WT and PV. The stochastic response model of RL is:

[0096]

[0097] To sum up, the adjusted power expectation value embedded in the FAS of the standardized SRSM samples is:

[0098]

[0099] In the formula,

[0100]

[0101] As can be seen from the above formula, the SRSM model maps the random response characteristics of SRSM samples to the coefficients through the known matrix H(ξ). -1 and maps the random response characteristics of SRSM samples to the coefficients Since the above formula is linear, introducing the SRSM model will not change the convexity of the microgrid optimal scheduling model, which is the guarantee for the optimization model to be solved.

[0102] S6: The system constructs a multi-objective function. When meeting extreme constraint conditions, minimizing the operation cost and its fluctuation and suppressing the deviation from the plan caused by uncertainty should be the goal of the stochastic optimization strategy for the microgrid day-ahead scheduling. In addition, in response to low-carbon environmental protection, the carbon emission cost also needs to be considered in the optimization objective.

[0103] It should be noted that the construction of the multi-objective function is as follows:

[0104] In the proposed microgrid optimal scheduling model, the scheduling strategy for the economic, stable and reliable operation of the microgrid under WP&L uncertainty is simulated. Therefore, when meeting extreme constraint conditions, minimizing the operation cost and its fluctuation and suppressing the deviation from the plan caused by uncertainty should be the goal of the stochastic optimization strategy for the microgrid day-ahead scheduling. In addition, in response to low-carbon environmental protection, the carbon emission cost also needs to be considered in the optimization objective. Thus, the multi-objective function for the microgrid optimal scheduling considering uncertainty can be expressed as:

[0105]

[0106] In the formula,

[0107] f1 represents the expected value of the microgrid operation cost, which is used to measure the average level of the microgrid operation cost during the entire scheduling period. Among them, is the mathematical expectation of the microgrid operation cost Y Cop(t) at time t, which reflects the average trend of the operation cost in this period. Since the expected value of the random response output of the microgrid operation cost is equal to the first element of the corresponding Hermite chaos polynomial coefficient matrix,

[0108] f2 represents the variance of the microgrid operation cost, which is used to measure the fluctuation degree of the operation cost during the scheduling period. The smaller the variance, the more stable the operation cost. Among them, D[Y Cop(t) is the variance of the microgrid operation cost Y Cop(t) at time t, which reflects the degree of dispersion of the operation cost around the expected value. In the stochastic response surface method, since the variance D[Y Cop(t) of the random response output Y Cop(t) of the microgrid operation cost is equal to its chaos polynomial coefficient A Cop(t)The sum of the squares of all elements in the matrix except the first element, and the number of standardized SRSM samples N of the second-order three-dimensional SRSM model in this paper a = 10. Therefore, when summing the variances of the operating costs in each period within the scheduling cycle,

[0109] f3 represents the expected value of the carbon emission cost of the microgrid, which is used to evaluate the average carbon emission cost of the microgrid throughout the scheduling cycle, so as to reflect the low-carbon performance of the microgrid operation. Among them, is the carbon emission cost of the microgrid at time t The mathematical expectation of reflects the average level of the carbon emission cost in this period. Since the expected value of the stochastic response output of the microgrid operating cost is equal to the first element of the corresponding Hermite chaos polynomial coefficient matrix,

[0110] contains the Hermite chaos polynomial coefficients related to the operating cost (when K = OP) or carbon emission cost (when K = C) of the microgrid at time t. H(ξ) is the Hermite coefficient matrix, which is determined by the standardized SRSM samples and is used in the SRSM model to construct the relationship between the stochastic response and the standardized samples. H(ξ) -1 is its inverse matrix, which is used to solve the chaos polynomial coefficients. is a cost-related variable. When K = OP, it represents the microgrid operating cost including stochastic response in the period; when K = C, it represents the microgrid carbon emission cost including stochastic response in the period.

[0111] In summary, a conclusion can be drawn. Through the above steps, the accuracy of the SRSM model and the functions and advantages of the day-ahead scheduling strategy proposed in this paper are obtained.

[0112] In summary, through the optimized microgrid scheduling method based on the stochastic response surface method scheduling strategy proposed by the present invention, the uncertainty factors of WP&L can be comprehensively considered, just like building a solid defense line for the operation of the microgrid, successfully suppressing the rising trend of operating costs and carbon emissions. The distribution trends of its operating costs and carbon emissions are more concentrated and significantly reduced compared with other schemes. Especially in the face of complex scenarios with random fluctuations, this strategy has strong adaptability. Even in the case of large fluctuations, it can still accurately evaluate the response of WP&L random fluctuations. After being verified in various scenarios in all directions, it provides a rock-solid guarantee for the microgrid to achieve stable, economic, efficient, green and low-carbon operation in an environment full of uncertainties, and strongly promotes the improvement of energy utilization efficiency and the optimization of environmental benefits.

[0113] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0114] Embodiment 2

[0115] The second embodiment of the present invention provides an optimized microgrid scheduling system based on a random response surface method scheduling strategy, which is characterized in that it includes a data acquisition and preprocessing module, an abnormal behavior modeling module, a GAN abnormal generation module, an abnormal feature recognition module, and an optimal response strategy module.

[0116] Data acquisition and preprocessing module: Multi-modal data acquisition and preprocessing;

[0117] Abnormal behavior modeling module: Abnormal behavior modeling;

[0118] GAN abnormal generation module: Using a generative adversarial network to generate simulated abnormal samples;

[0119] Abnormal feature recognition module: Based on the trained model, identifying the abnormal features of the load behavior and classifying them;

[0120] Optimal response strategy module: Generating an optimal response strategy through reinforcement learning, dynamically adjusting the load or cutting off the power supply, minimizing the response cost and shortening the service restoration time.

[0121] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solutions of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solutions, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, etc., which can store program codes.

[0122] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definitional sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0123] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0124] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0125] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

[0126] Example 3

[0127] Reference Figures 2 - 3, which is the third embodiment of the present invention, provides an optimized microgrid scheduling method based on a stochastic response surface method scheduling strategy. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0128] Taking a microgrid composed of WT, PV, MT, FC, DE, ESS and demand response as an example. The present invention selects four different representative days as cases and conducts a cluster analysis on the meteorological data for nearly one year. The interconnected operation scenario and island operation scenario are set as follows:

[0129] Scenario 1: Interconnected operation scenario of prediction curve 1.

[0130] Scenario 2: Interconnected operation scenario of prediction curve 2.

[0131] Scenario 3: Island operation scenario, prediction curve 3, and the planned islanding period is from 14:00 to 18:00.

[0132] Scenario 4: Island operation scenario, prediction curve 4, and the planned islanding period is the whole day.

[0133] For the above 4 scenarios, the proposed SO strategy based on SRSM is used to optimize the day-ahead microgrid scheduling calculation. And the conventional RO strategy and the conventional deterministic optimization (DO) strategy are respectively used for comparative verification.

[0134] The experimental results are attached Figure 2 and attached Figure 3 show that Scenario 1 and Scenario 2 have the same prediction curve, but the random fluctuations in Scenario 2 are more intense. The fluctuations in the operating costs of the RO strategy and the DO strategy will increase with the fluctuations of WP&L, while the fluctuations of the SO strategy change relatively less. When the uncertainty of the prediction curve increases, the PDF curves of the operating costs and carbon emission costs of each strategy will shift in the positive direction. Among them, the PDF curve of the SO strategy is more concentrated, and the operating costs and carbon emission costs of the SO strategy are much lower than those of other strategies. It can be seen that the proposed SO model can accurately evaluate the response to the random fluctuations of WP&L, and at the same time can reduce the operating costs and carbon emissions.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An optimized microgrid dispatching method based on a random response surface method dispatching strategy is characterized by: include, Standardized random variables; Construct the coefficient matrix of the Hermite chaotic polynomial; Classify source load types and build short-term prediction models; Build a scheduling model; The system constructs multi-objective functions.

2. The method for optimizing microgrid dispatching based on random response surface method dispatching strategy according to claim 1, characterized in that: The standardized random variables are transformed into independent standard normal distribution variables by using Nataf transformation.

3. The method for optimizing microgrid dispatching based on random response surface method dispatching strategy according to claim 2, characterized in that: The coefficient matrix of constructing the Hermite chaotic polynomial includes that the mapping relationship between the random response output Y and the random response output obeying the independent standard normal random distribution can be constructed into a chaotic polynomial according to the Hermite orthogonal polynomial.

4. The method for optimizing microgrid dispatching based on random response surface method dispatching strategy according to claim 3, characterized in that: The source load type classification is performed based on whether it can be flexibly adjusted.

5. The method for optimizing microgrid dispatching based on random response surface method dispatching strategy according to claim 4, characterized in that: The short-term prediction model is constructed based on K-MEANS clustering and long short-term memory neural network to construct the short-term prediction model of WP&L. The WP&L historical data is used for training and verification of the long short-term memory neural network model to obtain the Gaussian distribution function and parameters of the model prediction error.

6. The method for optimizing microgrid dispatching based on random response surface method dispatching strategy according to claim 5, characterized in that: The construction of the scheduling model includes a FAS microgrid scheduling model and a NAS microgrid scheduling model.

7. The method for optimizing microgrid dispatching based on random response surface method dispatching strategy according to claim 6, characterized in that: The system construction multi-objective function is expressed as: In the formula, f1 represents the expected value of the microgrid operation cost, E[Y Cop(t) ] is the operating cost Y of the microgrid in period t Cop(t) The mathematical expectation of f2 represents the variance of the microgrid operation cost, D[Y Cop(t) ] is the operating cost Y of the microgrid in period t Cop(t) The variance of f3 represents the expected value of the carbon emission cost of the microgrid, is the carbon emission cost of the microgrid in period t The mathematical expectation of Contains the Hermite chaotic polynomial coefficients related to the operating cost or carbon emission cost of the microgrid in time period t. H(ξ) is the Hermite coefficient matrix, H(ξ) -1 is the inverse matrix, It is a cost-related variable. When K = OP, it indicates that the time period contains the microgrid operation cost of random response; when K = C, it indicates that the time period contains the microgrid carbon emission cost of random response.

8. A system for optimizing microgrid dispatching method based on random response surface method dispatching strategy according to any one of claims 1 to 7, characterized in that: It includes variable standardization module, coefficient matrix construction module, source load classification prediction module, scheduling model construction module and multi-objective optimization module.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images