Virtual power plant economic optimization scheduling method based on fusion optimization algorithm

Abstract

The invention discloses a virtual power plant economic optimization scheduling method based on a fusion optimization algorithm, and belongs to the technical field of virtual power plant energy scheduling. The method aims at solving the problems that an existing power dispatching algorithm lacks the capacity of fully dispatching distributed resources and flexible loads, and an optimal power load distribution scheme cannot be obtained. Comprising the following steps: acquiring a load value at a predicted target moment, a photovoltaic power generation output value at the target moment, a wind power generation output value at the target moment, charge and discharge potential data of the distributed energy storage equipment and constraint conditions of the distributed energy storage equipment; and a virtual power plant economic optimization objective function is established, and a moss group optimization algorithm is adopted to carry out power supply scheduling distribution at a target moment, so that common power distribution output of the distributed photovoltaic power generation equipment, the distributed wind power generation equipment, the distributed thermal power generating unit and the distributed energy storage equipment meets a load value at the target moment, and optimal economic performance is obtained. The method is used for economic optimization scheduling of the virtual power plant.

Description

Technical Field

[0001] The invention relates to an economic optimization dispatching method for a virtual power plant based on a fusion optimization algorithm, and belongs to the technical field of energy dispatching of virtual power plants. Background Art

[0002] With the rapid economic development and the continuous growth of population, the demand for electricity is increasing day by day, and renewable energy is developing rapidly. Photovoltaic power generation has also gradually changed from centralized to distributed and centralized, and has gradually become an important part of the power system together with wind power generation. However, due to the frequent fluctuations in the output of renewable energy, its reliability is poor, and it has strong indirectness and instability. The instability of output will bring great difficulties to the peak and frequency regulation of the power grid. Renewable energy has brought development to the power system, but also challenges, such as demand-side management, grid stability and other issues that need to be solved urgently.

[0003] In order to further enhance the competitiveness of distributed energy in the power market and strengthen the control over distributed energy, the distributed energy, energy storage system and traditional power system are organically integrated to improve energy utilization efficiency and enhance the flexibility and stability of the power system.

[0004] The emergence of virtual power plants is an innovation of the traditional power system model. Through intelligent technology and network connection, centralized management and optimized dispatch of energy are realized. When faced with the complexity of virtual power plant VPP systems, traditional power dispatch algorithms lack the ability to fully dispatch distributed resources and flexible loads, and often find it difficult to balance economic benefits, emission reduction, and energy efficiency. Therefore, in order to maximize economic benefits while ensuring the stable operation of the power system, a virtual power plant economic optimization dispatch method is needed. Summary of the invention

[0005] In view of the problem that the existing power dispatching algorithms lack the ability to fully dispatch distributed resources and flexible loads and cannot obtain the optimal solution for power load distribution, the present invention provides a virtual power plant economic optimization dispatching method based on a fusion optimization algorithm.

[0006] A virtual power plant economic optimization scheduling method based on a fusion optimization algorithm of the present invention comprises:

[0007] The load value at the target moment is predicted through an integrated learning algorithm based on the historical load data of the virtual power plant; at the same time, the photovoltaic power generation output value of the distributed photovoltaic power generation equipment at the target moment is predicted through an integrated learning algorithm based on the photovoltaic power generation related data, and the wind power generation output value of the distributed wind power generation equipment at the target moment is predicted through an integrated learning algorithm based on the wind power generation related data;

[0008] Obtaining charging and discharging potential data of distributed energy storage equipment; and determining constraints of distributed energy storage equipment;

[0009] According to the load value at the target moment, the photovoltaic power generation output value at the target moment, the wind power generation output value at the target moment and the charging and discharging potential data, as well as the constraints of the distributed energy storage equipment and the economic optimization objective function of the virtual power plant, the moss swarm optimization algorithm is used to perform power supply scheduling and allocation at the target moment, so that the common distribution output of the distributed photovoltaic power generation equipment, the distributed wind power generation equipment, the distributed thermal power units and the distributed energy storage equipment can meet the load value at the target moment and obtain the optimal economic performance.

[0010] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the prediction method of the photovoltaic power generation output value at the target time includes:

[0011] According to the influence of sunshine intensity, temperature and terrain conditions at the corresponding historical moment, combined with meteorological data, the photovoltaic power generation output value of distributed photovoltaic power generation equipment at the target moment is predicted;

[0012] According to the influence of wind speed, wind direction and humidity at the corresponding historical moment and combined with meteorological data, the wind power output value of distributed wind power generation equipment at the target moment is predicted.

[0013] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the charging and discharging potential data of the distributed energy storage device include the state of charge, health state, voltage and current;

[0014] Extracting time characteristics and derived variables of charge and discharge potential data;

[0015] Use the random forest model to predict the state of charge at the target time based on the corresponding historical data;

[0016] The LSTM model is used to predict time features based on the corresponding historical data and capture the dynamic changes of the charging and discharging potential data.

[0017] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the constraints of the distributed energy storage equipment include:

[0018] Energy balance constraints:

[0019] E t+1 =E t +η charge ·P charge,t -η dlischarge ·P dlscharge,t ,

[0020] Where E t+1 is the stored energy of the distributed energy storage device at time t+1, E t is the stored energy of the distributed energy storage device at time t, η charge is the charging efficiency of distributed energy storage equipment, Pcharge,t is the charging rate of the distributed energy storage device at time t, η dlischarge is the energy release efficiency of distributed energy storage equipment, P dlscharge,t is the energy release rate of the distributed energy storage device at time t;

[0021]

[0022] In the formula is the maximum charging rate, is the maximum energy release rate;

[0023] E min ≤E t ≤E max ,

[0024] Where E min is the minimum value of reserve energy, E max The maximum value of reserve energy.

[0025] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the virtual power plant economic optimization objective function is:

[0026] min total =C OM +C fuel +C gridbuy +C gridsell +C storge +C L +C CO2 ,

[0027] Where M total is the total cost of power distribution, C OM is the total power generation cost, C fuel is the fuel cost, C gridbuy is the cost of purchasing electricity from the upper grid, C gridsell is the cost of selling electricity to the upper grid, C storge is the energy storage cost, C L is the dispatch compensation cost of thermal power units, C CO2 The cost of carbon emissions.

[0028] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the moss colony optimization algorithm includes:

[0029] Define decision variables I i :

[0030] I i =[P load ,P pv-i ,P wind-i , P Gt-i , P battery-i ],

[0031] i=1,2,3,……,n;

[0032] Define the moss population F:

[0033] F=[I1,I2,...,I n ],

[0034] Where P load is the load power, which is determined according to the load value at the target time, P pv-i is the photovoltaic power in the i-th power supply dispatching allocation strategy, which is determined according to the photovoltaic power generation output value at the target time, P wind-i is the wind power in the i-th power supply dispatching allocation strategy, which is determined according to the wind power output value at the target time, P Gt-i is the thermal power in the i-th power supply dispatch allocation strategy, P battery-i The energy storage power in the i-th power supply dispatch allocation strategy is determined according to the charging and discharging potential data;

[0035] Initialize the moss population and randomly generate the moss population F within the range of the decision variable. i As a moss individual;

[0036] Iterative calculation: Calculate moss individuals I i The fitness value of each objective of the virtual power plant economic optimization objective function is obtained, and the initial Pareto optimal solution set is obtained;

[0037] Select the moss individuals whose fitness values ​​on each target in the initial Pareto optimal solution set are greater than the fitness threshold, expand to a better area in the solution space, find new moss individuals to add to the initial Pareto optimal solution set, and make the initial Pareto optimal solution set concentrate on the optimal solution area until the moss individuals in the concentrated Pareto optimal solution set reach the target number;

[0038] The current position and fitness value information of each moss individual is shared among the moss individuals in the concentrated Pareto optimal solution set, and the moss individuals with fitness values ​​greater than the fitness threshold are retained through a competition mechanism, so that the number of moss individuals in the concentrated Pareto optimal solution set is equal to the number of moss individuals in the initial Pareto optimal solution set; then return to the next round of iterative calculation until the termination condition is met; the power supply scheduling allocation strategy corresponding to the moss individual with the largest fitness value on each target in the final Pareto optimal solution set is selected as the final power supply scheduling allocation strategy.

[0039] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the fitness matrix F0 of the moss population F is:

[0040]

[0041] Where F n4 represents the fitness value of the nth moss individual on the fourth goal;

[0042] When sorting the initial Pareto optimal solution set, the indicator matrix D is used to represent the dominance relationship between moss individuals. If moss individual i dominates moss individual i′, then D ii′ =1, otherwise D ii′ =0;i′=1,2,3,...,n;

[0043]

[0044] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, assuming that the number of moss individuals whose fitness values ​​on various objectives are greater than the fitness threshold in the initial Pareto optimal solution is m, then the m moss individuals expand to a better area in the solution space, and the updated position I new It is expressed as:

[0045] I new =I+Δ,

[0046] I=[I1,I2,...,I m ];

[0047] Where Δ is the individual expansion matrix:

[0048]

[0049] Where δ m is the expansion amount of the mth moss individual.

[0050] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention,

[0051] Set F max and F min The maximum and minimum fitness values ​​of each target, and the normalized fitness value F ij The calculation method of ′ is:

[0052]

[0053] Wherein j = 1, 2, 3, 4;

[0054] Normalized fitness value F ij ′ is in the interval [0, 1];

[0055] The competition mechanism is realized by operating the fitness matrix F0. Each value in the fitness matrix F0 is normalized to form a new matrix R:

[0056]

[0057] In the competition phase, the position I of the retained moss individual is selected according to the value of R final for:

[0058] I final =I[R>τ],

[0059] Where τ is the set fitness threshold.

[0060] According to the virtual power plant economic optimization scheduling method based on the fusion optimization algorithm of the present invention, the termination condition is that the mutation rate is less than the mutation rate threshold ∈:

[0061]

[0062] In the formula is the optimal fitness value of the Tth iteration.

[0063] Beneficial effects of the invention: The invention can solve the problem of the lack of ability to fully dispatch distributed resources and flexible loads in existing scheduling algorithms, select the power supply scheduling allocation strategy through the moss colony optimization algorithm, output the optimal load plan while ensuring the stability and economy of the virtual power plant, and then adjust the allocation plan of each distributed resource according to the load plan output by the virtual power plant, thereby improving the economy and stability of the virtual power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is an overall flow chart of the economic optimization dispatching method of a virtual power plant based on a fusion optimization algorithm according to the present invention;

[0065] Figure 2 It is a flow chart for predicting wind power output value;

[0066] Figure 3 It is a flow chart for predicting photovoltaic power generation output value;

[0067] Figure 4 It is the virtual power plant load forecasting flow chart;

[0068] Figure 5 It is the flow chart of moss colony optimization algorithm. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0071] The present invention will be further described below in conjunction with the accompanying drawings, but is not intended to be a limitation of the present invention.

[0072] Combination Figures 1 to 4 As shown, the present invention provides a virtual power plant economic optimization scheduling method based on a fusion optimization algorithm, comprising:

[0073] The load value at the target moment is predicted through an integrated learning algorithm based on the historical load data of the virtual power plant; at the same time, the photovoltaic power generation output value of the distributed photovoltaic power generation equipment at the target moment is predicted through an integrated learning algorithm based on the photovoltaic power generation related data, and the wind power generation output value of the distributed wind power generation equipment at the target moment is predicted through an integrated learning algorithm based on the wind power generation related data;

[0074] Obtaining charging and discharging potential data of distributed energy storage equipment; and determining constraints of distributed energy storage equipment;

[0075] According to the load value at the target moment, the photovoltaic power generation output value at the target moment, the wind power generation output value at the target moment and the charging and discharging potential data, as well as the constraints of the distributed energy storage equipment and the economic optimization objective function of the virtual power plant, the moss swarm optimization algorithm is used to perform power supply scheduling and allocation at the target moment, so that the common distribution output of the distributed photovoltaic power generation equipment, the distributed wind power generation equipment, the distributed thermal power units and the distributed energy storage equipment can meet the load value at the target moment and obtain the optimal economic performance.

[0076] This implementation method predicts the load and distributed power generation output of the virtual power plant based on the historical load distribution of the virtual power plant and the performance parameters of the distributed power generation equipment, combined with the local specific weather forecast information. By obtaining the status of the distributed energy storage equipment in the virtual power plant, the potential for charging and discharging of the energy storage equipment is obtained. According to the predicted data information and decision variables of the virtual power plant, the moss population of the economic scheduling plan of the virtual power plant is initialized; the fitness of the moss population is calculated, and the Pareto optimal solution set is obtained according to the multi-objective function. According to the minimum value of each objective of the Pareto optimal solution set, the moss individuals expand to a better area in the solution space according to their fitness values, and the Pareto solution set is concentrated in the optimal solution area; the moss individuals share their current position and fitness information to help other individuals refer to excellent scheduling strategies. The moss individuals will compete with each other, and the poorer scheduling plans will be eliminated, and the plans with higher fitness will be retained; until the power supply scheduling allocation strategy is finally determined.

[0077] Combination Figure 2 and Figure 3As shown, the prediction method of photovoltaic power generation output value at the target time includes:

[0078] According to the influence of sunshine intensity, temperature and terrain conditions on distributed photovoltaic power generation equipment at the corresponding historical moment, combined with meteorological data, the photovoltaic power generation output value of distributed photovoltaic power generation equipment at the target moment is predicted;

[0079] According to the influence of wind speed, wind direction and humidity on the output power of distributed wind power generation equipment at the corresponding historical moment, combined with meteorological data, the wind power output value of distributed wind power generation equipment at the target moment is predicted.

[0080] The original data is input into the base model of the integrated learning algorithm and introduced into the system. The original data includes historical meteorological data, geographical data, external influencing factors and other variables. The pre-processed data is trained by the base model. Finally, the prediction results of all base models are integrated and then the integrated prediction results are processed to obtain the final output prediction value of distributed power generation equipment. The joint probability distribution of the output prediction value of distributed power generation equipment is shown as follows:

[0081] f(v,t,p,h)=C te ·C sh ·C in ·f v (v; k, λ)f t (t; μ t ,σ t )f p (p; μ p ,σ p )f h (h; μ h ,σ h )

[0082] Among them, f t (t; μ t ,σ t ) is the probability distribution of temperature, f p (p; μ p ,σ p ) is the probability distribution of air pressure, f h (h; μ h ,σ h ) is the probability distribution of humidity, C te is the terrain correction factor, C sh is the error correction coefficient, C in It is the correction factor for mutual influence between units.

[0083] Furthermore, the constraints in this embodiment include economic optimization constraints of multiple virtual power plants and network topology constraints of multiple virtual power plants.

[0084] The charging and discharging potential data of distributed energy storage devices include state of charge SOC, state of health SOH, voltage and current;

[0085] Extracting time characteristics and derived variables of charge and discharge potential data;

[0086] Use the random forest model to predict the state of charge at the target time based on the corresponding historical data;

[0087] The LSTM model is used to predict time features based on the corresponding historical data and capture the dynamic changes of the charging and discharging potential data.

[0088] In this implementation, the constraints of the distributed energy storage device include:

[0089] Energy balance constraints:

[0090] E t+1 =E t +η charge ·P charge,t -η dlischarge ·P dlscharge,t ,

[0091] Where E t+1 is the stored energy of the distributed energy storage device at time t+1, E t is the stored energy of the distributed energy storage device at time t, η charge is the charging efficiency of distributed energy storage equipment, P charge,t is the charging rate of the distributed energy storage device at time t, η dlischarge is the energy release efficiency of distributed energy storage equipment, P dlscharge,t is the energy release rate of the distributed energy storage device at time t;

[0092] The charging and discharging power constraints of the energy storage device are as follows:

[0093]

[0094] In the formula is the maximum charging rate, is the maximum energy release rate;

[0095] The energy storage constraint of the energy storage device is as follows:

[0096] E min ≤E t ≤E max ,

[0097] Where E min is the minimum value of reserve energy, E max The maximum value of reserve energy.

[0098] The energy storage equipment information obtained above is a necessary constraint condition for optimal scheduling.

[0099] Furthermore, the economic optimization objective function of the virtual power plant is:

[0100] min M total =C OM +C fuel +C gridbuy +C gridsell +C storge +C L +C CO2 ,

[0101] Where M total is the total cost of power distribution, C OM is the total power generation cost, C fuel is the fuel cost, C gridbuy is the cost of purchasing electricity from the upper grid, C gridsell is the cost of selling electricity to the upper grid, C storge is the energy storage cost, C L is the dispatch compensation cost of thermal power units, C CO2 The cost of carbon emissions.

[0102] The moss optimization algorithm of this embodiment proposes the following rules based on the growth and ecological behavior of mosses in nature:

[0103] (1) In the algorithm, the fitness value of an individual is equivalent to the survival ability of the moss individual. Individuals with high fitness represent better solutions in target optimization, simulating the process of natural selection.

[0104] (2) Mosses occupy new areas by reproduction and expansion. Through individual expansion, individuals move to better positions in the solution space according to their fitness.

[0105] (3) Moss individuals can transmit information through chemical signals or other means to help each other find better growth conditions. Individuals in the moss algorithm will share their own fitness and location information, which will encourage other individuals to learn from excellent strategies, thereby enhancing the overall search capability.

[0106] (4) As moss individuals expand, the moss community gradually grows, and the number of candidate solutions in the solution space also increases. In order to prevent too many solutions or poor quality solutions from occupying resources, a competition mechanism is introduced into the moss optimization algorithm: in the solution space, nutrients are limited, and moss individuals will compete with each other for nutrient resources. If the fitness of a moss individual is not as good as that of the individuals around it, it will lose nutrients and decline, or even die.

[0107] (5) By comparing the fitness values ​​of moss individuals, individuals with poor fitness are screened and eliminated, and individuals with higher fitness are retained. This process can reduce the computational overhead of the algorithm and at the same time encourage moss individuals to concentrate in the optimal solution area.

[0108] (6) Mosses can quickly adjust their growth strategies when faced with environmental changes. In the algorithm’s dynamic adjustment mechanism, individuals continuously update their positions and strategies based on changes in the environment and fitness to avoid local optimality.

[0109] This implementation method uses the moss optimization algorithm to search for the best solution with economic benefits as the goal, and obtains a scheduling plan with the lowest operating cost for the virtual power plant.

[0110] The initial stage of the moss optimization algorithm is similar to planting moss. A certain number of candidate solutions are randomly generated in the solution space, called "moss individuals", which can be regarded as the initial colony of moss. Each moss individual corresponds to a specific solution, and its position is determined by the decision variable of the problem.

[0111] Combination Figure 5 As shown, the moss colony optimization algorithm includes:

[0112] Define decision variables I i :

[0113] I i =[P load ,P pv-i ,P wind-i , P Gt-i , P battery-i ],

[0114] i=1,2,3,……,n;

[0115] The parameters of the moss optimization algorithm are set and the population is initialized; assuming that the initial population size is F, the initial population is positioned within the value range of the decision variable and the initial moss population is randomly generated, and each individual represents a possible scheduling strategy. The initial individuals are generated based on historical data.

[0116] Define the moss population F:

[0117] F=[I1,I2,...,I n ],

[0118] Where P load is the load power, which is determined according to the load value at the target time, P pv-i is the photovoltaic power in the i-th power supply dispatching allocation strategy, which is determined according to the photovoltaic power generation output value at the target time, P wind-i is the wind power in the i-th power supply dispatching allocation strategy, which is determined according to the wind power output value at the target time, P Gt-i is the thermal power in the i-th power supply dispatch allocation strategy, P battery-i The energy storage power in the i-th power supply dispatch allocation strategy is determined according to the charging and discharging potential data;

[0119] Initialize the moss population and randomly generate the moss population F within the range of the decision variable. i As a moss individual;

[0120] Iterative calculation: Calculate moss individuals I i The fitness value of each objective of the virtual power plant economic optimization objective function is obtained, and the initial Pareto optimal solution set is obtained;

[0121] Select the moss individuals whose fitness values ​​on each target in the initial Pareto optimal solution set are greater than the fitness threshold, expand to a better area in the solution space, find new moss individuals to add to the initial Pareto optimal solution set, and make the initial Pareto optimal solution set concentrate on the optimal solution area until the moss individuals in the concentrated Pareto optimal solution set reach the target number;

[0122] The current position and fitness value information of each moss individual is shared among the moss individuals in the concentrated Pareto optimal solution set, and the moss individuals with fitness values ​​greater than the fitness threshold are retained through a competition mechanism, so that the number of moss individuals in the concentrated Pareto optimal solution set is equal to the number of moss individuals in the initial Pareto optimal solution set; then return to the next round of iterative calculation until the termination condition is met; the power supply scheduling allocation strategy corresponding to the moss individual with the largest fitness value on each target in the final Pareto optimal solution set is selected as the final power supply scheduling allocation strategy.

[0123] The fitness function is defined according to the objective function of the problem, and is usually measured by multiple objectives such as economy, cost, emissions, and renewable energy utilization rate to measure the pros and cons of each solution in the virtual power plant scheduling. For multi-objective optimization problems, the fitness function can comprehensively consider multiple objectives to achieve balanced optimization of multiple objectives. By calculating the fitness value of each individual on each objective, its performance under a specific objective is evaluated. It helps to determine which individual performs better in meeting different objectives. Evaluating fitness values ​​can help select and retain individuals with excellent performance in each iteration. Individuals with high fitness will be retained first to enhance the overall performance of the group. Assuming that there are n individuals in the group, each individual has 4 objective functions in this implementation.

[0124] In this implementation, the fitness matrix F0 of the moss population F is constructed as:

[0125]

[0126] Where F n4 represents the fitness value of the nth moss individual on the fourth goal;

[0127] Adaptive evaluation allows the algorithm to identify non-dominated solutions and form the Pareto frontier. The individuals on the Pareto frontier represent the best trade-offs between different objectives, providing decision makers with a variety of options.

[0128] When sorting the initial Pareto optimal solution set, the indicator matrix D is used to represent the dominance relationship between moss individuals. If moss individual i dominates moss individual i′, then D ii′ =1, otherwise D ii′ =0;i′=1,2,3,...,n;

[0129]

[0130] Moss individuals will expand like bryophytes and grow towards the better areas in the solution space. This process is achieved through the expansion mechanism, that is, each moss individual chooses the best expansion direction according to the surrounding environment and nutrient distribution. Generally, moss individuals tend to grow towards areas with richer nutrients and higher fitness. The expansion step size determines the distance the moss individual moves during each expansion process. The expansion of individuals can be achieved through vector and matrix operations.

[0131] Assuming that the number of moss individuals whose fitness values ​​on each target are greater than the fitness threshold in the initial Pareto optimal solution is m, then m moss individuals expand to a better area in the solution space, and the updated position I new It is expressed as:

[0132] I new =I+Δ,

[0133] I=[I1,I2,...,I m ];

[0134] Where Δ is the individual expansion matrix:

[0135]

[0136] Where δ m is the expansion amount of the mth moss individual.

[0137] The step size in the moss optimization algorithm is dynamically adjusted. Usually, the step size is large in the early stage to perform global search. As the algorithm iterates, the step size gradually decreases and turns to local search. In each iteration, the moss individual moves in the solution space according to the expansion direction and step size to generate a new moss position.

[0138] In the process of information sharing, individuals can communicate with each other through the maximum and minimum values ​​of the fitness matrix.

[0139] Set F max and F minThe maximum and minimum fitness values ​​of each target, and the normalized fitness value F ij The calculation method of ′ is:

[0140]

[0141] Wherein j = 1, 2, 3, 4;

[0142] Normalized fitness value F ij ′ is in the interval [0, 1]; it is convenient for comparing and sharing information.

[0143] The competition mechanism is realized by operating the fitness matrix F0. Each value in the fitness matrix F0 is normalized to form a new matrix R:

[0144]

[0145] In the competition phase, the position I of the retained moss individual is selected according to the value of R final for:

[0146] I final =I[R>τ],

[0147] Where τ is the set fitness threshold.

[0148] The termination condition is that the mutation rate is less than the mutation rate threshold ∈:

[0149] If after several consecutive iterations, the change in fitness of all individuals is less than the set threshold ∈, the improvement is considered insufficient and the algorithm terminates:

[0150]

[0151] In the formula is the optimal fitness value of the Tth iteration.

[0152] The determination of insufficient improvement is an important termination mechanism of the moss algorithm. By monitoring the change in fitness value or population diversity, it is determined whether the algorithm is stuck. It ensures that the algorithm can be terminated in time when it is close to the optimal solution, avoiding unnecessary calculations, while maintaining a certain global search capability.

[0153] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that the features described in conjunction with a single embodiment may be used in other described embodiments.

Claims

1. A virtual power plant economic optimization scheduling method based on a fusion optimization algorithm, characterized in that include, The load value at the target moment is predicted through an integrated learning algorithm based on the historical load data of the virtual power plant; at the same time, the photovoltaic power generation output value of the distributed photovoltaic power generation equipment at the target moment is predicted through an integrated learning algorithm based on the photovoltaic power generation related data, and the wind power generation output value of the distributed wind power generation equipment at the target moment is predicted through an integrated learning algorithm based on the wind power generation related data; Obtaining charging and discharging potential data of distributed energy storage equipment; and determining constraints of distributed energy storage equipment; According to the load value at the target moment, the photovoltaic power generation output value at the target moment, the wind power generation output value at the target moment and the charging and discharging potential data, as well as the constraints of the distributed energy storage equipment and the economic optimization objective function of the virtual power plant, the moss swarm optimization algorithm is used to perform power supply scheduling and allocation at the target moment, so that the common distribution output of the distributed photovoltaic power generation equipment, the distributed wind power generation equipment, the distributed thermal power units and the distributed energy storage equipment can meet the load value at the target moment and obtain the optimal economic performance.

2. The economic optimization scheduling method for a virtual power plant based on a fusion optimization algorithm according to claim 1 is characterized in that: The prediction methods of photovoltaic power generation output value at the target time include: According to the influence of sunshine intensity, temperature and terrain conditions at the corresponding historical moment, combined with meteorological data, the photovoltaic power generation output value of distributed photovoltaic power generation equipment at the target moment is predicted; According to the influence of wind speed, wind direction and humidity at the corresponding historical moment and combined with meteorological data, the wind power output value of distributed wind power generation equipment at the target moment is predicted.

3. The economic optimization scheduling method for a virtual power plant based on a fusion optimization algorithm according to claim 2 is characterized in that: The charging and discharging potential data of distributed energy storage devices include state of charge, health status, voltage and current; Extracting time characteristics and derived variables of charge and discharge potential data; Use the random forest model to predict the state of charge at the target time based on the corresponding historical data; The LSTM model is used to predict time features based on the corresponding historical data and capture the dynamic changes of the charging and discharging potential data.

4. The economic optimization scheduling method for a virtual power plant based on a fusion optimization algorithm according to claim 3 is characterized in that: The constraints of distributed energy storage devices include: Energy balance constraints: E t+1 =E t +n charge ·P charge,t -or dlischarge ·P dlscharge,t , Where E t+1 is the stored energy of the distributed energy storage device at time t+1, E t is the stored energy of the distributed energy storage device at time t, η charge is the charging efficiency of distributed energy storage equipment, P charge,t is the charging rate of the distributed energy storage device at time t, η dlischarge is the energy release efficiency of distributed energy storage equipment, P dlscharge,t is the energy release rate of the distributed energy storage device at time t; In the formula is the maximum charging rate, is the maximum energy release rate; AND min ≤E t ≤E max , Where E min is the minimum value of reserve energy, E max The maximum value of reserve energy.

5. The economic optimization scheduling method for virtual power plants based on fusion optimization algorithm according to claim 4 is characterized in that: The economic optimization objective function of the virtual power plant is: minM total =C OM +C fuel +C gridbuy +C gridsell +C storge +C L +C CO2 , Where M total is the total cost of power distribution, C OM is the total power generation cost, C fuel is the fuel cost, C gridbuy is the cost of purchasing electricity from the upper grid, C gridsell is the cost of selling electricity to the upper grid, C storge is the energy storage cost, C L is the dispatch compensation cost of thermal power units, C CO2 The cost of carbon emissions.

6. The virtual power plant economic optimization scheduling method based on the fusion optimization algorithm according to claim 5 is characterized in that: The moss colony optimization algorithm includes: Define decision variables I i : I i =[P load ,P pv-i ,P wind-i ,P Gt-i ,P battery-i ], i=1,2,3,……,n; Define the moss population F: F=[I1,I2,...,I n ], Where P load is the load power, which is determined according to the load value at the target time, P pv-i is the photovoltaic power in the i-th power supply dispatching allocation strategy, which is determined according to the photovoltaic power generation output value at the target time, P wind-i is the wind power in the i-th power supply dispatching allocation strategy, which is determined according to the wind power output value at the target time, P Gt-i is the thermal power in the i-th power supply dispatch allocation strategy, P battery-i The energy storage power in the i-th power supply dispatch allocation strategy is determined according to the charging and discharging potential data; Initialize the moss population and randomly generate the moss population F within the range of the decision variable. i As a moss individual; Iterative calculation: Calculate moss individuals I i The fitness value of each objective of the virtual power plant economic optimization objective function is obtained, and the initial Pareto optimal solution set is obtained; Select the moss individuals whose fitness values ​​on each target in the initial Pareto optimal solution set are greater than the fitness threshold, expand to a better area in the solution space, find new moss individuals to add to the initial Pareto optimal solution set, and make the initial Pareto optimal solution set concentrate on the optimal solution area until the moss individuals in the concentrated Pareto optimal solution set reach the target number; The current position and fitness value information of each moss individual is shared among the moss individuals in the concentrated Pareto optimal solution set, and the moss individuals with fitness values ​​greater than the fitness threshold are retained through a competition mechanism, so that the number of moss individuals in the concentrated Pareto optimal solution set is equal to the number of moss individuals in the initial Pareto optimal solution set; then return to the next round of iterative calculation until the termination condition is met; the power supply scheduling allocation strategy corresponding to the moss individual with the largest fitness value on each target in the final Pareto optimal solution set is selected as the final power supply scheduling allocation strategy.

7. The method for economic optimization scheduling of a virtual power plant based on a fusion optimization algorithm according to claim 6 is characterized in that: The fitness matrix F0 of the moss population F is: Where F n4 represents the fitness value of the nth moss individual on the fourth goal; When sorting the initial Pareto optimal solution set, the indicator matrix D is used to represent the dominance relationship between moss individuals. If moss individual i dominates moss individual i′, then D ii′ =1, otherwise D ii′ =0;i′=1,2,3,...,n; 8. The economic optimization scheduling method for virtual power plants based on fusion optimization algorithm according to claim 7 is characterized in that: Assuming that the number of moss individuals whose fitness values ​​on each target are greater than the fitness threshold in the initial Pareto optimal solution is m, then m moss individuals expand to a better area in the solution space, and the updated position I new It is expressed as: I new =I+Δ, I=[I1,I2,*,I m ]; Where Δ is the individual expansion matrix: Where δ m is the expansion amount of the mth moss individual.

9. The virtual power plant economic optimization scheduling method based on the fusion optimization algorithm according to claim 8 is characterized in that: Set F max and F min The maximum and minimum fitness values ​​of each target, and the normalized fitness value F ij The calculation method of ′ is: Wherein j = 1, 2, 3, 4; Normalized fitness value F ij ′ is in the interval [0, 1]; The competition mechanism is realized by operating the fitness matrix F0. Each value in the fitness matrix F0 is normalized to form a new matrix R: In the competition phase, the position I of the retained moss individual is selected according to the value of R final for: I final =I[R>τ], Where τ is the set fitness threshold.

10. The virtual power plant economic optimization scheduling method based on the fusion optimization algorithm according to claim 9 is characterized in that: The termination condition is that the mutation rate is less than the mutation rate threshold ∈: In the formula is the optimal fitness value of the Tth iteration.

Images