Scheduling method applied to multi-energy microgrid

Abstract

The invention relates to a scheduling method applied to a multi-energy microgrid, and the method comprises the steps: obtaining the output data and meteorological information of wind power photovoltaic, inputting the output data and meteorological information into a load prediction model, and obtaining a wind power photovoltaic net load prediction curve; according to the net load prediction curve, the optimal start-stop plan and output scheduling scheme for solving the gas turbine and the fuel cell are obtained by combining the minimum start-stop time and climbing rate constraint for solving the gas turbine and the fuel cell; the charging and discharging state of the energy storage equipment is obtained, a charging and discharging optimization model with the minimum scheduling cost as the target is established according to the charging and discharging state, and a charging and discharging strategy of the energy storage equipment is obtained; and integrating the optimal start-stop plan and the output scheduling scheme for solving the gas turbine and the fuel cell and the charging and discharging strategy of the energy storage equipment to form a complete energy scheduling scheme. According to the invention, coordinated optimization scheduling of various energy devices is realized, and the operation efficiency and economical efficiency of the multi-energy micro-grid are improved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and particularly to a scheduling method applied to a multi - energy microgrid. Background Art

[0002] In the optimal scheduling of a multi - energy microgrid, there are coupling characteristics and mutual influences among multiple energy systems, which pose challenges to modeling and optimal solution. The output of renewable energy sources such as wind power and photovoltaic power is intermittent and volatile, while conventional energy sources such as gas turbines and fuel cells are restricted by operation constraints such as start - stop time and ramp rate. In addition, the charge - discharge strategy of energy storage devices also has an important impact on the energy balance and economy of the system. When establishing an optimal scheduling model, it is necessary to comprehensively consider the complementarity and coordination among various energy sources, as well as the dynamic changes in load demand, in order to obtain a globally optimal scheduling plan. However, due to the complex non - linear relationship and time - varying characteristics among energy systems, traditional optimization methods are difficult to be directly applied. How to establish an accurate and efficient multi - energy coupling optimization model and use a suitable algorithm for solution on the premise of meeting the operation constraints of each device and system safety constraints is a key technical problem to be solved urgently. At the same time, in actual operation, it is also necessary to dynamically adjust and optimize the scheduling strategy according to real - time data to adapt to the changing external conditions and system states, which puts forward higher requirements for the robustness and real - time performance of the optimization algorithm. Summary of the Invention

[0003] In order to solve the problems existing in the above - mentioned prior art, the purpose of the present invention is to provide a scheduling method applied to a multi - energy microgrid, which realizes the coordinated optimal scheduling of multiple energy devices, effectively improves the operation efficiency and economy of the energy system, and provides strong support for the stable operation of the smart grid.

[0004] To achieve the above purpose, the present invention provides the following solutions:

[0005] A scheduling method applied to a multi - energy microgrid, comprising:

[0006] Obtain the output data and meteorological information of wind power and photovoltaic power, input the output data and the meteorological information into a load prediction model, and obtain the net load prediction curve of wind power and photovoltaic power; the load prediction model is obtained by training a support vector machine using a training set;

[0007] According to the net load prediction curve, combined with the minimum start - stop time and ramp rate constraints of a gas turbine and a fuel cell, obtain the optimal start - stop plan and output scheduling plan of the gas turbine and the fuel cell;

[0008] Obtain the charge and discharge status of the energy storage device, and based on the charge and discharge status, establish a charge and discharge optimization model with the goal of minimizing the scheduling cost, and obtain the charge and discharge strategy of the energy storage device;

[0009] Integrate the optimal start-stop plan and output scheduling scheme of the gas turbine and fuel cell and the charge and discharge strategy of the energy storage device to form a complete energy scheduling scheme.

[0010] Optionally, obtaining the start-stop plans of the gas turbine and the fuel cell includes:

[0011] According to the net load prediction curve, adopt a mixed integer programming algorithm to establish a combined optimization model for the operation of the gas turbine and the fuel cell. The combined optimization model includes: decision variables and an objective function; the decision variables are the start-stop status and output levels of the gas turbine and fuel cell in each time period, and the objective function is to minimize the power generation cost;

[0012] In the combined optimization model, introduce the minimum start-stop time constraints of the gas turbine and fuel cell, that is, the continuous operation or shutdown time of the equipment shall not be less than the set minimum time threshold, and introduce the ramp rate constraints of the gas turbine and fuel cell, that is, the rate of change of the equipment output shall not exceed the set maximum ramp rate threshold;

[0013] Use the branch and bound algorithm to solve the combined optimization model to obtain the optimal start-stop plan and output scheduling scheme of the gas turbine and fuel cell in each time period with the goal of minimizing the power generation cost.

[0014] Optionally, obtaining the start-stop status and output levels of the gas turbine and the fuel cell in each time period includes:

[0015] Power balance constraint:

[0016] P gt,t +P fc,t +P re,t =L t ;

[0017] where, L t is the net load in time period t, P gt,t is the output of the gas turbine in time period t, P fc,t is the output of the fuel cell in time period t, P re,t is the output of renewable energy in time period t;

[0018] Equipment output constraint:

[0019] P min,gt ·u gt,t ≤P gt,t ≤P max,gt ·ugt,t ;

[0020] P min,fc ·u fc,t ≤P fc,t ≤P max,fc ·u fc,t ;

[0021] Among them, P min,gt is the minimum output of the gas turbine, u gt,t is the start-stop state of the gas turbine at time period t, P gt,t is the output of the gas turbine at time period t, P max,gt is the maximum output of the gas turbine, P min,fc is the minimum output of the fuel cell, u fc,t is the start-stop state of the fuel cell at time period t, P fc,t is the output of the fuel cell at time period t, P max,fc is the maximum output of the fuel cell.

[0022] Optionally, obtaining the objective function includes:

[0023]

[0024] Among them, A is the objective function, T is the total number of time periods, C gt,t is the unit power generation cost of the gas turbine at time period t, P gt,t is the output of the gas turbine at time period t, C fc,t is the unit power generation cost of the fuel cell at time period t, P fc,t is the output of the fuel cell at time period t, C start,gt is the start-stop cost of the gas turbine, u gt,t is the start-stop state of the gas turbine at time period t, u gt,t-1 is the start-stop state of the gas turbine at time period t - 1, C start,fc is the start-stop cost of the fuel cell, u fc,t-1 is the start-stop state of the fuel cell at time period t - 1.

[0025] Optionally, obtaining the minimum start-stop time constraint and ramp rate constraint of the gas turbine and the fuel cell includes:

[0026] Obtaining the minimum start-stop time constraint of the gas turbine:

[0027]

[0028] Among them, T min,gt -1 is the minimum continuous operation time of the gas turbine, ugt,t′ is the state of the gas turbine at different time periods;

[0029] Obtain the minimum start-stop time constraint of the fuel cell:

[0030]

[0031] where T min,fc is the minimum continuous operation time of the fuel cell, and u fc,t′ is the state of the fuel cell at different time periods;

[0032] Obtain the ramp rate constraint of the gas turbine and the fuel cell:

[0033]

[0034] where P gt,t is the output of the gas turbine at time period t, P gt,t-1 is the output of the gas turbine at time period t - 1, R gt is the maximum ramp rate of the gas turbine, P fc,t is the output of the fuel cell at time period t, Pf c,t-1 is the output of the fuel cell at time period t - 1, and R fc is the maximum ramp rate of the fuel cell.

[0035] Optionally, obtaining the charge-discharge strategy of the energy storage device includes:

[0036] Obtain the charge-discharge state of the energy storage device; the charge-discharge state includes: capacity and charge-discharge efficiency;

[0037] According to the charge-discharge state, establish a state transition model of the energy storage device to describe the change of the charge-discharge state of the energy storage device in different time periods;

[0038] Obtain the real-time electricity price information and load demand data of the power grid, calculate the real-time electricity price information and the load demand data using a dynamic programming algorithm, obtain the operating cost under different charge-discharge decisions in each time period, and based on the operating cost and the state transition model of the energy storage device, obtain the minimum total cost from time period t to T;

[0039] Starting from time period T, recursively backward to time period t, determine the optimal charge-discharge power for each time period, and generate the charge-discharge strategy of the energy storage device according to the optimal charge-discharge power for each time period.

[0040] Optionally, establishing the state transition model of the energy storage device includes:

[0041]

[0042] where E t+1 is the remaining power of the energy storage device at the end of time period t + 1, and E tis the remaining power of the energy storage device at the end of time period t, P c,t is the charging power during time period t, η c is the charging efficiency, η d is the discharging efficiency.

[0043] Optionally, obtaining the minimum total cost includes:

[0044]

[0045] wherein, V t (E t ) is the minimum total cost during time period t, C elec,t is the grid electricity price during time period t, C depr,t is the depreciation cost of the energy storage device during time period t.

[0046] Optionally, determining the optimal charging and discharging power for each time period includes:

[0047]

[0048] wherein, is the optimal charging and discharging power for each time period, argmin Pc,t,Pd,t is to find the combination of charging power and discharging power that minimizes the total cost during time period t.

[0049] The beneficial effects of the present invention are:

[0050] By integrating the historical output of wind power and photovoltaic power, weather forecasts, and load forecasts, the present invention generates a net load curve for the next 24 hours, and uses a mixed-integer programming algorithm to optimize the start-stop plans of gas turbines and fuel cells. At the same time, a dynamic programming algorithm is used to optimize the charging and discharging strategies of energy storage devices, and integrate them with the scheduling plans of traditional power generation devices to form a complete energy scheduling plan. The present invention also introduces a rolling optimization mechanism, dynamically adjusts the scheduling plan according to the real-time collected data, and continuously updates the prediction model and optimization algorithm parameters to continuously improve the prediction accuracy and optimization effect, realizes the coordinated optimization scheduling of multiple energy devices, effectively improves the operation efficiency and economy of the energy system, and provides strong support for the stable operation of the smart grid. Description of the Drawings

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1This is a flowchart of a scheduling method applied to a multi - energy microgrid according to an embodiment of the present invention. Specific embodiments

[0053] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0054] To make the above - mentioned objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0055] As Figure 1 shown, this embodiment discloses a scheduling method applied to a multi - energy microgrid, including:

[0056] Obtain the output data and meteorological information of wind power and photovoltaic power, input the output data and meteorological information into the load prediction model, and obtain the net load prediction curve of wind power and photovoltaic power; the load prediction model is obtained by training a support vector machine using a training set; according to the net load prediction curve, combined with the minimum start - stop time and ramp rate constraints of the gas turbine and fuel cell, obtain the optimal start - stop plan and output scheduling scheme of the gas turbine and fuel cell; obtain the charge - discharge state of energy storage devices, and according to the charge - discharge state, establish a charge - discharge optimization model with the goal of minimizing the scheduling cost, and obtain the charge - discharge strategy of energy storage devices; integrate the optimal start - stop plan and output scheduling scheme of the gas turbine and fuel cell and the charge - discharge strategy of energy storage devices to form a complete energy scheduling scheme.

[0057] S101. Obtain the historical output data and meteorological forecast information of wind power and photovoltaic power, and generate a net load curve for the next 24 hours in combination with the load prediction model.

[0058] Obtain the historical output data of wind power and photovoltaic power generation, and establish the prediction models for wind power and photovoltaic power generation output. Obtain the meteorological forecast information for the next 24 hours, and input the meteorological forecast information into the wind power and photovoltaic power generation output prediction models to obtain the predicted output curves of wind power and photovoltaic power generation for the next 24 hours. Obtain the historical load data and establish the load prediction model. Input the meteorological forecast information for the next 24 hours into the load prediction model to obtain the predicted load curve for the next 24 hours. Integrate the predicted output curves of wind power and photovoltaic power generation for the next 24 hours with the predicted load curve to obtain the predicted net load curve for the next 24 hours. If the predicted net load curve shows a negative value, it is judged that there may be wind and light curtailment phenomena during this period, and corresponding measures need to be taken. According to the predicted net load curve, determine the power generation plans and operation modes for each period of the next 24 hours to ensure the safe and stable operation of the power grid.

[0059] Exemplarily, the prediction of wind power and photovoltaic power output is an important basis for power grid dispatching. First, historical output data needs to be obtained, which usually includes power generation, wind speed, light intensity, etc. Taking wind power as an example, the wind speed and power generation data for each hour in the past year can be collected. Using these data, a prediction model based on machine learning can be established, such as a support vector machine or a neural network. After the model training is completed, by inputting the weather forecast data for the next 24 hours, the wind power output prediction curve can be obtained. The prediction of photovoltaic power generation is similar, but more attention is paid to sunshine duration and cloud cover. For example, the prediction model may find that at noon on a sunny summer day, the output of a 100 MW photovoltaic power station can reach 90 MW, while on a cloudy day it may only be 30 MW. This kind of prediction provides an important reference for power grid dispatching. Load prediction is equally important, and it takes more factors into account, including weather, holidays, economic activities, etc. For example, in the hot afternoon on a working day, the use of air conditioners leads to a peak in electricity consumption, and the load may be 30% higher than usual. The load prediction model needs to consider these complex factors. By combining the new energy output prediction and the load prediction, the net load prediction curve can be obtained. This curve reflects the load that traditional power generation needs to bear. If the net load is negative at a certain time period, it means that the new energy generation exceeds the electricity demand, and it may be necessary to limit the new energy generation or increase energy storage to balance the power grid. For example, assume that at 3 am in spring in a certain area, the predicted total load is 5000 MW, the wind power output is 2000 MW, and the photovoltaic output (due to night) is 0 MW. At this time, the net load is 3000 MW, and the traditional power source needs to provide this part of the power. However, if at noon, the total load is 8000 MW, the wind power output is 3000 MW, and the photovoltaic output is 6000 MW, the net load will become -1000 MW, which means that the new energy generation is excessive. In this case, the power grid dispatching needs to take measures. Excessive power can be absorbed by adjusting the output of traditional power sources, starting pumped-storage power stations, transmitting power to neighboring power grids, etc. If these measures still cannot completely absorb the excess power, it may be necessary to limit part of the new energy generation. Through this precise prediction and flexible dispatching, the utilization of new energy can be maximized while ensuring the safe and stable operation of the power grid. This not only improves the energy utilization efficiency but also helps to reduce carbon emissions, which is of great significance for achieving the "dual carbon" goal. With the continuous progress of prediction technology and the increasing richness of dispatching means, the power grid will be able to better adapt to the challenges of high-proportion new energy access and promote the clean transformation of the energy structure.

[0060] Further, obtaining the start-stop plans of the gas turbine and fuel cell includes: according to the net load prediction curve, using the mixed integer programming algorithm to establish a combined optimization model for the operation of the gas turbine and fuel cell. The combined optimization model includes: decision variables and an objective function. The decision variables are the start-stop states and output levels of the gas turbine and fuel cell in each time period, and the objective function is to minimize the power generation cost. In the combined optimization model, introduce the minimum start-stop time constraints of the gas turbine and fuel cell, that is, the continuous operation or shutdown time of the equipment shall not be less than the set minimum time threshold, and introduce the ramp rate constraints of the gas turbine and fuel cell, that is, the rate of change of the equipment output shall not exceed the set maximum ramp rate threshold. Use the branch and bound algorithm to solve the combined optimization model to obtain the optimal start-stop plans and output scheduling schemes of the gas turbine and fuel cell in each time period with the goal of minimizing the power generation cost.

[0061] S102. According to the change trend of the net load curve, use the mixed integer programming algorithm to solve the start-stop plans of the gas turbine and fuel cell, satisfying the minimum start-stop time and ramp rate constraints of the equipment.

[0062] According to the predicted net load curve, use the mixed integer programming algorithm to establish an optimization model for the combined operation of the gas turbine and fuel cell. The decision variables are the start-stop states and output levels of the gas turbine and fuel cell in each time period, and the objective function is to minimize the power generation cost. In the optimization model, introduce the minimum start-stop time constraints of the gas turbine and fuel cell, that is, the continuous operation or shutdown time of the equipment shall not be less than the set minimum time threshold. Through the setting of constraint conditions, ensure the safety and reliability of the equipment start-stop process. In the optimization model, introduce the ramp rate constraints of the gas turbine and fuel cell, that is, the rate of change of the equipment output shall not exceed the set maximum ramp rate threshold. Through the setting of constraint conditions, ensure the smoothness and controllability of the equipment output adjustment process. Use the branch and bound algorithm to solve the mixed integer programming model to obtain the optimal start-stop plans and output scheduling schemes of the gas turbine and fuel cell in each time period, achieving the goal of minimizing the power generation cost. According to the results of the optimization solution, generate the start-stop instruction sequences of the gas turbine and fuel cell, and send the start-stop instructions to the gas turbine and fuel cell equipment through the automatic control system to realize the automatic start-stop control of the equipment. During the operation of the equipment, collect the operation parameters of the gas turbine and fuel cell in real time, monitor the actual operation status of the equipment. If it is found that the operation parameters of the equipment deviate from the optimized scheduling scheme, trigger an alarm message to notify the duty personnel for inspection and handling to ensure the safe and reliable operation of the equipment according to the optimized scheduling scheme.

[0063] Exemplarily, the prediction of the net load curve is a key link in power system dispatching. Time series analysis methods can capture trends, seasonality, and periodic characteristics in historical data, thereby accurately predicting future net load changes. For example, the autoregressive integrated moving average model (ARIMA) can effectively handle non-stationary time series data. By performing differencing operations, the non-stationary sequence is transformed into a stationary sequence, and then modeling and prediction are carried out. Mixed-integer programming is an effective tool for solving the combined optimization problem of gas turbines and fuel cells. The decision variables include the start-stop status (0-1 integer variable) and the output level (continuous variable) of the equipment. The objective function usually considers factors such as fuel cost and start-stop cost. For example, a power plant has 3 gas turbines and 2 sets of fuel cells, and an optimal operation plan needs to be arranged within 24 hours. By establishing a mathematical model, the optimal operation status and output level of each device per hour can be obtained. The minimum start-stop time constraint is an important condition to ensure the safe operation of the equipment. Frequent start-stop will exacerbate equipment wear and reduce service life. Assuming that the minimum start time of the gas turbine is 2 hours and the minimum stop time is 3 hours, corresponding constraint conditions need to be added to the optimization model to ensure that the continuous operation or stop time of the equipment is not less than these thresholds. The ramp rate constraint reflects the dynamic characteristics of the equipment's output regulation. For gas turbines, their ramp rate is usually between 1% and 5% of the rated power per minute. For example, for a 100MW gas turbine, its maximum ramp rate may be 3MW / minute. In the optimization model, it is necessary to limit the output change between adjacent two time periods not to exceed the ramp rate limit value to ensure the safe and stable operation of the equipment. The branch and bound algorithm is a commonly used method for solving mixed-integer programming problems. It continuously branches (fixing the integer variables to different values) and bounds (calculating the solution of the relaxed problem) to narrow the feasible solution space and finally find the global optimal solution. In practical applications, mature optimization solvers (such as CPLEX, Gurobi, etc.) can be used to implement the branch and bound algorithm to improve the solving efficiency. The automated execution of the optimization results is the key to realizing intelligent dispatching. By converting the obtained start-stop instructions and output dispatching plan into control instructions in a standard format and sending them to each device through the SCADA system or other automated control systems, the automated control of the equipment can be achieved. This not only improves the dispatching efficiency but also reduces the risk of human operation errors. The real-time monitoring and alarm mechanism is an important means to ensure the effective execution of the optimization plan. By collecting the operation parameters of gas turbines and fuel cells (such as output, efficiency, temperature, etc.) and comparing them with the optimized dispatching plan in real time, the operation deviation of the equipment can be detected in time. By setting reasonable alarm thresholds, when the deviation exceeds the preset range, the system automatically triggers an alarm message to remind the on-duty personnel to check and handle it, so as to ensure that the equipment always operates according to the optimal plan.

[0064] Further, obtaining the start-stop status and output level of the gas turbine and fuel cell in each time period includes:

[0065] Power balance constraint:

[0066] P gt,t +P fc,t +P re,t = L t ;

[0067] where L t is the net load at time period t, P gt,t is the output of the gas turbine at time period t, P fc,t is the output of the fuel cell at time period t, P re,t is the output of renewable energy at time period t;

[0068] Equipment output constraint:

[0069] P min,gt ·u gt,t ≤ P gt,t ≤ P max,gt ·u gt,t ;

[0070] P min,fc ·u fc,t ≤ P fc,t ≤ P max,fc ·u fc,t ;

[0071] where P min,gt is the minimum output of the gas turbine, u gt,t is the start-stop state of the gas turbine at time period t, P gt,t is the output of the gas turbine at time period t, P max,gt is the maximum output of the gas turbine, P min,fc is the minimum output of the fuel cell, u fc,t is the start-stop state of the fuel cell at time period t, P fc,t is the output of the fuel cell at time period t, P max,fc is the maximum output of the fuel cell.

[0072] Furthermore, obtaining the objective function includes:

[0073]

[0074] where A is the objective function, T is the total number of time periods, C gt,t is the unit power generation cost of the gas turbine at time period t, P gt,t is the output of the gas turbine at time period t, C fc,t is the unit power generation cost of the fuel cell at time period t, P fc,t is the output of the fuel cell at time period t, C start,gt is the start-stop cost of the gas turbine, ugt,t is the start-stop state of the gas turbine at time period t, u gt,t-1 is the start-stop state of the gas turbine at time period t - 1, C start,fc is the start-stop cost of the fuel cell, u fc,t-1 is the start-stop state of the fuel cell at time period t - 1.

[0075] Furthermore, obtaining the minimum start-stop time constraints and ramp rate constraints of the gas turbine and the fuel cell includes:

[0076] Obtaining the minimum start-stop time constraint of the gas turbine:

[0077]

[0078] where T min,gt -1 is the minimum continuous operation time of the gas turbine, ugt,t′ is the state of the gas turbine at different time periods;

[0079] Obtaining the minimum start-stop time constraint of the fuel cell:

[0080]

[0081] where T min,fc is the minimum continuous operation time of the fuel cell, u fc,t′ is the state of the fuel cell at different time periods;

[0082] Obtaining the ramp rate constraints of the gas turbine and the fuel cell:

[0083]

[0084] where P gt,t is the output of the gas turbine at time period t, P gt,t-1 is the output of the gas turbine at time period t - 1, R gt is the maximum ramp rate of the gas turbine, P fc,t is the output of the fuel cell at time period t, Pf c,t-1 is the output of the fuel cell at time period t - 1, R fc is the maximum ramp rate of the fuel cell.

[0085] Further, obtaining the charge and discharge strategy of the energy storage device includes: obtaining the charge and discharge state of the energy storage device; the charge and discharge state includes: capacity and charge and discharge efficiency; according to the charge and discharge state, establishing a state transition model of the energy storage device to describe the change of the charge and discharge state of the energy storage device in different time periods; obtaining the real-time electricity price information and load demand data of the power grid, using the dynamic programming algorithm to calculate the real-time electricity price information and load demand data, obtaining the operating cost under different charge and discharge decisions in each time period, and based on the operating cost and combining the state transition model of the energy storage device, obtaining the minimum total cost from time period t to T; starting from time period T, recursively reverse to time period t, determining the optimal charge and discharge power for each time period, and generating the charge and discharge strategy of the energy storage device according to the optimal charge and discharge power for each time period.

[0086] Further, establishing the state transition model of the energy storage device includes:

[0087]

[0088] where E t+1 is the remaining power of the energy storage device at the end of time period t + 1, E t is the remaining power of the energy storage device at the end of time period t, P c,t is the charging power in time period t, η c is the charging efficiency, η d is the discharging efficiency.

[0089] Further, obtaining the minimum total cost includes:

[0090]

[0091] where V t (E t ) is the minimum total cost in time period t, C elec,t is the electricity price of the power grid in time period t, C depr,t is the depreciation cost of the energy storage device in time period t.

[0092] Further, determining the optimal charge and discharge power for each time period includes:

[0093]

[0094] where is the optimal charge and discharge power for each time period, argmin Pc,t,Pd,t is to find the combination of charging power and discharging power that minimizes the total cost in time period t.

[0095] S103. For the energy storage device, establish a charge and discharge optimization model based on dynamic programming, with the goal of minimizing the system operation cost, and determine the charge and discharge power of the energy storage device.

[0096] According to parameters such as the capacity and charge-discharge efficiency of the energy storage device, establish the state transition equation of the energy storage device to describe the change of the charge-discharge state of the energy storage device in different time periods; obtain the real-time electricity price information and load demand data of the power grid as the input conditions of the dynamic programming model; adopt the dynamic programming algorithm to calculate the system operation cost under different charge-discharge decisions in each time period, including electricity cost expenditure and the depreciation cost of the energy storage device; through the forward recursion and backward solution of dynamic programming, obtain the optimal charge-discharge strategy within the entire operation cycle to minimize the total system operation cost; determine the charge-discharge power of the energy storage device in each time period according to the optimal charge-discharge strategy to generate a detailed charge-discharge plan; convert the optimal charge-discharge strategy into the control instruction of the energy storage device to adjust the charge-discharge power of the energy storage device in real time to ensure the execution of the strategy; continuously monitor the operation state of the energy storage device and the power grid situation, and dynamically adjust the charge-discharge strategy according to the actual situation to ensure the safe and economic operation of the system.

[0097] Exemplarily, the state transition equation of the energy storage device is a mathematical model that describes the changes in the charge and discharge states of the energy storage device in different time periods. The real-time electricity price information of the power grid and the load demand data are the key inputs of the dynamic programming model. Taking an industrial park as an example, its peak and valley electricity prices are 1.2 yuan / kWh and 0.5 yuan / kWh respectively, and the load demand is relatively high during the day on weekdays and relatively low at night and on weekends. These data provide a basis for optimizing the charge and discharge strategy. The dynamic programming algorithm finds the optimal strategy by calculating the system operation cost under different charge and discharge decisions for each time period. Suppose the electricity price in a certain period is 0.8 yuan / kWh, the cost of charging 10 kWh of the energy storage device is 8 yuan, plus a depreciation cost of 0.5 yuan, and the total cost is 8.5 yuan. By comparing the costs of different decisions, it can be determined whether to charge in this period. Determining the optimal charge and discharge strategy requires considering the entire operation cycle. For example, in a 24-hour cycle, the possible optimal strategy is to charge at night during the low valley electricity price and discharge during the day at the high peak electricity price, so as to maximize the economic benefits. Specifically for each hour, a detailed plan such as "charge from 22:00 to 06:00, standby from 06:00 to 10:00, discharge from 10:00 to 14:00" may be formed. Transforming the optimal strategy into control instructions is the key to realizing automatic operation. For example, when the strategy indicates to start charging at a certain moment, the control system will send an instruction to start charging to the energy storage device and adjust the charging power according to the plan. This precise control ensures the effective execution of the strategy. Dynamically adjusting the charge and discharge strategy is a necessary means to cope with changes in the actual situation. For example, when there is a large deviation between the actual load and the prediction, the system may need to advance or delay the discharge time. Another example is that when a sudden power grid fault occurs, the energy storage system may need to immediately switch to the emergency power supply mode and temporarily put aside the original economic operation strategy. This flexibility ensures the safe and economic operation of the system in various situations. Through this optimized scheduling method, the energy storage system can effectively smooth the load fluctuation and reduce the electricity cost. For example, after a certain factory adopts this method, it can save about 50,000 yuan in electricity bills per month, improve the power supply reliability at the same time, and reduce production interruptions caused by power fluctuations. This not only brings direct economic benefits, but also improves the overall energy utilization efficiency, laying a foundation for realizing a more extensive smart grid and renewable energy integration.

[0098] S104. Integrate the charge and discharge strategy of the energy storage device with the start-stop plans of the gas turbine and the fuel cell to form a complete energy scheduling plan.

[0099] Obtain the real-time operation data and prediction data of each device according to the charge and discharge status of the energy storage device and the operation status of the gas turbine and fuel cell. Analyze the historical operation data of each device through machine learning algorithms to establish a charge and discharge strategy model for the energy storage device and start-stop plan models for the gas turbine and fuel cell. Integrate the charge and discharge strategy model of the energy storage device and the start-stop plan models of the gas turbine and fuel cell to form a comprehensive energy dispatch model. According to the comprehensive energy dispatch model, combined with the current real-time operation status and prediction data of each device, use a heuristic optimization algorithm to generate an energy dispatch plan for each time period. If the generated energy dispatch plan does not meet the system operation constraint conditions, return to the previous step, adjust the parameters of the optimization algorithm, and regenerate the dispatch plan. Convert the generated energy dispatch plan into control instructions and send them to the control systems of the energy storage device, gas turbine, and fuel cell to achieve coordinated control of each device. According to the actual operation conditions of each device after executing the dispatch instructions, make real-time corrections and updates to the comprehensive energy dispatch model to improve the reliability and economy of subsequent dispatch plans.

[0100] Exemplarily, the establishment and application of the integrated energy scheduling model are the key to realizing the coordinated operation of multiple energy devices. First, obtain the real-time operation data and prediction data of energy storage devices, gas turbines, and fuel cells. These data include the charge and discharge status, power generation, efficiency, etc. of the devices. For example, the charge status of the energy storage device may be 80%, and it is expected that the electricity price will be low in the next 4 hours, which is suitable for continued charging; the current power generation of the gas turbine is 5 MW, and it is expected that the load demand will increase in the next 2 hours; the fuel cell is in standby state and can be started at any time. Analyze the historical operation data through machine learning algorithms to establish the operation models of each device. For energy storage devices, the support vector machine (SVM) algorithm can be used to predict the optimal charge and discharge time and power according to historical charge and discharge data and electricity price data. The start-stop plan models of gas turbines and fuel cells can use decision tree algorithms to predict the best start-stop time considering factors such as load demand and fuel cost. Integrate the device models to form an integrated energy scheduling model, which takes into account the mutual influence between devices and the overall system benefits. For example, when the energy storage device discharges, the power generation of the gas turbine can be reduced, thereby reducing fuel consumption and operating costs. Based on this model, combined with real-time data and prediction data, heuristic optimization methods such as genetic algorithms are used to generate a scheduling plan. Suppose that in a certain period, the energy storage device discharges 2 MW, the gas turbine generates 3 MW, and the fuel cell generates 1 MW, which can meet the load demand of 6 MW while minimizing the system operation cost. If the generated scheduling plan does not meet the constraint conditions, such as the device output exceeds the rated power or the system power balance is broken, the parameters of the optimization algorithm need to be adjusted to regenerate the plan. For example, a penalty factor can be increased to increase the cost of violating the constraints and guide the algorithm to generate a more reasonable plan. Convert the finally determined scheduling plan into control instructions, such as setting the discharge power of the energy storage device to 2 MW, starting the gas turbine and setting the output power to 3 MW, and starting the fuel cell and setting the output power to 1 MW. These instructions are sent to the control systems of each device through the communication network to achieve coordinated control. During the execution of the scheduling plan, continuously monitor the actual operation of each device. If there is a deviation, such as the actual discharge power of the energy storage device is only 1.8 MW, the scheduling model needs to be corrected in real time. Online learning algorithms, such as incremental learning methods, can be used to update the model parameters according to the new operation data to improve the accuracy and reliability of subsequent scheduling plans. This dynamic adjustment mechanism can adapt to changes in device performance and external environment to ensure the continuous and efficient operation of the system.

[0101] According to the integrated energy scheduling model, combined with the real-time operation status and prediction data of current devices, a heuristic optimization algorithm is used to generate the energy scheduling plan for each period.

[0102] According to the real-time operation status data of each device in the integrated energy system, obtain the current operating condition parameters of the device, including information such as the load and efficiency of the device, and form a device state feature vector. Train the historical operation data through machine learning algorithms to establish a prediction model of the device operation status and energy consumption. According to the current device state feature vector, predict the energy consumption of the device in the next period of time. Model the energy scheduling problem as a multi-objective optimization problem, where the objective functions include minimizing the energy cost and maximizing the energy utilization efficiency, etc. At the same time, consider conditions such as device operation constraints and energy balance to form an integrated energy scheduling optimization model. Use heuristic optimization algorithms, such as genetic algorithms and particle swarm algorithms, to solve the integrated energy scheduling optimization model and obtain the optimal energy scheduling plan for each period, including the output of each device and the charge and discharge plan of energy storage. According to the optimal scheduling plan, combined with the real-time operation status and prediction data of the device, perform rolling optimization on the scheduling plan, and adjust the operation parameters of each device in real time to ensure the executability and optimality of the scheduling plan. During the execution of the scheduling, continuously monitor the device operation status and energy balance. If there are deviations, trigger the re-optimization of the scheduling plan to ensure the safe and stable operation of the integrated energy system. Send the optimized scheduling plan to the control systems of each device, implement the scheduling plan through the automatic control system, and at the same time feedback the scheduling execution situation to the integrated energy management system to form a closed-loop control.

[0103] Exemplarily, the real-time scheduling optimization of an integrated energy system is a complex process involving multiple key steps and technologies. First of all, obtaining the real-time operating status data of equipment is the foundation. For example, for a gas turbine, parameters such as its current load, fuel consumption rate, and exhaust gas temperature can be collected; for energy storage equipment, information such as charge and discharge power and remaining capacity can be obtained. These data constitute the equipment status feature vector, providing a basis for subsequent analysis. Using machine learning algorithms to predict equipment energy consumption is the key to improving scheduling efficiency. Taking the support vector machine (SVM) as an example, a model can be trained using historical operation data. The inputs include features such as time, ambient temperature, and load demand, and the output is the predicted value of equipment energy consumption. This method can capture the non-linear relationship between equipment energy consumption and changes in time and operating conditions, improving the prediction accuracy. Modeling the energy scheduling problem as a multi-objective optimization problem is an effective way to maximize the comprehensive benefits. The objective function can include minimizing energy costs, minimizing carbon emissions, maximizing equipment life, etc. The constraint conditions include equipment operation limitations (such as minimum start-up time), energy balance, etc. For example, in the integrated energy system of an industrial park, the objective function can be set as: minimizing the sum of the power purchase cost from the power grid and the natural gas procurement cost, while maximizing the utilization rate of renewable energy. Using heuristic optimization algorithms to solve the scheduling model is an effective method for dealing with large-scale complex problems. Taking the genetic algorithm as an example, the output of equipment at each time period can be encoded as a chromosome, and the quality of the solution can be continuously optimized through operations such as crossover and mutation. In practical applications, a fitness function can be designed in combination with the characteristics of the problem, such as considering the weighted sum of factors such as energy cost and environmental impact. Rolling optimization is an important means to ensure the real-time performance and robustness of the scheduling plan. For example, the scheduling plan is updated once an hour, and the latest load prediction and equipment status information are used to re-optimize the scheduling plan for the next 24 hours. This method can effectively cope with uncertain factors such as load prediction errors and equipment failures. Real-time monitoring and dynamic adjustment are the keys to ensuring the safe and stable operation of the system. For example, when the output of photovoltaic power generation suddenly drops, the system can quickly increase the output of the gas turbine or start the energy storage equipment to discharge, maintaining the power balance of the system. This rapid response mechanism can effectively prevent the occurrence of large-scale power outages. Finally, executing the optimized scheduling plan through an automated control system and forming a closed-loop control is an important link in realizing intelligent scheduling. For example, a distributed control system (DCS) can be used to send scheduling instructions to each device, and at the same time, the real-time operation data of the device is collected and fed back to the energy management system to achieve the precise execution and timely adjustment of the scheduling plan. This closed-loop control mechanism can significantly improve the operation efficiency and reliability of the system, laying a foundation for the intelligent operation of the integrated energy system.

[0104] S105. According to the real-time collected wind power and photovoltaic output data and load change information, adopt the rolling optimization method to dynamically adjust the energy scheduling plan to obtain the optimized energy scheduling plan.

[0105] According to the pre-established historical output data models of the wind farm and the photovoltaic power station, combined with the current real-time collected meteorological data such as wind speed and irradiance, the support vector machine regression algorithm is used to predict the wind power and photovoltaic output curves in the future for a period of time. Obtain the real-time load data of the current power grid. According to the load prediction model, the time series analysis method is used to predict the load change curve in the future for a period of time. Input the predicted wind power output, photovoltaic output and load prediction data into the pre-established energy optimization scheduling model. This model comprehensively considers multiple objectives such as power generation cost, power balance, and equipment operation constraints, and uses the mixed integer programming algorithm to solve, obtaining an initial energy optimization scheduling plan. During the actual operation process, continuously collect the real-time output data of wind power and photovoltaic power and the power grid load data. If it is found that the deviation between the actual data and the predicted data exceeds the preset threshold, the rolling optimization process is triggered. Input the real-time data into the optimization scheduling model, and at the same time consider the influence of the executed scheduling plan. Using the rolling horizon optimization method, based on the optimization results of the previous time period, the scheduling plan for the subsequent time period is dynamically adjusted to obtain an updated optimization scheduling plan. If the optimized scheduling plan meets the real-time balance and equipment operation constraints, the plan is sent to each power generation equipment and load control equipment for execution; if it does not meet the constraints, return to step 5 to re-perform rolling optimization until all constraints are met. During the process of energy scheduling optimization, continuously monitor the change trends of wind power, photovoltaic output and load data. When it is found that the data change exceeds the preset range, the rolling optimization process is triggered in time to ensure the dynamic matching of the scheduling plan and the actual operation conditions, and improve the energy utilization efficiency and the reliability of the power grid operation.

[0106] Exemplarily, the output prediction of wind farms and photovoltaic power plants is a key link in energy dispatch optimization. The support vector machine regression algorithm can effectively capture the impact of meteorological factors such as wind speed and irradiance on power generation by learning the non-linear relationships in historical data. For example, a certain wind farm uses this algorithm and combines real-time wind speed data to predict the output curve for the next 24 hours, with a prediction accuracy of over 90%. This provides a reliable data basis for subsequent dispatch optimization. Load prediction is equally crucial. The time series analysis method takes into account the periodic and trend characteristics of the load and can accurately grasp the changing pattern of electricity demand. A certain regional power grid uses this method, combines historical load data with factors such as weather and holidays, and achieves hourly load prediction with an average error controlled within 3%. This helps the dispatch department make preparations for power supply-demand balance in advance. The energy optimization dispatch model is the core of the entire system. The mixed integer programming algorithm can handle both continuous variables (such as power generation) and discrete variables (such as unit start-stop status) simultaneously, and find the optimal dispatch plan under the premise of meeting various constraints. A certain integrated energy system applies this algorithm, comprehensively considers factors such as fuel cost, environmental protection indicators, and equipment life, and achieves a 5% reduction in annual power generation cost while reducing carbon emissions by 20,000 tons. Rolling optimization is the key technology to ensure the real-time performance and adaptability of the dispatch plan. When there is a large deviation between the actual operation data and the predicted data, the system will automatically trigger the optimization process. For example, due to cloud cover changes, the actual output of a certain photovoltaic power plant is 30% lower than the predicted value. The rolling optimization system immediately adjusts the output of thermal power units in subsequent periods and increases the discharge of energy storage devices, effectively maintaining the power supply-demand balance of the power grid. This dynamic adjustment mechanism greatly improves the system's ability to respond to uncertainties. Continuous monitoring and timely optimization are the guarantees to ensure the effectiveness of the dispatch plan. By setting reasonable trigger thresholds, the system can timely initiate the optimization process when there are significant changes in wind power, photovoltaic output, or load. A certain smart grid system uses this mechanism and successfully copes with the drastic fluctuations in wind power output during an extreme weather event, avoiding large-scale power outages. This real-time response mechanism greatly enhances the safety and reliability of the power grid. The core of the entire energy dispatch optimization system lies in its closed-loop feedback mechanism. Through continuous data collection, analysis, optimization, and execution, the system can continuously learn and improve, adapting to the complex and changing energy environment. This intelligent dispatch method not only improves energy utilization efficiency but also provides technical support for the grid connection operation of large-scale renewable energy, promoting the clean transformation of the energy structure.

[0107] S106. Use the real-time collected data to update the parameters of the wind power and photovoltaic output prediction models, load prediction models, and optimization algorithms, improve the prediction accuracy and optimization effect, and provide support for the next round of energy optimization dispatch.

[0108] Real-time monitoring data of wind farms, photovoltaic power stations and load sides are obtained, and the data are preprocessed to remove outliers and missing values ​​to obtain normalized time series data. According to the preprocessed real-time data, the recursive least squares method is used to update the parameters of the wind power output prediction model online, so that the model can adapt to changes in influencing factors such as wind speed and wind direction, and improve the accuracy of wind power output prediction. Similarly, using real-time data, the parameters of the photovoltaic output prediction model are adaptively adjusted through the Kalman filter algorithm, so that the model can track changes in factors such as solar radiation intensity and temperature, and improve the accuracy of photovoltaic output prediction. For the load prediction model, the support vector machine algorithm with adaptive weights is used to dynamically adjust the model weights according to real-time load data, improve the accuracy of load prediction, and provide reliable load prediction results for optimized scheduling. While updating the parameters of each prediction model, the dispatchable capacity and load demand of wind farms and photovoltaic power stations are calculated according to real-time data as the input of the optimized scheduling algorithm. The improved particle swarm optimization algorithm is used to solve the energy optimization scheduling problem with the updated prediction results and dispatchable capacity as constraints, and the optimal output plan of wind farms and photovoltaic power stations in the next round of scheduling cycle is obtained. The optimized output plan is sent to each wind farm and photovoltaic power station to guide their energy production, and the optimization results are fed back to the prediction model to further improve the prediction and optimization effects, forming a closed-loop optimization.

[0109] Exemplarily, in an energy dispatch optimization system, data preprocessing is a crucial step. Taking a wind farm as an example, real-time monitoring data may contain outliers caused by equipment failures or communication interruptions. By using the moving median method, these outliers can be effectively identified and removed, ensuring the accuracy of subsequent analysis. For example, if there is a value in the wind speed data of a wind farm that significantly deviates within one hour, by comparing the data of the previous and subsequent periods, it can be determined that this value is an outlier and replaced with the average value of adjacent periods. Updating the parameters of the wind power output prediction model is the key to improving the prediction accuracy. The recursive least squares method can adjust the model parameters in real time to adapt to changes in wind conditions. Suppose the relationship between wind speed and power output in a wind farm was originally linear, but with the change of seasons, this relationship may become more complex. Through the recursive least squares method, the model can timely capture this change, adjust the parameters to reflect the new wind speed-power output relationship, thereby improving the prediction accuracy. Photovoltaic power output prediction also requires dynamic adjustment. The Kalman filter algorithm can effectively handle the uncertainties in photovoltaic power generation. For example, in cloudy weather, the solar radiation intensity may change frequently. The Kalman filter algorithm can continuously update the state estimation of the photovoltaic power output prediction model based on real-time observation data, making the prediction results closer to the actual situation. This method is particularly suitable for dealing with short-term changes in light intensity and helps improve the dispatching accuracy of photovoltaic power plants. Optimizing the load prediction model is crucial for balancing supply and demand. The support vector machine algorithm with adaptive weights can dynamically adjust the model weights according to the load characteristics of different periods. For example, the electricity load patterns may vary significantly between weekdays and weekends. This algorithm can identify these patterns and accordingly adjust the weights of the prediction model to make the prediction results more accurate. This method is particularly suitable for dealing with load fluctuations caused by seasonal changes and special events. In the optimized dispatch algorithm, the improved particle swarm optimization algorithm can effectively handle multi-objective optimization problems. For example, when considering both economy and environmental protection, the algorithm can formulate an optimal power output plan for wind farms and photovoltaic power plants. Suppose there are multiple wind farms and photovoltaic power plants in a certain area, the algorithm will calculate the optimal power output level of each power plant according to the predicted wind conditions, light conditions, and grid load demand, which not only meets the load demand but also maximizes the utilization rate of renewable energy. Finally, feeding back the optimization results to the prediction model to form a closed-loop optimization can continuously improve the overall performance of the system. For example, if there is a deviation between the actual effect and the expectation of a certain optimized dispatch, the system will analyze the reasons and use this information to adjust the parameters of the prediction model or the strategies of the optimization algorithm. This mechanism of continuous learning and optimization enables the entire system to adapt to the ever-changing external environment and maintain efficient operation.

[0110] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A scheduling method applied to a multi-energy microgrid, characterized in that: include: Obtaining wind power and photovoltaic power output data and meteorological information, inputting the output data and the meteorological information into a load forecasting model, and obtaining a wind power and photovoltaic net load forecasting curve; the load forecasting model is obtained by training a support vector machine using a training set; According to the net load prediction curve, combined with the minimum start-stop time and ramp rate constraints of the gas turbine and the fuel cell, the optimal start-stop plan and output scheduling scheme of the gas turbine and the fuel cell are obtained; Acquire the charge and discharge status of the energy storage device, establish a charge and discharge optimization model with the goal of minimizing the scheduling cost according to the charge and discharge status, and acquire the charge and discharge strategy of the energy storage device; The optimal start-stop plan and output scheduling plan of the gas turbine and fuel cell and the charging and discharging strategy of the energy storage device are integrated to form a complete energy scheduling plan.

2. The dispatching method applied to a multi-energy microgrid according to claim 1, characterized in that: Obtaining the start and stop plan of the gas turbine and the fuel cell includes: According to the net load prediction curve, a combined optimization model for the operation of the gas turbine and the fuel cell is established by using a mixed integer programming algorithm, wherein the combined optimization model includes: decision variables and an objective function; the decision variables are the start / stop states and output levels of the gas turbine and the fuel cell in each time period, and the objective function is minimizing the power generation cost; In the combined optimization model, the minimum start-stop time constraints of gas turbines and fuel cells are introduced, that is, the continuous operation or shutdown time of the equipment shall not be less than the set minimum time threshold, and the ramp rate constraints of gas turbines and fuel cells are introduced, that is, the rate of change of equipment output shall not exceed the set maximum ramp rate threshold; The combined optimization model is solved by using a branch and bound algorithm to obtain the optimal start and stop plan and output scheduling scheme of the gas turbine and the fuel cell in each time period with the goal of minimizing the power generation cost.

3. The dispatching method applied to a multi-energy microgrid according to claim 2, characterized in that: Obtaining the start / stop status and output level of the gas turbine and the fuel cell in each time period includes: Power balance constraints: P gt,t +P fc,t +P re,t =L t ; Among them, L t is the net load in period t, P gt,t is the output of the gas turbine in period t, P fc,t is the output of the fuel cell in period t, P re,t is the output of renewable energy in time period t; Equipment output constraints: P min,gt ·in gt,t ≤P gt,t ≤P max,gt ·in gt,t ; P min,fc ·in fc,t ≤P fc,t ≤P max,fc ·in fc,t ; Among them, P min,gt is the minimum output of the gas turbine, u gt,t is the start and stop state of the gas turbine in period t, P gt,t is the output of the gas turbine in period t, P max,gt is the maximum output of the gas turbine, P min,fc is the minimum output of the fuel cell, u fc,t is the start and stop state of the fuel cell in period t, P fc,t is the output of the fuel cell in period t, P max,fc The maximum output of the fuel cell.

4. The dispatching method applied to a multi-energy microgrid according to claim 2, characterized in that: Obtaining the objective function includes: Among them, A is the objective function, T is the total number of time periods, C gt,t is the unit power generation cost of the gas turbine in period t, P gt,t is the output of the gas turbine in period t, C fc,t is the unit power generation cost of the fuel cell in time period t, P fc,t is the output of the fuel cell in period t, C start,gt is the start-up and shutdown cost of the gas turbine, u gt,t is the start and stop state of the gas turbine in period t, u gt,t-1 is the start and stop state of the gas turbine in period t-1, C start,fc is the start-stop cost of the fuel cell, u fc,t-1 is the start and stop state of the fuel cell in time period t-1.

5. The dispatching method applied to a multi-energy microgrid according to claim 2, characterized in that: Obtaining the minimum start-stop time constraints and ramp rate constraints of the gas turbine and the fuel cell includes: Get the minimum start and stop time constraints of the gas turbine: Among them, T min,gt -1 is the minimum continuous operation time of the gas turbine, ugt,t′ is the state of the gas turbine in different time periods; Get the minimum start and stop time constraint of the fuel cell: Among them, T min,fc is the minimum continuous operation time of the fuel cell, u fc,t′ is the state of the fuel cell at different time periods; Get the ramp rate constraints of the gas turbine and the fuel cell: Among them, P gt,t is the output of the gas turbine in period t, P gt,t-1 is the output of the gas turbine at time period t-1, R gt is the maximum ramp rate of the gas turbine, P fc,t is the output of the fuel cell in period t, Pf c,t-1 is the output of the fuel cell in period t-1, R fc is the maximum ramp rate of the fuel cell.

6. The dispatching method for multi-energy microgrid according to claim 4, characterized in that: Obtaining the charging and discharging strategy of the energy storage device includes: Acquiring the charge and discharge state of the energy storage device; the charge and discharge state includes: capacity and charge and discharge efficiency; According to the charging and discharging state, a state transition model of the energy storage device is established to describe the change of the charging and discharging state of the energy storage device in different time periods; Obtaining real-time electricity price information and load demand data of the power grid, calculating the real-time electricity price information and the load demand data using a dynamic programming algorithm, obtaining the operating cost under different charging and discharging decisions in each time period, and obtaining the minimum total cost from time period t to T based on the operating cost combined with the state transition model of the energy storage device; Starting from time period T, the direction is reversed to time period t, the optimal charge and discharge power of each time period is determined, and the charge and discharge strategy of the energy storage device is generated according to the optimal charge and discharge power of each time period.

7. The dispatching method for multi-energy microgrid according to claim 6, characterized in that: Establishing the state transition model of the energy storage device includes: Among them, E t+1 is the remaining power of the energy storage device at the end of period t+1, E t is the remaining power of the energy storage device at the end of period t, P c,t is the charging power in time period t, η c is the charging efficiency, η d is the discharge efficiency.

8. The dispatching method for multi-energy microgrid according to claim 6, characterized in that: Obtaining the minimum total cost includes: Among them, V t (E t ) is the minimum total cost in period t, C elec,t is the grid electricity price in period t, C depr,t is the depreciation cost of the energy storage equipment in period t.

9. The dispatching method for multi-energy microgrid according to claim 6, characterized in that: Determining the optimal charging and discharging power for each period includes: in, is the optimal charge and discharge power for each period, argmin Pc,t,Pd,t To find the combination of charging power and discharging power that minimizes the total cost within time period t.

Images