A multi-time-scale microgrid energy management method

Abstract

The application discloses a kind of multi-time scale micro-grid energy management method, using low-cost router-based networking scheme, simple and easy, without complex field bus can realize communication interaction, combine day-ahead and intra-day optimization, intra-day optimization according to the latest, higher accuracy prediction information is carried out to the day-ahead scheduling scheme Rolling correction, reduce the influence of the uncertainty factor of prediction information, improve the accuracy and reliability of micro-grid scheduling plan, reduce the operation cost of micro-grid.Introduction and main grid power interaction, reduce the fluctuation brought by new energy power generation, further improve the stability of micro-grid, while the real-time electricity price of main grid and the bidding of distributed power generation into grid electricity price make the cost of micro-grid further optimized, in order to avoid the single of optimization target, environmental management cost is included in objective function, realize the comprehensive consideration to environmental benefit and economic benefit.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart grid, in particular to a multi-time-scale micro-grid energy management method. BACKGROUND

[0002] Nowadays, in order to cope with the impact of global warming on human society and ecological system, and solve the problem of energy shortage and environmental pollution of fossil energy, accelerate the development and utilization of renewable energy, and reduce carbon emissions have become the consensus of the world. However, the intermittency and uncertainty of renewable energy have an undeniable impact on the stability of the power grid. Only relying on continuous expansion of the power generation side cannot meet the development needs, and it is also necessary to cooperate with reasonable energy management optimization methods to coordinate and manage various distributed renewable energy and energy storage devices with different performances in the micro-grid, so as to realize the stable operation and optimal scheduling of the micro-grid. Therefore, the research on micro-grid energy management is of great significance.

[0003] At present, the communication networking scheme of micro-grid mostly adopts field bus technology, and there are many kinds of field buses, such as profibus bus, Modbus bus, CanOpen bus, etc. Different field buses cannot be compatible, and the software and hardware can usually only use a product of one company. There is a lack of interoperability and interchangeability between products of different manufacturers, and the integrability is poor. For a large-scale distributed system, a large number of I / O cables and laying construction not only increase the cost, but also increase the unreliability of the system, which cannot meet the growing requirements of the control network. The existing energy management technology is mostly through optimizing controllable power supply and energy storage to manage energy, but the randomness and volatility of clean energy (such as photovoltaic power generation) bring instability to the power grid, so it is necessary to equip larger energy storage or large power grid support. The existing energy management method mostly collects the day-ahead information before the dispatching day, only formulates a 24-hour dispatching plan with 1-hour cycle, but with the increase of time scale, the influence of uncertainty factors of load prediction and photovoltaic output prediction is increasing, and the accuracy of prediction will gradually decrease with the increase of time scale, and the day-ahead plan of 24 hours is stage type, and the mutation is large, so the actual operation cannot be completely executed according to the plan, so it is necessary to have optimization results of shorter time scale.

[0004] In order to solve the above problems, a micro-grid energy management method is disclosed in Chinese Patent No. CN109950919A, which includes the following steps: Step one: establishing an optimization model of the power grid, including photovoltaic, wind turbine, micro gas turbine, energy storage system and load; Step two: establishing a function with the minimum controllable power supply cost as the target based on the energy storage system; Step three: predicting the data of load and discharge power; Step four: calculating the charge and discharge power of the storage system; Step five: optimizing the charge and discharge control of the energy storage system. The present application comprehensively coordinates the needs of photovoltaic, micro gas turbine and user load through the charge and discharge power of the energy storage system, solves the problem of unequal distribution of possible source output and load demand, and realizes efficient use; by reducing the interference of prediction error on the charge and discharge control of the energy storage system, improving the flexibility of energy management, fully utilizing various types of energy, effectively reducing the peak load, reducing the impact of the load on the power grid, and realizing the economic and optimized use of energy.

[0005] For example, Chinese Patent No. CN112069676A discloses a micro-grid energy management method containing clean energy, belonging to the technical field of micro-grid of power system, including the following steps: Step 1: modeling the photovoltaic power generation system, wind power generation system, energy storage system, diesel generator system and micro gas turbine system in the micro-grid; Step 2: determining the total target function of the system; Step 3: establishing the constraint conditions of each system in the micro-grid; Step 4: improvement of NSGA-II algorithm; Step 5: combining the improved NSGA-II method with the constructed total target function to obtain the optimal solution of the target function. This method improves the defects existing in the existing micro-grid energy management strategy, ensures the reasonable distribution of distributed power generation, gas turbine, energy storage device and the like when the micro-grid is normally operated, and guarantees the maximum economic benefit and environmental protection benefit of the micro-grid operation.

[0006] At present, the existing micro-grid energy management technology still has deficiencies: the first patent establishes a target function by modeling, predicts and calculates the charge and discharge power of the energy storage system, and solves the problem of unequal distribution of possible source output and load demand. The second patent realizes the reasonable distribution of micro-grid system devices by modeling, determining the combination of improved NSGA-II algorithm and constructed total target function of the system, but the first patent predicts and calculates the charge and discharge power needed on the same day, and the accuracy of prediction decreases with the increase of time scale, and the second patent still has power grid instability due to the randomness and volatility of clean energy, and the existing technology still needs to be improved. SUMMARY

[0007] In view of the deficiencies of the prior art, the present application provides a multi-time-scale micro-grid energy management method to solve the above problems.

[0008] To achieve the above object, the present application is realized by the following technical solutions:

[0009] A multi-time scale micro-grid energy management method mainly comprises the following steps:

[0010] S1: information exchange of each device in the micro-grid is carried out by using a communication scheme based on router networking, and a Modbus / Tcp protocol is used as the communication protocol;

[0011] S2: a central controller executes and issues a real-time scheduling strategy of the micro-grid according to real-time collected system operation information, coordinates output conditions of the distributed power supply and the energy storage unit, and controls switching of the load and whether to purchase or sell electricity from the large power grid, so as to ensure efficient operation of the micro-grid system, and the collected photovoltaic output power and load historical data and weather forecast are used for prediction of the photovoltaic output power and the load;

[0012] S3: on the basis of the prediction information, the central controller collects real-time electricity price, comprehensively considers constraint conditions of each unit in the micro-grid, mathematically models each unit, takes minimization of operation cost and environmental treatment cost as an optimization target, formulates a 24-hour day-ahead scheduling scheme with 1 hour as a period, and transmits the day-ahead scheduling scheme to a local controller, the day-ahead scheduling scheme takes minimization of operation cost and environmental treatment cost of the micro-grid as an optimization target, and a target function is as follows:

[0013] min[C operation +C environment +C grid ] (1)

[0014] (1) in the formula, C operation is the operation cost of the micro-grid, C environment is the environmental treatment cost, and C grid is cost brought by purchase and sale of electricity between the micro-grid and the large power grid;

[0015] S4: according to the latest and more accurate intra-day prediction information, the central controller carries out online rolling correction on the day-ahead scheduling scheme to obtain an intra-day scheduling scheme with 15 minutes as a period on the scheduling day, improves accuracy of the scheduling scheme, reduces power error in actual operation of the micro-grid, and the optimization target of the intra-day scheduling scheme is to add a penalty cost to the optimization target of the day-ahead scheduling scheme, the purpose is to reduce difference between the intra-day scheduling scheme and the day-ahead scheduling scheme, and a target function is as follows:

[0016] min[C operation +C environment +C grid +C penalty ] (2)

[0017] (2) in the formula, C penalty is the penalty cost.

[0018] Furthermore, the microgrid includes multiple distributed generation units, each with a local controller. All local controllers are connected to a router via Cat6e Ethernet cables, and then to a central controller via Cat6e Ethernet cables. The local controllers are equipped with human-machine interfaces (HMIs) to facilitate on-site personnel in monitoring equipment status and performing operations. The central controller is responsible for collecting local information at the grid connection points of the distributed generation units and information such as the voltage and frequency of the mains grid. Based on this information, it completes all calculation tasks, including scheduling plan formulation, photovoltaic output power and load prediction, and issues instructions to the local controllers after completing the calculations.

[0019] Furthermore, the prediction of photovoltaic output power and load is based on the latest weather forecast information. First, wavelet analysis is used to perform time-frequency analysis on historical photovoltaic power and load data to improve the accuracy of the prediction results. Then, a BP neural network is used to obtain the final prediction. Since the day-ahead scheduling plan is formulated some time before the scheduling day, the accuracy of weather forecasts and other information on the scheduling day is relatively low, so predictions are made on a 1-hour cycle. In contrast, the intraday scheduling plan is formulated on the scheduling day itself, when weather forecasts and other information are more accurate, so more detailed predictions are made on a 15-minute cycle.

[0020] Furthermore, the day-ahead scheduling scheme aims to minimize the operating costs and environmental remediation costs of the microgrid. The operating cost C of the microgrid is defined in its objective function. operation This mainly includes maintenance costs for each distributed generation unit; environmental remediation costs C environment This mainly includes the treatment of pollutants such as nitrogen oxides and sulfides generated by distributed power generation, C grid This refers to the costs associated with the purchase and sale of electricity between microgrids and the main power grid.

[0021] Furthermore, the operating cost C of the microgrid operation The formula is:

[0022]

[0023] C i =χ i P i (1-2)

[0024] In equation (1-1), C i χ is the operating cost of the i-th distributed generation unit, and N is the number of distributed generation units; (1-2) where χ i It is the maintenance cost coefficient, P i It is the active power output of the i-th distributed generation unit.

[0025] Furthermore, the environmental remediation cost C environment The formula is:

[0026]

[0027] (1-3) where α n and α s is the cost coefficient of sulfur and nitrogen, and is the amount of sulfur and nitrogen emitted by the ith power generation unit in 1 hour, P i (t) is the active output power of the ith power generation unit at time t.

[0028] Further, the cost C grid brought by the purchase and sale of electricity between the microgrid and the main grid is:

[0029]

[0030] (1-4) where C buy and C sell represent the price of purchasing electricity from the main grid and selling electricity to the main grid, respectively, P grid is the power exchanged between the microgrid and the main grid, greater than 0 indicating the purchase of electricity from the main grid, and less than 0 indicating the sale of electricity to the main grid.

[0031] Further, the constraints in the microgrid mainly include overall constraints and unit constraints, and the overall constraint refers to the power balance in the microgrid:

[0032]

[0033] (3) where P load (t) is the active load demand at time t in the microgrid, P loss (t) is the total active power loss at time t in the microgrid.

[0034] The unit constraints include the power limit of the distributed power generation unit, the state of charge limit of the energy storage system, and the power change rate limit of the controllable unit:

[0035]

[0036] SOC min ≤ SOC(t) ≤ SOC max (3-2)

[0037]

[0038] (3-1) where and are the upper and lower limits of the active output power of the ith distributed power generation unit; (3-2) where SOC is the state of charge of the energy storage battery, SOC max and SOCmin are upper and lower limits of state of charge, which are used to prolong the life of the battery; (3-3) in which is the maximum power variation rate of the ith controllable unit.

[0039] Further, the accuracy of the day-ahead power prediction in the day-ahead scheduling scheme decreases with time, so the day-ahead scheduling scheme cannot meet the actual microgrid operation requirements, and the prediction results of power load and photovoltaic output need to be rolling corrected. In order to follow the day-ahead plan as much as possible, not to change the load operation state, and to meet the power balance, the output power limit of each unit and other constraints, the period is 15 minutes. When the first period of the scheduling day comes, the initial scheduling plan is executed according to the day-ahead scheduling plan, while the photovoltaic power and load information of the next period are predicted every period, and the next period of the day-ahead scheduling plan is made. The rolling correction of the scheduling arrangement of each unit in the next period makes the total output of the power generation unit and the actual power generation demand gradually approach, so as to ensure the safe and stable operation of the microgrid. The optimization objective of the day-ahead scheduling scheme is to add a penalty cost to the optimization objective of the day-ahead scheduling scheme. The purpose of this is to reduce the difference between the day-ahead scheduling scheme and the day-ahead scheduling scheme. The penalty cost C penalty in the objective function.

[0040]

[0041] The penalty cost C penalty is related to the difference between the power instruction value in the day-ahead scheduling scheme and the day-ahead scheduling scheme. In equation (2-1), n is the number of controllable power generation units; μ i is the penalty coefficient of the ith controllable power generation unit.

[0042] The constraint conditions not only meet the constraint conditions in the day-ahead scheduling scheme, but also limit the power adjustment amount of each power generation unit.

[0043]

[0044] Further, the microgrid can decide whether to buy or sell electricity from the main grid according to the real-time electricity price and the output information of each distributed power source in the microgrid, so as to achieve the purpose of stabilizing the operation of the microgrid and minimizing the operation cost.

[0045] The beneficial effects of the present application are that: a multi-time scale micro-grid energy management method adopts a low-cost router-based networking scheme, is simple and easy to implement, and can realize communication interaction without complex field buses, combines day-ahead optimization and intraday optimization, the intraday optimization performs rolling correction on the day-ahead scheduling scheme according to the latest and more accurate prediction information, reduces the influence of the uncertainty factors of the prediction information, improves the accuracy and reliability of the micro-grid scheduling plan, and reduces the operation cost of the micro-grid. The introduction of power interaction with the large power grid reduces the fluctuation caused by new energy power generation, further improves the stability of the micro-grid, and the bidding of the real-time electricity price of the large power grid and the distributed power generation grid price further optimizes the cost of the micro-grid. In order to avoid the single optimization target, the environmental governance cost is included in the target function, and the comprehensive consideration of environmental benefits and economic benefits is realized. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 The micro-grid system structure schematic diagram of an embodiment of the micro-grid energy management method of the present application is shown in the figure.

[0047] Figure 2 The micro-grid energy management system structure diagram of an embodiment of the micro-grid energy management method of the present application is shown in the figure.

[0048] Figure 3 The micro-grid energy management system flow chart of an embodiment of the micro-grid energy management method of the present application is shown in the figure.

[0049] Figure 4 The micro-grid communication networking scheme diagram of an embodiment of the micro-grid energy management method of the present application is shown in the figure.

[0050] Figure 5 The relationship diagram of day-ahead optimization and intraday optimization of the micro-grid energy management method of the present application is shown in the figure.

[0051] Figure 6 The photovoltaic output power and load power prediction flow chart of the micro-grid energy management method of the present application is shown in the figure.

[0052] Figure 7 The prediction error diagram of the micro-grid energy management method of the present application is shown in the figure.

[0053] Figure 8 The actual photovoltaic output power prediction result of the micro-grid energy management method of the present application is shown in the figure.

[0054] Figure 9 The load power prediction result of the micro-grid energy management method of the present application is shown in the figure.

[0055] Figure 10 The simulation result of the day-ahead scheduling scheme of the multi-time scale micro-grid energy management method of the application;

[0056] Figure 11 The simulation result of the day-ahead scheduling scheme of the multi-time scale micro-grid energy management method of the application; DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some of the embodiments of the application, but not all the embodiments of the application.

[0058] The application provides a technical solution: a multi-time scale micro-grid energy management method, mainly comprising the following steps:

[0059] S1: information exchange of each device in the micro-grid is performed by using a communication scheme based on router networking, and a Modbus / Tcp protocol is used as the communication protocol.

[0060] The micro-grid comprises a plurality of distributed power generation units, each of which has a local controller, all the local controllers are connected to a router through a super six category network cable, and then connected to a central controller through a super six category network cable. Meanwhile, the local controller is provided with a human-machine interface (HMI) to facilitate on-site staff to detect the equipment state and perform operation, the central controller is responsible for collecting local and partial information at the grid-connected point of the distributed power generation unit connected to the micro-grid and information such as the voltage and frequency of the large power grid, thereby completing all calculation tasks, including the formulation of the scheduling plan, the prediction of the photovoltaic output power and the load, and issuing instructions to the local controller after the calculation is completed.

[0061] S2: the central controller executes and issues the real-time scheduling strategy of the micro-grid according to the real-time collected system operation information, coordinates the output conditions of the distributed power source and the energy storage unit and the switching of the load, and controls whether to purchase or sell electricity from the large power grid, to ensure efficient operation of the micro-grid system, and uses the collected photovoltaic output power and load historical data and weather forecast to predict the photovoltaic output power and the load.

[0062] The central controller is responsible for processing data transmitted by the local controller, optimizing calculation to obtain optimal scheduling instructions of the distributed power supply, and sending to the local controller. The local controller is mainly responsible for monitoring the running state of each distributed power supply, sending data to the central processor, and receiving instructions from the central processor to control the active output of the distributed power supply, so as to achieve the purpose of unified control. The energy supply side includes the upper grid, biogas power generation, photovoltaic power generation, and small biomass power generation; the energy storage side is composed of storage batteries and pumped storage; the energy demand side contains various loads. The SCADA software in the central controller is responsible for collecting information in the microgrid for optimization calculation, and can also display real-time weather and real-time electricity price, generate reports, etc., as shown in FIG. 8. Figure 1

[0063] The microgrid grid-connected operation state is based on meeting the user side demand, and the storage battery is used for power supply priority during the power consumption peak period, and the power grid is used for power supply priority during the non-power consumption peak period. The central controller executes the real-time detection and regulation strategy of the distributed power generation and energy storage unit, the photovoltaic, biogas and small biomass systems adopt maximum power point tracking to maintain maximum power output, the pumped storage and energy storage system adjusts the running state according to the actual running state of the microgrid, the ordinary load and important load maintain the rated running state, and the load can be changed according to the user's needs and is not controlled by the microgrid energy management platform, so as to ensure the safety and reliability of the microgrid operation and make the economic benefit optimal. In the microgrid island operation state, the energy storage unit adjusts the charge and discharge power of the energy storage system according to the instructions issued by the central controller, each distributed power generation unit adopts maximum power point tracking control to maintain maximum power output, and the central controller issues real-time instructions according to the current state of the energy storage and distributed power generation unit to ensure the balance of energy supply and demand in the microgrid.

[0064] The real-time scheduling strategy process is that the central controller issues initialization instructions, initializes the controllable data variables and running state of each unit of the microgrid, performs microgrid inspection, judges whether each unit is faulty and whether there is a safety warning, and issues grid-connected or islanded instructions when everything is ready. The specific control process of the microgrid grid-connected state is as follows: at time t, when , and SOC max ≥ SOC(t) ≥ SOC min , if it is in the power consumption valley, the storage battery is charged from the power grid, and the microgrid purchases power from the power grid; if it is in the power consumption peak period, when the energy storage system outputs power according to the maximum discharge power, the power can meet the active power shortage of the microgrid, at this time the energy storage determines its output power according to the active power difference, the storage battery discharges to the microgrid, and the tie line power is 0; otherwise, the storage battery discharges to the microgrid while the microgrid purchases power from the power grid; SOC(t) ≥ SOC max ​When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. max When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. max When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above.

[0065] The specific control process of the micro-grid island state is as follows: at the time t, when When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. min When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. min When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. min When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. min When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. max When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. max When the time is at the low power consumption valley, the battery does not charge or discharge, and the micro-grid purchases power from the grid; when the power consumption peak is reached, the same adjustment process is performed as described above. Figure 3 P CMAX is the maximum charging power of the energy storage; and P DMAX is the maximum discharging power of the energy storage.

[0066] The specific manner of the photovoltaic output power prediction and the load prediction is as follows: for the embodiment in the application, the photovoltaic output power prediction is to predict the photovoltaic power sequence in a future period of time. Both the load sequence and the photovoltaic power sequence belong to non-stationary signals and have strong randomness, and are greatly affected by weather conditions, such as overcast, rainfall, gale, temperature, etc., which can cause changes in the load and the photovoltaic output power, and therefore, the two sequences have volatility. The prediction of the solar irradiance is the most important in the photovoltaic power prediction. First, the original data needs to be processed to analyze the internal regularity. For the two signals of the power load and the solar irradiance, not only all the frequency components need to be obtained, but also the time when each frequency component appears, that is, the signal frequency changes with time, that is, time-frequency analysis needs to be performed. Before prediction, the load / photovoltaic data sequence is decomposed into high-frequency components and low-frequency components by wavelet analysis. In this way, the components of different frequency scales of the original data are obtained by wavelet analysis. Then, the components on each frequency scale are predicted. The application adopts a BP neural network as the prediction algorithm, and the number of neurons in each layer in the network is as follows: 40 for each of the input layer and the output layer, and 10 for the hidden layer. The sample test data is the solar irradiance from 8 a.m. to 6 p.m. for 700 days in a certain place, the measurement interval is 15 minutes, the input variable is the solar irradiance change on a certain day in the training set, and the expected output is the solar irradiance change in the next day. After the BP neural network is trained by using the historical data of the solar irradiance for 700 days, the performance of the trained BP neural network is tested by using the solar irradiance data in the test set. There are 400 days of solar irradiance data in the test set, the data for the first 300 days are taken as the input, and the solar irradiance for the next 6 days is predicted. The predicted output is compared with the actual solar irradiance (the expected output) for the 6 days to obtain the prediction error as shown in Figure 7 The actual photovoltaic output power and the load prediction results are shown in Figure 8 , Figure 9 .

[0067] S3: On the basis of the prediction information, the central controller collects the real-time electricity price, comprehensively considers the constraint conditions of each unit in the microgrid, mathematically models each unit, takes the minimization of the operation cost and the environmental treatment cost as the optimization target, formulates a 24-hour day-ahead scheduling scheme with a period of 1 hour, and transmits the day-ahead scheduling scheme to the local controller. The objective function of the day-ahead scheduling scheme is as follows:

[0068] min[C operation +C environment +C grid ] (1)

[0069] (1) In the formula, C operationis the operation cost of the microgrid, mainly including the maintenance cost of each distributed generation unit, C environment is the environmental treatment cost, mainly including the treatment of pollutants such as nitride and sulfide caused by distributed generation, C grid is the cost caused by the purchase and sale of electricity between the microgrid and the main grid. Among them,

[0070]

[0071] C i = χ i P i (1-2)

[0072] (1-1) In the formula C i is the operation cost of the i-th distributed generation unit, and N is the number of distributed generation units; (1-2) In the formula χ i is the maintenance cost coefficient, P i is the active output power of the i-th distributed generation unit.

[0073]

[0074] (1-3) In the formula α n and α s are the cost coefficients of treating sulfide and nitride, and is the amount of sulfur and nitride emitted by the i-th generation unit in 1 hour, P i (t) is the active output power of the i-th generation unit at time t.

[0075]

[0076] (1-4) In the formula C buy and C sell respectively represent the purchase price of electricity from the main grid and the sale price of electricity to the main grid, P grid is the power exchanged between the microgrid and the main grid, greater than 0 indicates the purchase of electricity from the main grid, and less than 0 indicates the sale of electricity to the main grid.

[0077] S4: According to the latest more accurate intra-day prediction information, the central controller performs online rolling correction on the day-ahead scheduling scheme with a period of 15 minutes to obtain an intra-day scheduling scheme on the dispatch day, improve the accuracy of the scheduling scheme, and reduce the power error in the actual operation of the microgrid. The objective function of the intra-day scheduling scheme is:

[0078] min[C operation +C environment +C grid +C penalty ] (2)

[0079] (2) In the formula Cpenalty It is the cost of punishment.

[0080] The relationship between daily optimization and intraday optimization is as follows: Figure 5 As shown. The day-ahead optimization feature is that the day-ahead scheduling scheme aims to minimize the operating costs and environmental governance costs of the microgrid. Based on load and photovoltaic output forecasts, it collects information from various parts of the microgrid, considers the electricity price for the current time period, and takes into account the technical characteristics of each generating unit, while satisfying the constraints within the microgrid. The energy management system uses the Cplex mathematical solver to calculate the optimal scheduling strategy for the corresponding time period, obtaining the scheduling scheme that minimizes the microgrid's operating costs and ensuring the economical and stable operation of the microgrid. The optimization period is 24 hours, i.e., the length of one scheduling day, with 1-hour optimization intervals. During each optimization interval, the output power of each generating unit and the charging and discharging power of the energy storage system are considered constants that do not change. The accuracy of day-ahead power prediction in the aforementioned day-ahead dispatch scheme decreases over time, thus failing to meet the actual microgrid operation requirements. We need to continuously adjust the predicted power load and photovoltaic output, adhering as closely as possible to the day-ahead plan, without altering the load operation status, and satisfying constraints such as power balance and output power limits for each unit. Using a 15-minute cycle, the initial dispatch plan is executed according to the day-ahead plan at the start of the first cycle of the dispatch day. Simultaneously, each cycle forecasts the photovoltaic power and load information for the next cycle, formulating the intraday dispatch plan for the next cycle. This continuous adjustment of the dispatch arrangements for each unit at subsequent moments ensures that the total output of the generating units gradually approaches the actual power demand, guaranteeing the safe and stable operation of the microgrid. The optimization objective of the intraday dispatch scheme incorporates a penalty cost into the optimization objective of the day-ahead dispatch scheme. This aims to reduce the difference between the intraday and day-ahead dispatch schemes. Furthermore, to avoid excessive adjustments and increased costs, the intraday dispatch plan must strictly adhere to the day-ahead unit start-up and shutdown strategies. The intraday rolling optimization uses a 15-minute optimization interval, dividing the day into 96 time periods and considering the differences between peak, valley, and normal periods. The optimization scheduling cycle is 1 hour. Based on the latest power forecast information, the intraday output plan of each unit is adjusted according to the day-ahead plan with the objective of minimizing the total cost in the next stage. Its objective function includes:

[0081]

[0082] Penalty cost C penalty This is related to the difference in power command values ​​between the intraday dispatch scheme and the day-ahead dispatch scheme; (2-1) where n is the number of controllable generating units, μ i It is the penalty coefficient for the i-th controllable power generation unit.

[0083] The constraints must not only meet the constraints in the day-ahead dispatch scheme mentioned above, but also limit the power adjustment amount of each power generation unit.

[0084]

[0085] The micro-grid can decide whether to purchase or sell electricity from the main grid according to the real-time electricity price and the output information of each distributed power source in the micro-grid, so as to stabilize the operation of the micro-grid and minimize the operation cost.

[0086] The constraint conditions in the micro-grid mainly include overall constraints and unit constraints, and the overall constraint refers to the power balance in the micro-grid:

[0087]

[0088] (3) In the formula, P load (t) is the active load demand of the micro-grid at time t, P loss (t) is the total active power loss of the micro-grid at time t.

[0089] The unit constraints include the power limit of the distributed power generation unit, the state of charge limit of the energy storage system, and the power change rate limit of the controllable unit:

[0090]

[0091] SOC min ≤ SOC (t) ≤ SOC max (3-2)

[0092]

[0093] (3-1) In the formula, and are the upper and lower limits of the active output power of the i th distributed power generation unit; (3-2) In the formula, SOC is the state of charge of the energy storage battery, SOC max and SOC min are the upper and lower limits of the state of charge, which is to prolong the battery life; (3-3) In the formula, is the maximum power change rate of the i th controllable unit.

[0094] The day-ahead scheduling scheme obtained by using the Cplex solver is shown in Figure 10 It can be seen that the output of the power generation unit well follows the load curve change, and is affected by the time-of-use electricity price and the peak-valley difference of power load. The overall situation of the scheduling plan is that during the low price period, a large amount of electricity is purchased from the main grid, and the excess is stored in the energy storage unit. During the peak price period, the amount of electricity purchased from the main grid decreases significantly, and mainly relies on new energy power generation and energy storage unit output.

[0095] The day-ahead scheduling scheme obtained by using the Cplex solver is shown in Figure 11It can be seen that compared with the day-ahead optimization result, the day-ahead optimization unit adjustment is more frequent, but the overall result is better to follow the day-ahead scheduling scheme, which meets the requirement of the day-ahead scheduling scheme on the fine-tuning of the day-ahead scheduling scheme. The following table is the cost comparison of the day-ahead scheduling scheme and the day-ahead scheduling scheme:

[0096] Table 1 Economic cost of optimization results of each time scale

[0097]

[0098] The economic cost of the day-ahead scheduling scheme is low, but the prediction accuracy of the day-ahead optimization is low, the time scale of the day-ahead power scheduling is short, the prediction accuracy is high, the day-ahead rolling optimization is targeted at the economy of the current period, the frequency of the adjustment of the power generation unit is high, the global economic benefit is not considered, the cost is increased, but the actual use is more reliable.

[0099] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this, any skilled person in the art can make equivalent replacement or change according to the technical scheme and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

1. A multi-time scale microgrid energy management method, characterized in that: Mainly comprising the following steps: S1: using a router-based networking communication scheme for information exchange of each device in the micro-grid, and the communication protocol uses Modbus / TCP; S2: the central controller executes and issues the real-time scheduling strategy of the micro-grid according to the real-time collected system operation information, coordinates the output conditions of the distributed power supply and the energy storage unit, and controls the switching of the load, and whether to purchase and sell electricity with the main grid, and uses the collected photovoltaic output power and load historical data and weather forecast to predict the photovoltaic output power and load, decomposes the load / photovoltaic data sequence into high-frequency components and low-frequency components by wavelet time-frequency analysis before prediction, obtains the components of different frequency scales of the original data through wavelet analysis, predicts the components on each frequency scale, uses a BP neural network as a prediction algorithm, trains the BP neural network using historical solar irradiance data, and tests the performance of the trained BP neural network using solar irradiance data in the test set, and the purchase and sale of electricity between the micro-grid and the main grid is one of the scheduling decisions; S3: On the basis of the prediction information, the central controller collects real-time electricity price and time-of-use electricity price, comprehensively considers the constraint conditions of each unit in the microgrid, establishes a day-ahead optimization model, takes the sum of operation cost, environmental governance cost and electricity purchase and sale cost as the optimization objective, formulates a 24-hour day-ahead scheduling scheme with 1 hour as a cycle, and transmits it to the local controller. The day-ahead scheduling scheme takes minimizing the operation cost and environmental governance cost of the microgrid as the optimization objective, and its objective function is: (1) (1) wherein is the operation cost of the micro-grid, including the maintenance cost of each distributed power generation unit, is the environmental treatment cost, including the treatment of pollutants nitrogen oxides and sulfur oxides brought by the distributed power generation, is the cost brought by the purchase and sale of electricity between the micro-grid and the main grid; Operating cost of the micro-grid The formula is: In formula (1-1) is the operating cost of the mth distributed power generation unit, is the number of distributed power generation units; in formula (1-2) is the maintenance cost coefficient, is the active output power of the mth distributed power generation unit; the environmental governance cost The formula is: In formula (1-3) and are the cost coefficients of governing sulfur oxides and nitrogen oxides, is the amount of sulfur and nitrogen oxides emitted by the mth power generation unit in 1 hour, is the active output power of the mth power generation unit at time t. Costs incurred by the micro-grid buying and selling power from the main grid The formula is: In formula (1-4) And Pb and Ps represent the price of buying power from the main grid and selling power to the main grid, respectively, Pm is the power exchanged between the micro-grid and the main grid, greater than 0 indicates buying power from the main grid, less than 0 indicates selling power to the main grid; S4: According to the latest and more accurate intra-day prediction information, the central controller makes online rolling correction to the day-ahead scheduling scheme to obtain an intra-day scheduling scheme with a period of 15 minutes on the dispatch day, and the optimization objective function of the intra-day scheduling scheme is: wherein is a penalty cost. In order to follow the day-ahead plan as much as possible, not to change the load operation state, and to meet the power balance and the output power limit constraint of each unit, the scheduling period is 15 minutes. When the first period of the scheduling day comes, the initial scheduling plan is executed according to the day-ahead scheduling plan. At the same time, the photovoltaic power and load information of the next period are predicted every period, and the intraday scheduling plan of the next period is made. The scheduling arrangement of each unit at the next time is corrected, so that the total output of the power generation unit and the actual power generation demand gradually approach, and the microgrid is safely and stably operated. The objective function of the intraday scheduling scheme is: penalty cost related to the difference between the power instruction value in the intraday scheduling scheme and the day-ahead scheduling scheme. In formula (2-1), n is the number of controllable power generation units, is the penalty coefficient of the controllable power generation unit, is the power adjustment amount of each power generation unit. The constraint condition not only satisfies the constraint condition in the above day-ahead scheduling scheme, but also limits the power adjustment amount of each power generation unit, The constraint condition in the micro-grid mainly includes overall constraint and unit constraint, and the overall constraint refers to the power balance in the micro-grid: In the formula is the active load demand at time t in the micro-grid, is the total active power loss at time t in the micro-grid, and the unit constraint includes the power limit of the distributed power generation unit, the state of charge limit of the energy storage system, and the power change rate limit of the controllable unit: In the formula (3-1) is the upper and lower limit of the active output power of the i-th distributed power generation unit; in the formula (3-2) is the state of charge of the energy storage battery, is the upper and lower limit of the state of charge, and in the formula (3-3) is the maximum power change rate of the i-th controllable unit; The optimization model is solved by a mathematical programming solver Cplex to obtain executable day-ahead scheduling scheme and intraday scheduling scheme.

2. The multi-time scale microgrid energy management method of claim 1, wherein: The micro-grid includes a plurality of distributed power generation units, each of which has a local controller, all of which are connected to a router through a super six category network cable, and then connected to a central controller through a super six category network cable, and the local controller is equipped with a human-machine interface HMI to facilitate on-site staff to detect equipment status and operate, the central controller is responsible for collecting local information at the grid connection point of the distributed power generation unit accessing the micro-grid and the main grid voltage frequency information, thereby completing all calculation tasks, including scheduling plan formulation, photovoltaic output power and load prediction, and issuing instructions to the local controller after calculation.

3. The multi-time scale microgrid energy management method of claim 2, wherein: The constraint conditions in the micro-grid mainly include overall constraints and unit constraints, and the main target of the real-time regulation of the central controller in the grid-connected and island operation state of the micro-grid is to meet the overall constraints and unit constraints of the micro-grid, the output conditions of each unit of the real-time micro-grid and the switching of the load, and the prediction of photovoltaic output power and load is based on the latest weather forecast information, first through wavelet analysis to analyze the time-frequency of the historical data of photovoltaic power and load, improve the accuracy of the prediction result, and then through the BP neural network to obtain the result, the day-ahead scheduling scheme is formulated a period of time before the scheduling day, so the weather forecast information of the scheduling day has low accuracy, and the prediction is made in 1 hour period, while the intraday scheduling scheme is formulated on the scheduling day, the weather forecast information has high accuracy, and the prediction is made in 15 minute period, the micro-grid is taken as a hard constraint in the island state, and the economy is taken as the main constraint in the grid-connected state and is subject to the power limit of the tie line.

4. The multi-time scale microgrid energy management method of claim 3, wherein: The micro-grid can decide whether to purchase and sell electricity with the main grid according to the real-time electricity price and the output information of each distributed power supply in the micro-grid, so as to stabilize the operation of the micro-grid and minimize the operation cost.

Images