Multi-time-scale source-load-storage collaborative optimization scheduling method based on model prediction

Abstract

The invention provides a multi-time-scale source-load-storage collaborative optimization scheduling method based on model prediction, and the method comprises the following steps: obtaining photovoltaic and load historical data of a source-load-storage system for day-ahead prediction, and obtaining source-load day-ahead prediction data; reconstructing and sorting the day-ahead prediction data of the photovoltaic power and the load power and the local real-time electricity price data, dividing a scheduling time period and performing capacity calculation; calculating an energy storage day-ahead power instruction of each time period according to the source load day-ahead prediction data and the electricity price data; and performing intra-day rolling prediction on the photovoltaic output power and the load prediction power, and performing real-time scheduling of each time period according to a prediction result. According to the method, a day-ahead and intra-day two-stage scheduling mode is adopted, optimal distribution of energy storage power is carried out according to photovoltaic and load prediction results and local real-time electricity prices in the day ahead, the purpose of improving the energy utilization rate and energy storage operation economy is achieved, adjustment is carried out according to the real-time prediction results on the basis of day-ahead instructions in the day, and the energy utilization rate is improved. Safe and economical operation of the system is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of energy storage technology, and in particular relates to a multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction. Background Art

[0002] The penetration rate of new energy in the power grid continues to increase, and the proportion of power generation is getting larger and larger. However, the randomness of new energy power generation is relatively high, and it depends to a large extent on external environmental conditions, which can easily lead to problems such as wind and solar power abandonment. The development of energy storage technology has solved the above problems to a certain extent.

[0003] Energy storage technology is highly flexible and responsive. It can absorb excess energy and make up for power shortages through rapid charging and discharging, thus improving system reliability. At the same time, for industrial and commercial users, under the current background of peak-valley electricity prices, energy storage devices can obtain certain benefits through peak-valley arbitrage, which is highly economical. Therefore, developing source-load-storage coordinated optimization scheduling is an important way to improve the level of new energy consumption and ensure the maximum economic operation of energy storage.

[0004] At present, realizing new energy output forecasting, load forecasting, and coordinated optimization scheduling between new energy power generation, energy storage and load considering real-time electricity prices, minimizing the abandonment rate and improving economic efficiency are still urgent issues to be solved in source-load-storage systems. Summary of the invention

[0005] The problem to be solved by the present invention is to provide a multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction, which can simply and conveniently realize the collaborative optimization scheduling between source-load-storage, make full use of the flexibility of energy storage action, play the role of peak shaving and valley filling, ensure the economy of system operation, and improve the utilization rate of new energy.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction, comprising the following steps:

[0007] S1: Obtain the photovoltaic and load historical data of the source-load-storage system for day-ahead forecasting, and obtain the source-load day-ahead forecast data;

[0008] S2: Reconstruct and sort the day-ahead forecast data of PV power and load power as well as the local real-time electricity price data, divide the scheduling time period and perform capacity calculation;

[0009] S3: Calculate the energy storage day-ahead power instruction for each time period according to the source-load day-ahead forecast data and the electricity price data;

[0010] S4: Carry out daily rolling forecasts for PV output power and load forecast power, and perform real-time scheduling for each time period based on the forecast results.

[0011] Furthermore, in S1, the photovoltaic and load historical data of the source-load-storage system are obtained, and the data used for day-ahead prediction are all one point per hour.

[0012] Furthermore, in S2, the photovoltaic power and load power day-ahead prediction models both use long short-term memory networks, the input parameters of the photovoltaic power day-ahead prediction model are the temperature, humidity, light intensity and humidity of the day to be predicted obtained according to the weather forecast, and the output parameter is the photovoltaic output power of the day to be predicted; the input parameter of the load day-ahead prediction model is the load power one day before the day to be predicted, and the output parameter is the load power of the day to be predicted.

[0013] Further, the S2 comprises the following steps:

[0014] S21: input the real-time electricity price, day-ahead PV power and load power forecast data in the form of a column vector;

[0015] S22: reorganize the real-time electricity price, day-ahead PV power and load power forecast data into a 24×3 matrix A;

[0016] S23: Rearrange each row of the matrix A in the order of electricity prices from low to high, and obtain new electricity price data dPrice sorted by electricity price and corresponding photovoltaic power prediction data P PV and load power forecast data P load , get the position idx0 of the new data in dPrice in the original data;

[0017] S24: traverse dPrice and find the positions n1, n2, and n3 of all the points where the electricity price increases. If there are only two points where the electricity price increases, then n3 = 25; determine [1, n1-1] as the electricity price valley period, [n1, n2-1] as the electricity price normal period, [n2, n3-1] as the electricity price peak period, and [n3, 24] as the electricity price peak period;

[0018] S25: Find all P PV >P load Calculate the next day's P PV >P load The remaining energy E after photovoltaic power supplies the load at the moment free ;

[0019] S26: Calculate [n3,24] time period P PV <P load The total energy E after deducting the photovoltaic power from the load power at the moment load ;

[0020] S27: Calculate the discharge amount E1 of the energy storage [n2, n3-1].

[0021] Furthermore, in S25, the remaining energy E free The calculation formula is:

[0022]

[0023] Where dt is the time interval in h.

[0024] Furthermore, in S26, the total energy E load The calculation formula is:

[0025]

[0026] Furthermore, in S27, the calculation formula of the discharge amount E1 is:

[0027] E1=E use -E load

[0028] Among them, E use is the available capacity of the energy storage device, in kWh, and the formula is:

[0029] E use =E max *(SOC max -SOC min )

[0030] Among them, E max is the maximum energy storage capacity in kWh, SOC max and SOC min are the maximum and minimum states of charge of the energy storage device, respectively.

[0031] Further, the S3 comprises the following steps:

[0032] S31: input the data processed by S2 and the energy of each time period calculated;

[0033] S32: Determine whether E free <E use , if so, then the energy storage power P in the period [1,n1-1] es =-P max , where P max is the maximum charge and discharge power of the energy storage device. The energy storage device is charged until the energy of the energy storage device reaches (E use -E free ), otherwise [1,n1-1] time period P es =0, energy storage does not work;

[0034] S33: i=n1, start loop;

[0035] S34: Determine whether P PV(i)<P load (i) If so, then P es (i)=-(P PV (i)-P load (i)), otherwise, go to S35;

[0036] S35: Determine whether i∈[n1,n2-1], if so, let P es (i) = 0, the energy storage does not work, otherwise, determine whether i∈[n2,n3-1], if so, let P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the [n2,n3-1] period reaches E1, otherwise, P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the period n3 to 24 reaches E load ;

[0037] S36: Determine whether i < 24, if so, set i = i + 1 and return to S34, otherwise, exit the loop;

[0038] S37: So far, the energy storage power P of all time periods is obtained es , P es Reorder according to idx0, obtain the energy storage power sorted by time and output it.

[0039] Further, the S4 comprises the following steps:

[0040] S41: Perform rolling forecasts on the photovoltaic power and load power at the next moment every 10 minutes during the day, and the forecast model still uses the LSTM model;

[0041] S42: [1, n1-1] The energy storage power in the corresponding period is issued according to the day-ahead instruction;

[0042] S43: [n2-1,24] The energy storage power command for the corresponding period is adjusted point by point according to the predicted photovoltaic power and load power during the day. The power command formula is the same as the day-ahead dispatch.

[0043] Furthermore, in S41, the input parameters of the photovoltaic power intraday prediction model are the temperature, humidity, light intensity, humidity and photovoltaic output power 24 hours before the prediction time obtained according to the weather forecast, and the output parameter is the photovoltaic output power at the prediction time; the input parameters of the load intraday prediction model are the load power 24 hours before the prediction time, and the output parameter is the load power at the prediction time.

[0044] The advantages and positive effects of the present invention are:

[0045] The present invention adopts a two-stage scheduling method, that is, the day-ahead and day-within-day. The energy storage power is optimally allocated according to the photovoltaic and load forecast results and the local real-time electricity price on the day-ahead, so as to achieve the purpose of improving energy utilization and the economic efficiency of energy storage operation. During the day, adjustments are made based on the day-ahead instructions according to the real-time forecast results to ensure the safe and economical operation of the system. The present invention can simply and conveniently realize the coordinated optimization scheduling among source, load and storage, make full use of the flexibility of energy storage action, play the role of peak shaving and valley filling, ensure the economic efficiency of system operation, and improve the utilization rate of new energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of the overall structure of an embodiment of the present invention.

[0047] Figure 2 It is a flow chart of data reconstruction, sorting and energy storage energy calculation in an embodiment of the present invention.

[0048] Figure 3 It is a flow chart of the day-ahead and intra-day source-load-storage coordinated optimization scheduling method according to an embodiment of the present invention.

[0049] Figure 4 It is a power diagram of each unit under the day-ahead source-load-storage collaborative optimization scheduling method based on model prediction in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0052] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] The embodiments of the present invention are further described below in conjunction with the accompanying drawings:

[0054] like Figure 1 As shown, a multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction includes the following steps:

[0055] S1: Obtain the photovoltaic and load historical data of the source-load-storage system for day-ahead forecasting, and obtain the source-load day-ahead forecasting data. Specifically, obtain the photovoltaic and load historical data of the source-load-storage system, and the data used for day-ahead forecasting are all one point per hour.

[0056] S2: Reconstruct and sort the day-ahead forecast data of photovoltaic power and load power as well as the local real-time electricity price data, divide the scheduling time period and perform capacity calculation. Specifically, the photovoltaic power and load power day-ahead forecast models both use long short-term memory networks. The input parameters of the photovoltaic power day-ahead forecast model are the temperature, humidity, light intensity and humidity of the day to be predicted obtained from the weather forecast, and the output parameter is the photovoltaic output power of the day to be predicted; the input parameter of the load day-ahead forecast model is the load power one day before the day to be predicted, and the output parameter is the load power of the day to be predicted. Specifically, if Figure 2 As shown, S2 includes the following steps:

[0057] S21: Input the real-time electricity price, day-ahead photovoltaic power and load power forecast data in the form of a column vector.

[0058] S22: Reorganize the real-time electricity price, day-ahead PV power, and load power forecast data into a 24×3 matrix A.

[0059] S23: Rearrange each row of the matrix A in the order of electricity prices from low to high, and obtain new electricity price data dPrice sorted by electricity price and corresponding photovoltaic power prediction data P PV and load power forecast data P load , get the position idx0 of the new data in dPrice in the original data.

[0060] S24: Traverse dPrice and find the locations n1, n2, and n3 of all electricity price increase points. If there are only two electricity price increase points, then n3=25; determine [1, n1-1] as the electricity price valley period, [n1, n2-1] as the electricity price normal period, [n2, n3-1] as the electricity price peak period, and [n3, 24] as the electricity price peak period.

[0061] S25: Find all P PV >P load Calculate the next day's P PV >P load The remaining energy E after photovoltaic power supplies the load at the moment free ; Residual energy E free The calculation formula is:

[0062]

[0063] Where dt is the time interval in h.

[0064] S26: Calculate [n3,24] time period P PV <P load The total energy E after deducting the photovoltaic power from the load power at the moment load Total energy E load The calculation formula is:

[0065]

[0066] S27: Calculate the discharge amount E1 of the energy storage [n2, n3-1]. The calculation formula of the discharge amount E1 is:

[0067] E1=E use -E load

[0068] Among them, E use is the available capacity of the energy storage device, in kWh, and the formula is:

[0069] E use =E max *(SOC max -SOC min )

[0070] Among them, E maxis the maximum energy storage capacity in kWh, SOC max and SOC min are the maximum and minimum states of charge of the energy storage device, respectively.

[0071] S3: Calculate the energy storage day-ahead power instruction for each time period based on the source-load day-ahead forecast data and electricity price data. Figure 3 As shown, S3 includes the following steps:

[0072] S31: Input the data processed by S2 and the energy calculated for each time period.

[0073] S32: Determine whether E free <E use , if so, then the energy storage power P in the period [1,n1-1] es =-P max , where P max is the maximum charge and discharge power of the energy storage device. The energy storage device is charged until the energy of the energy storage device reaches (E use -E free ), otherwise [1,n1-1] time period P es =0, energy storage does not work.

[0074] S33: i=n1, start loop.

[0075] S34: Determine whether P PV (i)<P load (i) If so, then P es (i)=-(P PV (i)-P load (i)), otherwise go to S35.

[0076] S35: Determine whether i∈[n1,n2-1], if so, let P es (i) = 0, the energy storage does not work, otherwise, determine whether i∈[n2,n3-1], if so, let P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the [n2,n3-1] period reaches E1, otherwise, P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the period n3 to 24 reaches E load .

[0077] S36: Determine whether i<24, if so, set i=i+1 and return to S34, otherwise, exit the loop.

[0078] S37: So far, the energy storage power P of all time periods is obtained es , P es Reorder according to idx0, obtain the energy storage power sorted by time and output it.

[0079] S4: Perform daily rolling forecasts on photovoltaic output power and load forecast power, and perform real-time scheduling for each period based on the forecast results. Specifically, S4 includes the following steps:

[0080] S41: Perform rolling prediction of the photovoltaic power and load power at the next moment every 10 minutes during the day, and the prediction model still uses the LSTM model. Among them, the input parameters of the photovoltaic power intraday prediction model are the temperature, humidity, light intensity, humidity and photovoltaic output power 24 hours before the prediction time obtained according to the weather forecast, and the output parameter is the photovoltaic output power at the prediction time; the input parameter of the load intraday prediction model is the load power 24 hours before the prediction time, and the output parameter is the load power at the prediction time.

[0081] S42: The energy storage power in the corresponding period of [1,n1-1] is issued according to the day-ahead instruction.

[0082] S43: [n2-1,24] The energy storage power command for the corresponding period is adjusted point by point according to the predicted photovoltaic power and load power during the day. The power command formula is the same as the day-ahead dispatch.

[0083] The present invention will be further described below in conjunction with specific embodiments:

[0084] S1: Obtain the historical data of photovoltaic and load of a certain industrial and commercial park in Henan Province. The data used for day-ahead prediction are all one point per hour. The day-ahead prediction models for photovoltaic power and load power both use long short-term memory networks (LSTM). The input of the photovoltaic power day-ahead prediction model is the temperature, humidity, light intensity and humidity of the day to be predicted obtained according to the weather forecast, and the output is the photovoltaic output power of the day to be predicted; the input of the load day-ahead prediction model is the load power one day before the day to be predicted, and the output is the load power of the day to be predicted.

[0085] S2: Reconstruct and sort the day-ahead forecast data of PV power and load power as well as the local real-time electricity price data, divide the scheduling time period and perform capacity calculation.

[0086] S21: Input the real-time electricity price, day-ahead photovoltaic power and load power forecast data in the form of a column vector.

[0087] S22: Reorganize the real-time electricity price, day-ahead PV power, and load power forecast data into a 24×3 matrix A.

[0088] S23: Rearrange each row of the matrix A in the order of electricity prices from low to high, and obtain new electricity price data (dPrice, in yuan / kWh) and corresponding photovoltaic power prediction data (P PV , in kW) and load power forecast data (P load , in kW), get the position of the new data in dPrice in the original data (idx0).

[0089] S24: Traverse dPrice and find the positions of all points where electricity prices increase, which are n1=9, n2=17, and n3=21. So far, it can be determined that [1, n1-1] is the electricity price valley period, [n1, n2-1] is the electricity price normal period, [n2, n3-1] is the electricity price peak period, and [n3, 24] is the electricity price spike period.

[0090] S25: Find all P PV >P load Calculate the next day's P PV >P load The remaining energy E after photovoltaic power supplies the load at the moment free , the calculation formula is as follows:

[0091]

[0092] S26: Calculate [n3,24] time period P PV <P load The total energy E after deducting the photovoltaic power from the load power at the moment load (Unit: kWh):

[0093]

[0094] Wherein, dt is the time interval, which is 1h.

[0095] S27: Calculate the discharge amount E1 (in kWh) of the energy storage [n2, n3-1]:

[0096] E1=E use -E load

[0097] Among them, E use is the available capacity of the energy storage device, in kWh, and the formula is:

[0098] E use =E max *(SOC max -SOC min )

[0099] Among them, E max The maximum energy storage capacity is 3960kWh, SOC max and SOC min are the maximum and minimum states of charge of the energy storage device, which are 0.98 and 0.03 respectively.

[0100] S3: Calculate the energy storage day-ahead power instruction for each time period based on the source-load day-ahead forecast data and electricity price data.

[0101] S31: Input the data processed by S2 and the energy calculated for each time period.

[0102] S32: Judgement E free <E use ? If so, then the energy storage power P in the period [1,n1-1] is es =-P max (P max is the maximum charge and discharge power of the energy storage device, which is 1000kW), and the energy storage device is charged until the energy of the energy storage device reaches (E use -E free ), otherwise [1,n1-1] time period P es =0, energy storage does not work.

[0103] S33: i=n1, start loop.

[0104] S34: Determine P PV (i)<P load (i)? If yes, then P es (i)=-(P PV (i)-P load (i)), otherwise go to S35.

[0105] S35: Determine i∈[n1,n2-1]? If so, let P es (i)=0, energy storage does not work, otherwise, judge i∈[n2,n3-1]? If so, let P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the [n2,n3-1] period reaches E1, otherwise, P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the period n3-24 reaches E load .

[0106] S36: Check if i is less than 24? If so, set i=i+1 and return to S34; otherwise, exit the loop.

[0107] S37: So far, the energy storage power P of all time periods is obtained es , P es Reorder according to idx0, obtain the energy storage power sorted by time and output it. Figure 4 The power diagram of each unit under the day-ahead source-load-storage coordinated optimization scheduling method based on model prediction. In the figure, PPV is the day-ahead photovoltaic power prediction value, Pload is the day-ahead load power prediction value, and Pes is the power instruction of the energy storage device. Figure 4 It can be seen that the energy storage device is charged when the electricity price is low and there is surplus photovoltaic power, and discharged when the electricity price is peak, which can obtain maximum profit through the peak-valley price difference. At the same time, this method also reduces the system's abandonment rate and improves the level of new energy consumption.

[0108] S4: Carry out daily rolling forecasts for PV output power and load forecast power, and perform real-time scheduling for each time period based on the forecast results.

[0109] S41: Perform rolling prediction of the photovoltaic power and load power at the next moment every 10 minutes during the day. The prediction model still uses the LSTM model, where the input of the photovoltaic power daily prediction model is the temperature, humidity, light intensity, humidity and photovoltaic output power 24 hours before the prediction time obtained according to the weather forecast, and the output is the photovoltaic output power at the prediction time; the input of the load daily prediction model is the load power 24 hours before the prediction time, and the output is the load power at the prediction time.

[0110] S42: The energy storage power in the corresponding period from 1 to n1-1 is issued according to the day-ahead instruction.

[0111] S43: The energy storage power command for the period n2-1 to 24 is adjusted point by point according to the predicted photovoltaic power and load power for the day, and the power command formula is the same as the day-ahead dispatch.

[0112] The advantages and positive effects of the present invention are:

[0113] The present invention adopts a two-stage scheduling method, that is, the day-ahead and day-within-day. The energy storage power is optimally allocated according to the photovoltaic and load forecast results and the local real-time electricity price on the day-ahead, so as to achieve the purpose of improving energy utilization and the economic efficiency of energy storage operation. During the day, adjustments are made based on the day-ahead instructions according to the real-time forecast results to ensure the safe and economical operation of the system. The present invention can simply and conveniently realize the coordinated optimization scheduling among source, load and storage, make full use of the flexibility of energy storage action, play the role of peak shaving and valley filling, ensure the economic efficiency of system operation, and improve the utilization rate of new energy.

[0114] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction, characterized by: The following steps are included: S1: Obtain the photovoltaic and load historical data of the source-load-storage system for day-ahead forecasting, and obtain the source-load day-ahead forecast data; S2: Reconstruct and sort the day-ahead forecast data of PV power and load power as well as the local real-time electricity price data, divide the scheduling time period and perform capacity calculation; S3: Calculate the energy storage day-ahead power instruction for each time period according to the source-load day-ahead forecast data and the electricity price data; S4: Carry out daily rolling forecasts for PV output power and load forecast power, and perform real-time scheduling for each time period based on the forecast results.

2. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 1 is characterized by: In S1, the photovoltaic and load historical data of the source-load-storage system are obtained, and the data used for day-ahead prediction are all one point per hour.

3. A multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 1 or 2, characterized in that: In S2, the photovoltaic power and load power day-ahead prediction models both use long short-term memory networks. The input parameters of the photovoltaic power day-ahead prediction model are the temperature, humidity, light intensity and humidity of the day to be predicted obtained according to the weather forecast, and the output parameter is the photovoltaic output power of the day to be predicted; the input parameter of the load day-ahead prediction model is the load power one day before the day to be predicted, and the output parameter is the load power of the day to be predicted.

4. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 3 is characterized by: The S2 comprises the following steps, S21: input the real-time electricity price, day-ahead PV power and load power forecast data in the form of a column vector; S22: reorganize the real-time electricity price, day-ahead PV power and load power forecast data into a 24×3 matrix A; S23: Rearrange each row of the matrix A in the order of electricity prices from low to high, and obtain new electricity price data dPrice sorted by electricity price and corresponding photovoltaic power prediction data P PV and load power forecast data P load , get the position idx0 of the new data in dPrice in the original data; S24: traverse dPrice and find the positions n1, n2, and n3 of all the points where the electricity price increases. If there are only two points where the electricity price increases, then n3 = 25; determine [1, n1-1] as the electricity price valley period, [n1, n2-1] as the electricity price normal period, [n2, n3-1] as the electricity price peak period, and [n3, 24] as the electricity price peak period; S25: Find all P PV >P load Calculate the next day's P PV >P load The remaining energy E after photovoltaic power supplies the load at the moment free ; S26: Calculate [n3,24] time period P PV <P load The total energy E after deducting the photovoltaic power from the load power at the moment load ; S27: Calculate the discharge amount E1 of the energy storage [n2, n3-1].

5. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 4 is characterized by: In S25, the remaining energy E free The calculation formula is: Where dt is the time interval in h.

6. A multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 4 or 5, characterized in that: In S26, the total energy E load The calculation formula is:

7. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 4 or 5, characterized in that: In the step S27, the calculation formula of the discharge amount E1 is: E1=E use -HAVE BEEN load Among them, E use is the available capacity of the energy storage device, in kWh, and the formula is: IN use =E max *(SOC max -SOC min ) Among them, E max is the maximum energy storage capacity in kWh, SOC max and SOC min are the maximum and minimum states of charge of the energy storage device, respectively.

8. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 1 or 2, characterized in that: The S3 comprises the following steps, S31: input the data processed by S2 and the energy of each time period calculated; S32: Determine whether E free <E use , if so, then the energy storage power P in the period [1,n1-1] es =-P max , where P max is the maximum charge and discharge power of the energy storage device. The energy storage device is charged until the energy of the energy storage device reaches (E use -E free ), otherwise [1,n1-1] time period P es =0, energy storage does not work; S33: i=n1, start loop; S34: Determine whether P PV (i)<P load (i) If so, then P es (i)=-(P PV (i)-P load (i)), otherwise, go to S35; S35: Determine whether i∈[n1,n2-1], if so, let P es (i) = 0, the energy storage does not work, otherwise, determine whether i∈[n2,n3-1], if so, let P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the [n2,n3-1] period reaches E1, otherwise, P es (i) = min[(P load (i)-P PV (i)),P max ], the energy storage discharges until the total energy released in the period n3 to 24 reaches E load ; S36: Determine whether i < 24, if so, set i = i + 1 and return to S34, otherwise, exit the loop; S37: So far, the energy storage power P of all time periods is obtained es , P es Reorder according to idx0, obtain the energy storage power sorted by time and output it.

9. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 1 or 2, characterized in that: The S4 comprises the following steps, S41: Perform rolling forecasts on the photovoltaic power and load power at the next moment every 10 minutes during the day, and the forecast model still uses the LSTM model; S42: [1, n1-1] The energy storage power in the corresponding period is issued according to the day-ahead instruction; S43: [n2-1,24] The energy storage power command for the corresponding period is adjusted point by point according to the predicted photovoltaic power and load power during the day. The power command formula is the same as the day-ahead dispatch.

10. The multi-time scale source-load-storage collaborative optimization scheduling method based on model prediction according to claim 9 is characterized by: In S41, the input parameters of the photovoltaic power intraday prediction model are the temperature, humidity, light intensity, humidity and photovoltaic output power 24 hours before the prediction time obtained according to the weather forecast, and the output parameter is the photovoltaic output power at the prediction time; the input parameters of the load intraday prediction model are the load power 24 hours before the prediction time, and the output parameter is the load power at the prediction time.

Images