Photovoltaic energy storage adjusting method and system based on machine learning

Abstract

The invention relates to the technical field of photovoltaic energy storage adjustment, in particular to a photovoltaic energy storage adjustment method and system based on machine learning. Comprising the following steps: constructing a photovoltaic power prediction model based on historical monitoring data, and generating an expected power curve in a single adjustment period according to the photovoltaic power prediction model; setting a plurality of monitoring sub-cycles according to the expected power curve, and setting a primary energy storage strategy according to all the monitoring sub-cycles; a deviation evaluation value is generated according to the preset feedback time node, and whether a correction instruction is generated or not is judged according to the deviation evaluation value. A plurality of monitoring sub-cycles are set based on an expected power curve, a proper energy storage point is selected to perform charge and discharge control by analyzing a power fluctuation state in each monitoring sub-cycle, and early warning is performed on prediction deviation of a photovoltaic prediction model in time by analyzing actual photovoltaic power and environmental parameters. The overall loss of the energy storage system caused by photovoltaic power fluctuation is reduced, and the service life of the energy storage system is prolonged.

Description

Technical Field

[0001] The present application relates to the field of photovoltaic energy storage regulation technology, and in particular to a photovoltaic energy storage regulation method and system based on machine learning. Background Art

[0002] With the rapid development of renewable energy, the proportion of photovoltaic power generation in the power system continues to increase. However, photovoltaic power generation is intermittent and fluctuating, and its output power is significantly affected by factors such as weather and sunlight intensity, resulting in challenges to grid stability. To alleviate this problem, energy storage systems are widely used to smooth power fluctuations and achieve peak load shifting.

[0003] However, since photovoltaic power generation forecasts are affected by sudden weather changes, traditional point prediction methods have large errors, causing the energy storage strategy to deviate from the optimal value. At the same time, it will also cause excessive charging and discharging of the energy storage system, aggravate the loss of energy storage equipment, and greatly reduce the service life of the battery in the energy storage system. Summary of the Invention

[0004] The purpose of this application is: to solve the above technical problems, this application provides a photovoltaic energy storage regulation method and system based on machine learning, aiming to improve the regulation efficiency of the photovoltaic energy storage system.

[0005] In some embodiments of the present application, a photovoltaic energy storage regulation method based on machine learning is provided, comprising: Build a photovoltaic power prediction model based on historical monitoring data, and generate the expected power curve within a single regulation cycle according to the photovoltaic power prediction model; Set multiple monitoring sub-cycles based on the expected power curve, and set the first-level energy storage strategy based on all monitoring sub-cycles; A deviation evaluation value is generated according to a preset feedback time node, and whether a correction instruction is generated is determined based on the deviation evaluation value.

[0006] In some embodiments of the present application, when multiple monitoring sub-periods are set according to the expected power curve, the following steps are included: Set multiple auxiliary indicators based on historical monitoring data; Get the expected environment data package for the current adjustment cycle; Generate expected fluctuation curves of various auxiliary indicators based on the expected environmental data package; Generate monitoring evaluation value b based on all expected fluctuation curves and expected power curves; Set the power difference w according to the monitoring evaluation value b; Setting multiple monitoring sub-cycles according to the power difference w and the expected power curve; Establish monitoring sub-period sequence A, A=(a1, a2…a i …a n ), where ai is the i-th monitoring sub-cycle; n is the number of monitoring sub-cycles.

[0007] In some embodiments of the present application, when generating the monitoring evaluation value b, the process includes: b=e1*Q1*[ β 1i *p i ]+e2*Q2*[ β 2i *s i ]; Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; θ1 is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; p i is the volatility evaluation value generated based on the expected volatility curve of the i-th auxiliary indicator; is the number of predicted evaluation indicators; β 2i is the influencing factor of the i-th prediction evaluation index; s i The reference value of the i-th prediction evaluation index generated based on the expected fluctuation curve of the current adjustment period.

[0008] In some embodiments of the present application, when setting the first-level energy storage strategy based on all monitoring sub-cycles, it includes: Set multiple energy storage points according to energy storage equipment parameters; Establish the energy storage point sequence C, C=(c1,c2…c i …c m ), where c i is the i-th energy storage point; m is the number of energy storage points; According to the monitoring sub-cycle sequence A, the i-th monitoring sub-cycle is set as the target sub-cycle; Generate the expected loss value v of the target sub-cycle; Set the screening sub-strategy for the target sub-period according to the expected loss value v; Obtaining the expected grid demand for the target sub-cycle; Setting the energy storage sub-strategy for the target sub-cycle based on the screening sub-strategy and the expected demand of the power grid; The energy storage sub-strategy for each monitoring sub-cycle is set in sequence, and a first-level energy storage strategy is generated based on all energy storage sub-strategies.

[0009] In some embodiments of the present application, generating the expected loss value v of the target sub-cycle includes: generating a power sub-curve of a target sub-cycle according to the expected power curve; generating an environmental sub-data packet of a target sub-period according to the expected environmental data packet; Generate an expected loss value v based on the environment sub-data packet and the power sub-curve; v= η i *j i ]; in, is the number of loss evaluation indicators; η i is the influencing factor of the i-th loss evaluation index; j i is the reference value of the i-th loss evaluation index in the target sub-cycle.

[0010] In some embodiments of the present application, when setting a screening sub-strategy, it includes: Generate equipment evaluation values for each energy storage point within the target sub-cycle; Establish equipment evaluation value series D, D=(d1, d2…d i …d m ), where d i is the equipment evaluation value of the i-th energy storage point in the target sub-cycle; m is the number of energy storage points; According to the expected loss value v set the target sub-cycle equipment evaluation value threshold D1; Set the number of backup energy storage points u according to the expected loss value v; If d i >D1, set the i-th energy storage point as the energy storage point to be controlled within the target sub-cycle; A screening sub-strategy for the target sub-cycle is generated according to the number u of backup energy storage points and all energy storage points to be controlled in the target sub-cycle.

[0011] In some embodiments of the present application, determining whether to generate a correction instruction based on a deviation evaluation value includes: Get the monitoring evaluation value b of the current monitoring sub-cycle: Set multiple feedback time nodes according to the monitoring evaluation value b; Generate the deviation evaluation value f of the current feedback time node; f=e3*Q3*[ β 1i *(k i -k' i )2]+e4*Q4*H; Among them, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth coefficient; is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; k i is the actual reference value of the i-th auxiliary indicator at the current feedback time node; k' iis the expected reference value of the i-th auxiliary indicator at the current feedback time node; H is the power prediction deviation value at the current feedback time node; Preset deviation evaluation value threshold F1; If f>F1, the current feedback time node generates a first-level correction instruction.

[0012] In some embodiments of the present application, a photovoltaic energy storage regulation system based on machine learning is provided, comprising: A prediction unit is used to build a photovoltaic power prediction model based on historical monitoring data, and generate an expected power curve within a single regulation cycle according to the photovoltaic power prediction model; The central control unit is used to set multiple monitoring sub-cycles according to the expected power curve and set the first-level energy storage strategy based on all monitoring sub-cycles; A correction unit, configured to generate a deviation evaluation value according to a preset feedback time node, and determine whether to generate a correction instruction according to the deviation evaluation value; Wherein, the central control unit includes: A first processing module is used to set multiple auxiliary indicators based on historical monitoring data; Get the expected environment data package for the current adjustment cycle; Generate expected fluctuation curves of various auxiliary indicators based on the expected environmental data package; Generate monitoring evaluation value b based on all expected fluctuation curves and expected power curves; Set the power difference w according to the monitoring evaluation value b; Setting multiple monitoring sub-cycles according to the power difference w and the expected power curve; Establish monitoring sub-period sequence A, A=(a1, a2…a i …a n ), where a i is the i-th monitoring sub-cycle; n is the number of monitoring sub-cycles; Among them, when generating the monitoring evaluation value b, it includes: b=e1*Q1*[ β 1i *p i ]+e2*Q2*[ β 2i *s i ]; Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; θ1 is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; p i is the volatility evaluation value generated based on the expected volatility curve of the i-th auxiliary indicator; is the number of predicted evaluation indicators; β2i is the influencing factor of the i-th prediction evaluation index; s i The reference value of the i-th prediction evaluation index generated based on the expected fluctuation curve of the current adjustment period.

[0013] In some embodiments of the present application, the central control unit further includes: A second processing module is used to set multiple energy storage points according to energy storage device parameters; Establish the energy storage point sequence C, C=(c1,c2…c i …c m ), where c i is the i-th energy storage point; m is the number of energy storage points; According to the monitoring sub-cycle sequence A, the i-th monitoring sub-cycle is set as the target sub-cycle; Generate the expected loss value v of the target sub-cycle; Set the screening sub-strategy for the target sub-period according to the expected loss value v; Obtaining the expected grid demand for the target sub-cycle; Setting the energy storage sub-strategy for the target sub-cycle based on the screening sub-strategy and the expected demand of the power grid; The energy storage sub-strategy for each monitoring sub-cycle is set in sequence, and a first-level energy storage strategy is generated based on all energy storage sub-strategies.

[0014] In some embodiments of the present application, the correction unit includes: The first correction module is used to obtain the monitoring evaluation value b of the current monitoring sub-cycle and set multiple feedback time nodes according to the monitoring evaluation value b; The second correction module is used to generate a deviation evaluation value f of the current feedback time node; f=e3*Q3*[ β 1i *(k i -k' i )2]+e4*Q4*H; Among them, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth coefficient; is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; k i is the actual reference value of the i-th auxiliary indicator at the current feedback time node; k' i is the expected reference value of the i-th auxiliary indicator at the current feedback time node; H is the power prediction deviation value at the current feedback time node; The third correction module is used to preset the deviation evaluation value threshold F1; If f>F1, the third correction module generates a first-level correction instruction for the current feedback time node.

[0015] Compared with the prior art, the photovoltaic energy storage regulation method and system based on machine learning in the embodiment of the present application has the following beneficial effects: Based on the expected power curve, multiple monitoring sub-cycles are set. By analyzing the power fluctuation status within each monitoring sub-cycle, appropriate energy storage points are selected for charging and discharging control, realizing overall regulation of the energy storage system, reducing the overall loss of the energy storage system caused by photovoltaic power fluctuations, and improving the operating life of the energy storage system.

[0016] By analyzing actual photovoltaic power and environmental parameters, timely warnings can be issued for prediction deviations of the photovoltaic prediction model, and periodic monitoring and adjustment of the working status of the energy storage system can be achieved. The output or storage of electric energy can be controlled, the efficiency of smoothing photovoltaic output fluctuations can be improved, and the stable operation of the system can be guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a photovoltaic energy storage regulation method based on machine learning in the preferred embodiment of the present application. DETAILED DESCRIPTION

[0018] The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0019] In the description of this application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply 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 a limitation on this application.

[0020] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.

[0021] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances.

[0022] like Figure 1 As shown, a photovoltaic energy storage regulation method based on machine learning in a preferred embodiment of the present application is characterized by comprising: S101: Build a photovoltaic power prediction model based on historical monitoring data, and generate an expected power curve within a single regulation cycle according to the photovoltaic power prediction model; S102: Setting multiple monitoring sub-cycles according to the expected power curve, and setting a primary energy storage strategy according to all monitoring sub-cycles; S103: Generate a deviation evaluation value according to a preset feedback time node, and determine whether to generate a correction instruction based on the deviation evaluation value.

[0023] Specifically, a single regulation cycle is preferably one day, and the duration of the regulation cycle can be set according to actual operating parameters of the photovoltaic field.

[0024] Specifically, when multiple monitoring sub-cycles are set according to the expected power curve, it includes: Set multiple auxiliary indicators based on historical monitoring data; Get the expected environment data package for the current adjustment cycle; Generate expected fluctuation curves of various auxiliary indicators based on the expected environmental data package; Generate monitoring evaluation value b based on all expected fluctuation curves and expected power curves; Set the power difference w according to the monitoring evaluation value b; Setting multiple monitoring sub-cycles according to the power difference w and the expected power curve; Establish monitoring sub-period sequence A, A=(a1, a2…a i …a n ), where a i is the i-th monitoring sub-cycle; n is the number of monitoring sub-cycles.

[0025] Specifically, the larger the monitoring evaluation value is, the greater the possibility of power fluctuation of the photovoltaic system in the current regulation cycle is, and the smaller the corresponding power difference is.

[0026] Specifically, the expected power value z1 of the start time node of the current adjustment cycle is obtained, and the first point that differs from the expected power value z1 by a single power difference is found on the expected power curve, and the point is set as the first end time node. A monitoring sub-cycle is established according to the time interval between the start time node and the first end time node. Then, the expected power value z2 of the first end time node is obtained, and the first point that differs from the expected power value z2 by a single power difference is found on the expected power curve as the second end point. A monitoring sub-cycle is established according to the time interval between the first end time node and the second end time node, and so on, until the end time node of the current adjustment cycle is reached, and all monitoring sub-cycles are obtained to construct a monitoring sub-cycle sequence.

[0027] Specifically, when generating the monitoring evaluation value b, it includes: b=e1*Q1*[ β 1i *p i ]+e2*Q2*[ β 2i *s i ]; Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; θ1 is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; p i is the volatility evaluation value generated based on the expected volatility curve of the i-th auxiliary indicator; is the number of predicted evaluation indicators; β 2i is the influencing factor of the i-th prediction evaluation index; s i The reference value of the i-th prediction evaluation index generated based on the expected fluctuation curve of the current adjustment period.

[0028] Specifically, auxiliary indicators include but are not limited to temperature, light intensity, cloud thickness, humidity, rain and snow, and other parameters that affect the power generation of the photovoltaic system. Corresponding influencing factors are set according to the degree of influence of each indicator on the power generation of the photovoltaic system. The greater the degree of influence, the greater the corresponding influencing factor.

[0029] Specifically, the fluctuation evaluation value refers to the expected curve of each auxiliary indicator in the current adjustment cycle generated based on the expected environmental data package of the current adjustment cycle collected. By evenly dividing the expected curve of a single auxiliary indicator, the expected value in each divided interval is obtained, and the corresponding fluctuation evaluation value is set according to the variance of all expected values. The larger the variance, the larger the corresponding fluctuation evaluation value. The larger the fluctuation evaluation value, the greater the possibility that the corresponding auxiliary indicator will fluctuate violently in the current adjustment cycle.

[0030] Specifically, the prediction and evaluation indicators include, but are not limited to, the confidence level of the photovoltaic prediction model, the fluctuation evaluation of the expected power curve, and other parameters. The corresponding impact factor is set according to the influence of each prediction and evaluation indicator on the energy storage strategy. The greater the influence, the greater the corresponding impact factor.

[0031] Specifically, all parameters in the model are normalized by presetting the third fixed coefficient and the fourth fixed coefficient, so that each parameter in the model is in the same value range.

[0032] In a preferred embodiment of the present application, when setting the first-level energy storage strategy according to all monitoring sub-cycles, it includes: Set multiple energy storage points according to energy storage equipment parameters; Establish the energy storage point sequence C, C=(c1,c2…c i …c m ), where c i is the i-th energy storage point; m is the number of energy storage points; According to the monitoring sub-cycle sequence A, the i-th monitoring sub-cycle is set as the target sub-cycle; Generate the expected loss value v of the target sub-cycle; Set the screening sub-strategy for the target sub-period according to the expected loss value v; Obtaining the expected grid demand for the target sub-cycle; Setting the energy storage sub-strategy for the target sub-cycle based on the screening sub-strategy and the expected demand of the power grid; The energy storage sub-strategy for each monitoring sub-cycle is set in sequence, and a first-level energy storage strategy is generated based on all energy storage sub-strategies.

[0033] Specifically, the energy storage system adopts a distributed energy storage method, with multiple energy storage batteries installed inside. Multiple energy storage points are set according to the equipment parameters of the energy storage batteries, and a single energy storage point represents an energy storage battery.

[0034] Specifically, the larger the expected loss value, the greater the possibility of power fluctuations in the target sub-cycle, and the greater the possibility of overcharging and discharging during the charging and discharging process.

[0035] Specifically, when generating the expected loss value v of the target sub-cycle, it includes: generating a power sub-curve of a target sub-cycle according to the expected power curve; generating an environmental sub-data packet of a target sub-period according to the expected environmental data packet; Generate an expected loss value v based on the environment sub-data packet and the power sub-curve; v= η i *j i ]; in, is the number of loss evaluation indicators; η i is the influencing factor of the i-th loss evaluation index; j i is the reference value of the i-th loss evaluation index in the target sub-cycle.

[0036] Specifically, the loss evaluation indicators include but are not limited to power fluctuation difference, power fluctuation probability, charge and discharge amount change difference and other parameters. The corresponding impact factor is set according to the influence of each loss evaluation indicator on the loss of the energy storage battery. The greater the influence, the greater the corresponding impact factor.

[0037] Specifically, when setting the screening sub-strategy, include: Generate equipment evaluation values for each energy storage point within the target sub-cycle; Establish equipment evaluation value series D, D=(d1, d2…d i …d m ), where d i is the equipment evaluation value of the i-th energy storage point in the target sub-cycle; m is the number of energy storage points; According to the expected loss value v set the target sub-cycle equipment evaluation value threshold D1; Set the number of backup energy storage points u according to the expected loss value v; If d i >D1, set the i-th energy storage point as the energy storage point to be controlled within the target sub-cycle; A screening sub-strategy for the target sub-cycle is generated according to the number u of backup energy storage points and all energy storage points to be controlled in the target sub-cycle.

[0038] Specifically, a corresponding equipment evaluation value is generated based on multiple parameters such as the historical operating time and failure probability of the energy storage battery at each equipment point. The larger the equipment evaluation value, the longer the expected life of the energy storage battery corresponding to the equipment point, the lower the failure rate, and the better the operating status.

[0039] Specifically, the greater the expected loss value, the larger the corresponding number of backup energy storage points. By setting up multiple backup energy storage points, when the photovoltaic power fluctuates violently, the backup energy storage points can be started in time for charging and discharging, avoiding the impact of photovoltaic power fluctuations on the energy storage system, improving the efficiency of smoothing photovoltaic output fluctuations, and ensuring the stable operation of the system.

[0040] It is understandable that in the above embodiment, multiple monitoring sub-cycles are set based on the expected power curve, and by analyzing the power fluctuation state within each monitoring sub-cycle, appropriate energy storage points are selected for charge and discharge control, thereby achieving overall regulation of the energy storage system, reducing the overall loss of the energy storage system caused by photovoltaic power fluctuations, and improving the operating life of the energy storage system. In a preferred embodiment of the present application, when determining whether to generate a correction instruction based on the deviation evaluation value, the method includes: Get the monitoring evaluation value b of the current monitoring sub-cycle: Set multiple feedback time nodes according to the monitoring evaluation value b; Generate the deviation evaluation value f of the current feedback time node; f=e3*Q3*[ β 1i *(k i -k' i )2]+e4*Q4*H; Among them, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth coefficient; is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; k i is the actual reference value of the i-th auxiliary indicator at the current feedback time node; k' i is the expected reference value of the i-th auxiliary indicator at the current feedback time node; H is the power prediction deviation value at the current feedback time node; Preset deviation evaluation value threshold F1; If f>F1, the current feedback time node generates a first-level correction instruction.

[0041] Specifically, the expected reference value is set according to the expected environmental parameters of the current adjustment period.

[0042] Specifically, the deviation evaluation value threshold can be set according to historical parameters.

[0043] Specifically, all parameters in the model are normalized by presetting the third fixed coefficient and the fourth fixed coefficient, so that each parameter in the model is in the same value range. Specifically, the first-level correction instruction corrects the expected power curve in the current regulation cycle and updates the first-level energy storage strategy of the current regulation cycle according to the correction result.

[0044] It is understandable that in the above embodiment, by analyzing the actual photovoltaic power and environmental parameters, timely warnings are issued for the prediction deviations of the photovoltaic prediction model, and periodic monitoring and adjustment of the working status of the energy storage system are achieved. The output or storage of electric energy is controlled, the efficiency of smoothing photovoltaic output fluctuations is improved, and the stable operation of the system is ensured.

[0045] Based on another preferred embodiment of a photovoltaic energy storage regulation method based on machine learning in any of the above preferred embodiments, this preferred embodiment provides a photovoltaic energy storage regulation system based on machine learning, including: A prediction unit is used to build a photovoltaic power prediction model based on historical monitoring data, and generate an expected power curve within a single regulation cycle according to the photovoltaic power prediction model; The central control unit is used to set multiple monitoring sub-cycles according to the expected power curve and set the first-level energy storage strategy based on all monitoring sub-cycles; A correction unit, configured to generate a deviation evaluation value according to a preset feedback time node, and determine whether to generate a correction instruction according to the deviation evaluation value; The central control unit includes: A first processing module is used to set multiple auxiliary indicators based on historical monitoring data; Get the expected environment data package for the current adjustment cycle; Generate expected fluctuation curves of various auxiliary indicators based on the expected environmental data package; Generate monitoring evaluation value b based on all expected fluctuation curves and expected power curves; Set the power difference w according to the monitoring evaluation value b; Setting multiple monitoring sub-cycles according to the power difference w and the expected power curve; Establish monitoring sub-period sequence A, A=(a1, a2…a i …a n ), where a i is the i-th monitoring sub-cycle; n is the number of monitoring sub-cycles; Among them, when generating the monitoring evaluation value b, it includes: b=e1*Q1*[ β 1i *p i ]+e2*Q2*[ β 2i *s i ]; Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; θ1 is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; p i is the volatility evaluation value generated based on the expected volatility curve of the i-th auxiliary indicator; is the number of predicted evaluation indicators; β 2i is the influencing factor of the i-th prediction evaluation index; s i The reference value of the i-th prediction evaluation index generated based on the expected fluctuation curve of the current adjustment period.

[0046] In a preferred embodiment of the present application, the central control unit further includes: A second processing module is used to set multiple energy storage points according to energy storage device parameters; Establish the energy storage point sequence C, C=(c1,c2…c i …c m ), where c i is the i-th energy storage point; m is the number of energy storage points; According to the monitoring sub-cycle sequence A, the i-th monitoring sub-cycle is set as the target sub-cycle; Generate the expected loss value v of the target sub-cycle; Set the screening sub-strategy for the target sub-period according to the expected loss value v; Obtaining the expected grid demand for the target sub-cycle; Setting the energy storage sub-strategy for the target sub-cycle based on the screening sub-strategy and the expected demand of the power grid; The energy storage sub-strategy for each monitoring sub-cycle is set in sequence, and a first-level energy storage strategy is generated based on all energy storage sub-strategies.

[0047] In a preferred embodiment of the present application, the correction unit includes: The first correction module is used to obtain the monitoring evaluation value b of the current monitoring sub-cycle and set multiple feedback time nodes according to the monitoring evaluation value b; The second correction module is used to generate a deviation evaluation value f of the current feedback time node; f=e3*Q3*[ β 1i *(k i -k' i )2]+e4*Q4*H; Among them, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth coefficient; is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; k i is the actual reference value of the i-th auxiliary indicator at the current feedback time node; k' i is the expected reference value of the i-th auxiliary indicator at the current feedback time node; H is the power prediction deviation value at the current feedback time node; The third correction module is used to preset the deviation evaluation value threshold F1; If f>F1, the third correction module generates a first-level correction instruction for the current feedback time node.

[0048] According to the first concept of the present application, multiple monitoring sub-cycles are set based on the expected power curve. By analyzing the power fluctuation state within each monitoring sub-cycle, appropriate energy storage points are selected for charging and discharging control, thereby achieving overall regulation of the energy storage system, reducing the overall loss of the energy storage system caused by photovoltaic power fluctuations, and improving the operating life of the energy storage system.

[0049] According to the second concept of this application, by analyzing the actual photovoltaic power and environmental parameters, timely warning of the prediction deviation of the photovoltaic prediction model is issued, and periodic monitoring and adjustment of the working status of the energy storage system is achieved, the output or storage of electric energy is controlled, the efficiency of smoothing photovoltaic output fluctuations is improved, and the stable operation of the system is ensured.

[0050] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and replacements can be made without departing from the technical principles of the present application. These improvements and replacements should also be regarded as the scope of protection of the present application.

Claims

1. A photovoltaic energy storage regulation method based on machine learning, characterized in that: include: Build a photovoltaic power prediction model based on historical monitoring data, and generate the expected power curve within a single regulation cycle according to the photovoltaic power prediction model; Set multiple monitoring sub-cycles based on the expected power curve, and set the first-level energy storage strategy based on all monitoring sub-cycles; A deviation evaluation value is generated according to a preset feedback time node, and whether a correction instruction is generated is determined based on the deviation evaluation value.

2. The photovoltaic energy storage regulation method based on machine learning according to claim 1, characterized in that: When multiple monitoring sub-cycles are set according to the expected power curve, including: Set multiple auxiliary indicators based on historical monitoring data; Get the expected environment data package for the current adjustment cycle; Generate expected fluctuation curves of various auxiliary indicators based on the expected environmental data package; Generate monitoring evaluation value b based on all expected fluctuation curves and expected power curves; Set the power difference w according to the monitoring evaluation value b; Setting multiple monitoring sub-cycles according to the power difference w and the expected power curve; Establish monitoring sub-period sequence A, A=(a1, a2…a i …a n ), where a i is the i-th monitoring sub-cycle; n is the number of monitoring sub-cycles.

3. The photovoltaic energy storage regulation method based on machine learning according to claim 2, characterized in that: When generating the monitoring evaluation value b, it includes: b=e1*Q1*[ β 1i *p i ]+e2*Q2*[ β 2i *s i ]: Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; θ1 is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; p i is the volatility evaluation value generated based on the expected volatility curve of the i-th auxiliary indicator; is the number of predicted evaluation indicators; β 2i is the influencing factor of the i-th prediction evaluation index; s i The reference value of the i-th prediction evaluation index generated based on the expected fluctuation curve of the current adjustment period.

4. The photovoltaic energy storage regulation method based on machine learning according to claim 3, characterized in that: When setting the first-level energy storage strategy based on all monitoring sub-cycles, it includes: Set multiple energy storage points according to energy storage equipment parameters; Establish the energy storage point sequence C, C=(c1,c2…c i …c m ), where c i is the i-th energy storage point; m is the number of energy storage points; According to the monitoring sub-cycle sequence A, the i-th monitoring sub-cycle is set as the target sub-cycle; Generate the expected loss value v of the target sub-cycle; Set the screening sub-strategy for the target sub-period according to the expected loss value v; Obtaining the expected grid demand for the target sub-cycle; Setting the energy storage sub-strategy for the target sub-cycle based on the screening sub-strategy and the expected demand of the power grid; The energy storage sub-strategy for each monitoring sub-cycle is set in sequence, and a first-level energy storage strategy is generated based on all energy storage sub-strategies.

5. The photovoltaic energy storage regulation method based on machine learning according to claim 4, characterized in that: When generating the expected loss value v of the target sub-cycle, it includes: generating a power sub-curve of a target sub-cycle according to the expected power curve; generating an environmental sub-data packet of a target sub-period according to the expected environmental data packet; Generate an expected loss value v based on the environment sub-data packet and the power sub-curve; v= or i *j i ]; in, is the number of loss evaluation indicators; η i is the influencing factor of the i-th loss evaluation index; j i is the reference value of the i-th loss evaluation index in the target sub-cycle.

6. The photovoltaic energy storage regulation method based on machine learning according to claim 4, characterized in that: When setting a filter sub-strategy, include: Generate equipment evaluation values for each energy storage point within the target sub-cycle; Establish equipment evaluation value series D, D=(d1, d2…d i …d m ), where d i is the equipment evaluation value of the i-th energy storage point in the target sub-cycle; m is the number of energy storage points; According to the expected loss value v set the target sub-cycle equipment evaluation value threshold D1; Set the number of backup energy storage points u according to the expected loss value v; If d i >D1, set the i-th energy storage point as the energy storage point to be controlled within the target sub-cycle; A screening sub-strategy for the target sub-cycle is generated according to the number u of backup energy storage points and all energy storage points to be controlled in the target sub-cycle.

7. The photovoltaic energy storage regulation method based on machine learning according to claim 6, characterized in that: When determining whether to generate a correction instruction based on the deviation evaluation value, it includes: Get the monitoring evaluation value b of the current monitoring sub-cycle: Set multiple feedback time nodes according to the monitoring evaluation value b; Generate the deviation evaluation value f of the current feedback time node; f=e3*Q3*[ β 1i *(k i -k' i )2]+e4*Q4*H; Among them, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth coefficient; is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; k i is the actual reference value of the i-th auxiliary indicator at the current feedback time node; k' i is the expected reference value of the i-th auxiliary indicator at the current feedback time node; H is the power prediction deviation value at the current feedback time node; Preset deviation evaluation value threshold F1; If f>F1, the current feedback time node generates a first-level correction instruction.

8. A photovoltaic energy storage regulation system based on machine learning, adopting the photovoltaic energy storage regulation method based on machine learning according to any one of claims 1 to 7, characterized in that: include: A prediction unit is used to build a photovoltaic power prediction model based on historical monitoring data, and generate an expected power curve within a single regulation cycle according to the photovoltaic power prediction model; The central control unit is used to set multiple monitoring sub-cycles according to the expected power curve and set the first-level energy storage strategy based on all monitoring sub-cycles; A correction unit, configured to generate a deviation evaluation value according to a preset feedback time node, and determine whether to generate a correction instruction according to the deviation evaluation value; Wherein, the central control unit includes: A first processing module is used to set multiple auxiliary indicators based on historical monitoring data; Get the expected environment data package for the current adjustment cycle; Generate expected fluctuation curves of various auxiliary indicators based on the expected environmental data package; Generate monitoring evaluation value b based on all expected fluctuation curves and expected power curves; Set the power difference w according to the monitoring evaluation value b; Setting multiple monitoring sub-cycles according to the power difference w and the expected power curve; Establish monitoring sub-period sequence A, A=(a1, a2…a i …a n ), where a i is the i-th monitoring sub-cycle; n is the number of monitoring sub-cycles; Among them, when generating the monitoring evaluation value b, it includes: b=e1*Q1*[ β 1i *p i ]+e2*Q2*[ β 2i *s i ]: Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; θ1 is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; p i is the volatility evaluation value generated based on the expected volatility curve of the i-th auxiliary indicator; is the number of predicted evaluation indicators; β 2i is the influencing factor of the i-th prediction evaluation index; s i The reference value of the i-th prediction evaluation index generated based on the expected fluctuation curve of the current adjustment period.

9. The photovoltaic energy storage regulation system based on machine learning according to claim 8, characterized in that: The central control unit further includes: A second processing module is used to set multiple energy storage points according to energy storage device parameters; Establish the energy storage point sequence C, C=(c1,c2…c i …c m ), where c i is the i-th energy storage point; m is the number of energy storage points; According to the monitoring sub-cycle sequence A, the i-th monitoring sub-cycle is set as the target sub-cycle; Generate the expected loss value v of the target sub-cycle; Set the screening sub-strategy for the target sub-period according to the expected loss value v; Obtaining the expected grid demand for the target sub-cycle; Setting the energy storage sub-strategy for the target sub-cycle based on the screening sub-strategy and the expected demand of the power grid; The energy storage sub-strategy for each monitoring sub-cycle is set in sequence, and a first-level energy storage strategy is generated based on all energy storage sub-strategies.

10. The photovoltaic energy storage regulation system based on machine learning according to claim 9, characterized in that: The correction unit includes: The first correction module is used to obtain the monitoring evaluation value b of the current monitoring sub-cycle and set multiple feedback time nodes according to the monitoring evaluation value b; The second correction module is used to generate a deviation evaluation value f of the current feedback time node; f=e3*Q3*[ β 1i *(k i -k' i )2]+e4*Q4*H; Among them, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth coefficient; is the number of auxiliary indicators; β 1i is the impact factor of the i-th auxiliary indicator; k i is the actual reference value of the i-th auxiliary indicator at the current feedback time node; k' i is the expected reference value of the i-th auxiliary indicator at the current feedback time node; H is the power prediction deviation value at the current feedback time node; The third correction module is used to preset the deviation evaluation value threshold F1; If f>F1, the third correction module generates a first-level correction instruction for the current feedback time node.

Images