Intelligent control method and system of household photovoltaic energy storage equipment

Abstract

The application discloses a kind of intelligent control method and system of domestic photovoltaic energy storage equipment.The intelligent control method of domestic photovoltaic energy storage equipment includes: obtaining weather forecast information and real-time acquisition domestic photovoltaic energy storage equipment operating data;Predict to obtain photovoltaic power generation power short-time prediction result, user load power short-time prediction result and net power consumption short-time prediction result;Net power consumption prediction curve is constructed based on the net power consumption short-time prediction result;When energy storage battery state of charge reaches preset upper limit threshold and there is negative value area that satisfies dispatchable condition in the net power consumption prediction curve, each dispatchable negative value area in the net power consumption prediction curve is executed positive-negative area matching to determine whether to trigger intelligent control strategy.Such, the intelligent control method of domestic photovoltaic energy storage equipment can be more intelligent and accurately control domestic photovoltaic energy storage equipment according to climate, improve photovoltaic self-use rate, reduce "abandon light" phenomenon.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent control of photovoltaic energy storage equipment, in particular to an intelligent control method and system for household photovoltaic energy storage equipment. BACKGROUND

[0002] With the transformation of global energy structure and the rapid development of renewable energy, household photovoltaic energy storage equipment has been more and more widely used. Household photovoltaic energy storage equipment is usually composed of photovoltaic arrays, energy storage batteries, hybrid inverters, etc., which can store the electrical energy generated by photovoltaic power generation in energy storage batteries and release it when needed to power user loads. Household photovoltaic energy storage equipment plays an important role in improving energy self-sufficiency, reducing electricity costs, and improving power reliability.

[0003] However, in areas with weak power grid infrastructure, due to unstable power grid, household photovoltaic energy storage equipment is often set to always maintain the full charge state of the energy storage battery, so as to serve as an uninterruptible power supply (UPS) in case of power failure. Although this operating strategy ensures the reliability of power supply in case of power grid failure, it can cause serious "light abandonment" phenomenon when combined with photovoltaic power generation. That is, when the photovoltaic power generation exceeds the household power load, the excess clean energy cannot be effectively stored and is wasted, resulting in a huge waste of clean energy. The "light abandonment" phenomenon not only reduces the economic benefit of household photovoltaic energy storage equipment, but also hinders the further development of renewable energy.

[0004] The current control strategy of household photovoltaic energy storage equipment is mainly to control the charging and discharging of the energy storage battery according to pre-set rules or thresholds, such as starting charging when the energy storage battery is below a certain threshold, and stopping charging when the energy storage battery is above a certain threshold. This method is simple to implement, but the control accuracy is low and it is difficult to adapt to complex and variable photovoltaic power generation and load fluctuations. Especially in areas with variable climate and different user power consumption habits, simple threshold control strategy is difficult to achieve optimal energy scheduling, resulting in serious "light abandonment" phenomenon.

[0005] Therefore, there is an urgent need for a more intelligent, efficient and reliable household energy storage system control method to improve photovoltaic self-use rate and reduce "light abandonment" phenomenon. SUMMARY

[0006] An advantage of the present application is to provide an intelligent control method and system for household photovoltaic energy storage equipment, wherein the intelligent control method for household photovoltaic energy storage equipment can accurately control the household photovoltaic energy storage equipment according to the climate, improve the photovoltaic self-use rate, reduce the "light abandonment" phenomenon, and improve the economy and reliability of user power consumption.

[0007] According to one aspect of the present application, a method for intelligent control of a household photovoltaic energy storage device is provided, which comprises the steps of:

[0008] S110: obtaining weather forecast information and real-time collected operation data of the household photovoltaic energy storage device, wherein the operation data comprises photovoltaic power generation power, user load power and state of charge of the energy storage battery;

[0009] S120: predicting photovoltaic power generation power, user load power and net power consumption in a future preset time window based on the operation data and the weather forecast information to obtain photovoltaic power generation power short-time prediction results, user load power short-time prediction results and net power consumption short-time prediction results;

[0010] S130: constructing a net power consumption prediction curve based on the net power consumption short-time prediction results;

[0011] S140: when the state of charge of the energy storage battery reaches a preset upper limit threshold and there is a negative value area in the net power consumption prediction curve that meets a dispatchable condition, performing positive-negative area matching on each dispatchable negative value area in the net power consumption prediction curve to determine whether to trigger an intelligent control strategy, wherein the dispatchable condition is that the area of the negative value area is greater than a preset area threshold, or the duration of the negative value area is greater than a preset time threshold, or the absolute value of the minimum net power of the negative value area is greater than a preset power threshold.

[0012] In an embodiment of the method for intelligent control of the household photovoltaic energy storage device according to the present application, step S140 comprises the steps of: extracting the dispatchable negative value area from the net power consumption prediction curve; calculating the area of each dispatchable negative value area; searching forward from the starting time of each dispatchable negative value area to find a target positive value area that meets a preset area matching condition; and when the target positive value area is found, triggering an intelligent control strategy, wherein the intelligent control strategy comprises: controlling the energy storage battery to start discharging at the starting time of the target positive value area and controlling the energy storage battery to start charging at the starting time of the dispatchable negative value area.

[0013] In an embodiment of the method for intelligent control of the household photovoltaic energy storage device according to the present application, searching forward from the starting time of each dispatchable negative value area to find a target positive value area that meets a preset area matching condition comprises the steps of: finding a plurality of candidate positive value areas that meet the preset area matching condition; and using a dynamic programming or branch and bound algorithm to select an optimal one from the plurality of candidate positive value areas as the target positive value area.

[0014] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, step S140, after searching forward from the starting time of each schedulable negative value area to find a target positive value area that meets the preset area matching condition, comprises the step of: when no target positive value area corresponding to the schedulable negative value area is found in the current future preset time window, performing positive-negative area matching between the schedulable negative value area and a positive value area of another future preset time window.

[0015] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, extracting the schedulable negative value area from the net power consumption prediction curve comprises the steps of: extracting an original negative value area from the net power consumption prediction curve, the area of the original negative value area being greater than an ultra-large area threshold, the ultra-large area threshold being greater than the preset area threshold; splitting the original negative value area into a plurality of original negative value areas; and extracting the schedulable negative value area from the plurality of original negative value areas.

[0016] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, extracting the schedulable negative value area from the net power consumption prediction curve comprises the steps of: extracting a plurality of original negative value areas from the net power consumption prediction curve, the area of each original negative value area being less than the preset area threshold; and performing a merging operation on the plurality of original negative value areas to obtain the schedulable negative value area.

[0017] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, step S140 comprises the steps of: extracting the schedulable negative value area from the net power consumption prediction curve; extracting a target positive value area from the net power consumption prediction curve based on a positive-negative area dynamic balance analysis mechanism; and triggering an intelligent control strategy when the target positive value area is searched, wherein the intelligent control strategy comprises: controlling the energy storage battery to start discharging at the starting time of the target positive value area and controlling the energy storage battery to start charging at the starting time of the schedulable negative value area.

[0018] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, extracting a target positive value area from the net power consumption prediction curve based on a positive-negative area dynamic balance analysis mechanism comprises the steps of: sampling positive and negative value areas in the time sequence direction from the net power consumption curve to obtain an initial negative value sampling time sequence vector and an initial positive value sampling time sequence vector;

[0019] calculating a time window phase correlation matrix based on the initial negative value sampling time sequence vector and the initial positive value sampling time sequence vector;

[0020] calculating a two-dimension full-time fluctuation adjustment factor and a weighting vector based on the initial negative value sampling time sequence vector and the initial positive value sampling time sequence vector;

[0021] calculating an area of an optimized schedulable negative value region and an area of a target positive value region based on the two-dimension full-time fluctuation adjustment factor, the weighting vector, the time window conforming phase correlation matrix, the initial negative value sampling time sequence vector and the initial positive value sampling time sequence vector to extract the target positive value region from the net power consumption prediction curve based on the area of the target positive value region.

[0022] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, sampling positive and negative value regions from the net power consumption curve in the time sequence direction to obtain an initial negative value sampling time sequence vector and an initial positive value sampling time sequence vector comprises the following steps:

[0023] sampling the schedulable negative value region from the net power consumption curve in the time sequence direction to obtain an initial negative value sampling time sequence vector;

[0024] extracting an initial positive value region from the net power consumption prediction curve;

[0025] sampling the initial positive value region from the net power consumption curve in the time sequence direction to obtain an initial positive value sampling time sequence vector, wherein the number of samples of the initial positive value region is the same as the number of samples of the schedulable negative value region.

[0026] In an embodiment of the intelligent control method of the household photovoltaic energy storage device according to the present application, step S120 comprises the following steps: using a plurality of candidate prediction models to process the operation data and the weather forecast information respectively to obtain a set of candidate photovoltaic power short-time prediction results, a set of candidate user load power short-time prediction results and a set of candidate net power consumption short-time prediction results; and calculating the photovoltaic power short-time prediction result, the user load power short-time prediction result and the net power consumption short-time prediction result based on the set of candidate photovoltaic power short-time prediction results, the set of candidate user load power short-time prediction results and the set of candidate net power consumption short-time prediction results.

[0027] According to another aspect of the present application, the present application provides an intelligent control system of a household photovoltaic energy storage device, which comprises a processor, the processor comprising:

[0028] a data acquisition module configured to acquire weather forecast information and real-time operation data of a household photovoltaic energy storage device, wherein the operation data comprises photovoltaic power, user load power and energy storage battery state of charge;

[0029] a prediction module, configured to predict photovoltaic power, user load power and net power consumption in a preset future time window based on the operation data and the weather forecast information to obtain a photovoltaic power short-time prediction result, a user load power short-time prediction result and a net power consumption short-time prediction result;

[0030] a net power consumption curve construction module, configured to construct a net power consumption prediction curve based on the net power consumption short-time prediction result;

[0031] a control strategy triggering module, configured to perform positive-negative area matching on each dispatchable negative value area in the net power consumption prediction curve to determine whether to trigger an intelligent control strategy when the state of charge of the energy storage battery reaches a preset upper limit threshold and there is a dispatchable negative value area in the net power consumption prediction curve, wherein the dispatchable condition is that the area of the negative value area is greater than a preset area threshold, or the duration of the negative value area is greater than a preset time threshold, or the absolute value of the minimum net power of the negative value area is greater than a preset power threshold.

[0032] The further purposes and advantages of the present application will be fully apparent from the following description and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0033] The above and other purposes, features and advantages of the present application will become more apparent from the following detailed description of embodiments of the present application, taken in conjunction with the accompanying drawings. The drawings provided in the appended hereto are for purposes of illustration only and the present application is not limited to the methods and embodiments illustrated in the figures.

[0034] Figure 1 FIG. 1 illustrates a flowchart of an intelligent control method of a household photovoltaic energy storage device according to an embodiment of the present application.

[0035] Figure 2 FIG. 5 illustrates a flowchart of step S120 of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0036] Figure 3 FIG. 6 illustrates a flowchart of step S122 of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0037] Figure 4 FIG. 7 illustrates another flowchart of step S122 of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0038] Figure 5 FIG. 8 illustrates a flowchart of step S140 of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0039] Figure 6 FIG. 13 illustrates a flow chart of step S141A of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0040] Figure 7 FIG. 14 illustrates another flow chart of step S141A of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0041] Figure 8 FIG. 15 illustrates a flow chart of step S143A of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0042] Figure 9 FIG. 16 illustrates another flow chart of step S140 of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0043] Figure 10 FIG. 17 illustrates another flow chart of step S142B of the intelligent control method of the household photovoltaic energy storage device according to an embodiment of the present application.

[0044] Figure 11 FIG. 18 illustrates a structure block diagram of the intelligent control system of the household photovoltaic energy storage device according to an embodiment of the present application. DETAILED DESCRIPTION

[0045] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part but not all of the embodiments of the present application. It should be understood that the present application is not limited to the described embodiments.

[0046] It can be understood that the term "a" should be understood as "at least one" or "one or more", that is, in an embodiment, the number of one element can be one, and in another embodiment, the number of the element can be multiple, and the term "a" cannot be understood as a limitation on the number. "Multiple" refers to greater than or equal to two.

[0047] Although ordinal numbers such as "first", "second" and the like will be used in describing various components, they are used not to limit those components. The terms are used only to distinguish one component from another. For example, a first component can be referred to as a second component, and similarly, a second component can also be referred to as a first component without departing from the teachings of the present application. The term "and / or" as used herein includes any and all combinations of one or more associated listed items.

[0048] The terminology used herein is for the purpose of describing various embodiments only and is not intended to be limiting. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, operations, elements, or components, but do not preclude the presence or addition of one or more other features, integers, operations, elements, components, or groups thereof.

[0049] As described above, the current household photovoltaic energy storage device control strategy is mainly to control the charging and discharging of the energy storage battery according to the pre-set rules or thresholds, for example, starting charging when the energy storage battery power is lower than a certain threshold, and stopping charging when the energy storage battery power is higher than a certain threshold. This method is simple to implement, but the control precision is low, and it is difficult to adapt to the complex and variable photovoltaic power generation and load fluctuation. Especially in areas with variable climate and different user electricity behavior habits, the simple threshold control strategy is difficult to achieve optimal energy scheduling, resulting in the "abandoned light" phenomenon is still serious.

[0050] Therefore, there is an urgent need for a more intelligent, efficient and reliable household energy storage system control method to improve the photovoltaic self-use rate and reduce the "abandoned light" phenomenon.

[0051] Based on this, the present application proposes an intelligent control method and system for household photovoltaic energy storage equipment, aiming to provide a more refined and intelligent control strategy, by predicting the photovoltaic power generation and user load power in the future for a period of time, constructing the net power consumption curve, and innovatively proposing and applying the positive and negative area matching algorithm, combined with the triggering mechanism based on the grid state, the energy storage battery state and the adjustable period, intelligently scheduling the charging and discharging behavior of the energy storage battery, to overcome the shortcomings of the prior art. The intelligent control method and system for household photovoltaic energy storage equipment can maximize the self-use ratio of photovoltaic power generation while prioritizing user electricity reliability during power grid failure, significantly reducing the "abandoned light" phenomenon, and improving the economy and reliability of user electricity.

[0052] Correspondingly, as Figures 1 to 10 shown, the intelligent control method for household photovoltaic energy storage equipment according to the embodiments of the present application is illustrated. As Figure 1As shown, the intelligent control method of the household photovoltaic energy storage device comprises the steps of: S110, acquiring weather forecast information and real-time collected operation data of the household photovoltaic energy storage device, wherein the operation data comprises photovoltaic power generation power, user load power and energy storage battery state of charge; S120, predicting photovoltaic power generation power, user load power and net power consumption in a future preset time window based on the operation data and the weather forecast information to obtain photovoltaic power generation power short-time prediction results, user load power short-time prediction results and net power consumption short-time prediction results; S130, constructing a net power consumption prediction curve based on the net power consumption short-time prediction results; S140, when the energy storage battery state of charge reaches a preset upper limit threshold and there is a negative value region in the net power consumption prediction curve that meets a dispatchable condition, performing positive-negative area matching on each dispatchable negative value region in the net power consumption prediction curve to determine whether to trigger an intelligent control strategy, wherein the dispatchable condition is that the negative value region area is greater than a preset area threshold, or the negative value region duration is greater than a preset time threshold, or the absolute value of the minimum net power of the negative value region is greater than a preset power threshold.

[0053] Specifically, in step S110, weather forecast information and real-time collected operation data of the household photovoltaic energy storage device are acquired. Specifically, the real-time collected operation data of the household photovoltaic energy storage device includes but is not limited to photovoltaic power generation power P_pv(t), user load power P_load(t), energy storage battery state of charge (SOC), grid state and the like. The data collection frequency can be set according to requirements, for example, 1 minute or 15 minutes.

[0054] In step S120, photovoltaic power generation power, user load power and net power consumption in a future preset time window are predicted based on the operation data and the weather forecast information to obtain photovoltaic power generation power short-time prediction results, user load power short-time prediction results and net power consumption short-time prediction results. Specifically, a plurality of candidate prediction models can be used to predict the operation data of the household photovoltaic energy storage device in the future preset time window.

[0055] Correspondingly, in an embodiment of the present application, as Figure 2 As shown, step S120 comprises the steps of: S121, using a plurality of candidate prediction models to respectively process the operation data and the weather forecast information to obtain a set of candidate photovoltaic power generation power short-time prediction results, a set of candidate user load power short-time prediction results and a set of candidate net power consumption short-time prediction results; S122, calculating the photovoltaic power generation power short-time prediction results, the user load power short-time prediction results and the net power consumption short-time prediction results based on the set of candidate photovoltaic power generation power short-time prediction results, the set of candidate user load power short-time prediction results and the set of candidate net power consumption short-time prediction results.

[0056] The set of candidate short-term photovoltaic power prediction results, the set of candidate short-term user load power prediction results, and the set of candidate short-term net power consumption prediction results respectively refer to a set of candidate short-term photovoltaic power prediction results, a set of candidate short-term user load power prediction results, and a set of candidate short-term net power consumption prediction results in a future preset time window, and the future preset time window The preset time window can be designed according to requirements, for example, 1 hour, 2 hours, 24 hours, 25 hours, and the like.

[0057] Predicting the operation data of the household photovoltaic energy storage device in the future preset time window is the basis of the intelligent control strategy, and the prediction accuracy directly affects the control effect. The present application adopts a data-driven intelligent prediction method, uses historical data and external information to learn the change law of photovoltaic power generation and user load. Considering the prediction requirements in different data quantities and application scenarios, the present application adopts a prediction strategy of multiple prediction model fusion, wherein the multiple prediction models include but are not limited to: a statistical prediction model, a machine learning model, and a deep learning model.

[0058] Correspondingly, in an example of the present application, the plurality of candidate prediction models include: a statistical prediction model, a machine learning model, and a deep learning model. Each candidate prediction model can be selected according to requirements, for example, the statistical prediction model can be ARIMA, Prophet, etc., which is suitable for a scenario with less data quantity or higher real-time requirement, and is used as a fast start and benchmark prediction model of the system; the machine learning model can be XGBoost, LightGBM, RandomForest, SVR, etc., which is suitable for a scenario with moderate data quantity and certain prediction accuracy requirement, and can capture the nonlinear relationship in the data. The deep learning model can be LSTM, GRU, TCN, Transformer, etc., which is suitable for a scenario with large data quantity and extremely high prediction accuracy requirement, and can deeply mine the complex patterns and long-term dependence relationship in the data.

[0059] Accordingly, in an example of the present application, in step S121, based on the historical data collected in step S110 (including weather forecast information and real-time collected operating data of household photovoltaic energy storage equipment), a statistical prediction model is used to process the operating data and the weather forecast information to obtain a first candidate photovoltaic power generation short-term prediction result, a first candidate user load power short-term prediction result, and a first candidate net power consumption short-term prediction result, and a machine learning model is used to process the operating data and the weather forecast information to obtain a second candidate photovoltaic power generation short-term prediction result, a second candidate user load power short-term prediction result, and a second candidate net power consumption short-term prediction result, and a deep learning model is used to process the operating data and the weather forecast information to obtain a third candidate photovoltaic power generation short-term prediction result, a third candidate user load power short-term prediction result, and a third candidate net power consumption short-term prediction result, thereby obtaining a set of candidate photovoltaic power generation short-term prediction results, a set of candidate user load power short-term prediction results, and a set of candidate net power consumption short-term prediction results.

[0060] Each candidate photovoltaic power generation power short-term prediction result (i.e., the first candidate photovoltaic power generation power short-term prediction result or the second candidate photovoltaic power generation power short-term prediction result or the third candidate photovoltaic power generation power short-term prediction result) includes the future preset time window Every minute or every The candidate photovoltaic power generation power prediction result of each candidate user load power prediction result (i.e., the first candidate user load power short-term prediction result or the second candidate user load power short-term prediction result or the third candidate user load power short-term prediction result) includes the future preset time window Every minute or every The candidate user load power prediction result of each candidate net power consumption short-term prediction result (i.e., the first candidate net power consumption short-term prediction result or the second candidate net power consumption short-term prediction result or the third candidate net power consumption short-term prediction result) includes the future preset time window Every minute or every Candidate net electricity consumption forecast results for each minute.

[0061] The set of candidate photovoltaic power short-time prediction results includes the first candidate photovoltaic power short-time prediction result, the second candidate photovoltaic power short-time prediction result, and the third candidate photovoltaic power short-time prediction result; the set of candidate user load power short-time prediction results includes the first candidate user load power short-time prediction result, the second candidate user load power short-time prediction result, and the third candidate user load power short-time prediction result; and the candidate net power consumption short-time prediction result includes the first candidate net power consumption short-time prediction result, the second candidate net power consumption short-time prediction result, and the third candidate net power consumption short-time prediction result.

[0062] The intelligent control method of the household photovoltaic energy storage device can adaptively select and switch different types of prediction models according to the amount of data accumulated in the actual operation time, and continuously optimize the model parameters by using online learning and transfer learning techniques, so as to ensure that the prediction module can maintain the best prediction accuracy throughout the entire life cycle of the system. The specific model selection and switching strategy will be described in detail in the specific embodiments. It is worth emphasizing that the core innovation of the present application lies in the positive and negative area matching control strategy in the back end, and the specific prediction algorithm is not a limitation of the present application. Any prediction method that can provide future photovoltaic power and user load power prediction results can be applied to the present application.

[0063] In an embodiment of the present application, the final short-time prediction result is obtained by weighting the prediction results of a plurality of candidate prediction models. Accordingly, as shown in Figure 3 S1221A, evaluating the plurality of candidate prediction models to obtain a plurality of model evaluation results; S1222A, determining the weight values of each candidate photovoltaic power short-time prediction result in the set of candidate photovoltaic power short-time prediction results based on the plurality of model evaluation results to obtain a set of first weight values; and S1223A, calculating the weighted sum of the set of candidate photovoltaic power short-time prediction results based on the set of first weight values to obtain the photovoltaic power short-time prediction result.

[0064] In step S1221A, the plurality of candidate prediction models are evaluated to obtain a plurality of model evaluation results. Specifically, the adaptability of the plurality of candidate prediction models to their application scenarios can be evaluated according to the amount of data of the weather forecast information input into the candidate prediction models and the real-time operation data of the household photovoltaic energy storage device, the expected prediction accuracy, the expected prediction real-time performance, the expected prediction time length, and other parameters, and the adaptability is taken as the plurality of model evaluation results.

[0065] In another embodiment of the present application, the prediction result of a candidate prediction model with higher credibility is selected from the prediction results of a plurality of candidate prediction models as the final short-time prediction result. Accordingly, as shown inFigure 4 As shown, step S122 comprises steps of: S1221B, evaluating the plurality of candidate prediction models to obtain a plurality of model evaluation results; S1222B, selecting, from the set of candidate photovoltaic power short-time prediction results, a candidate photovoltaic power short-time prediction result corresponding to a candidate prediction model with the best model evaluation result as the photovoltaic power short-time prediction result.

[0066] The method for calculating the user load power short-time prediction result can refer to the method for calculating the photovoltaic power short-time prediction result.

[0067] Correspondingly, in an embodiment of the present application, after step S1221A, step S122 further comprises steps of: S1224A, determining, based on the plurality of model evaluation results, a weight value of each candidate user load power short-time prediction result in the set of candidate user load power short-time prediction results to obtain a second set of weight values; S1225A, calculating, based on the second set of weight values, a weighted sum of the set of candidate user load power short-time prediction results to obtain the user load power short-time prediction result. Step S1224A and step S1225A can be performed before step S1222A or after step S1223A, or can be performed in parallel with step S1222A and step S1223A respectively.

[0068] In another embodiment of the present application, after step S1221B, step S122 further comprises a step of: S1223B, selecting, from the set of candidate user load power short-time prediction results, a candidate user load power short-time prediction result corresponding to a candidate prediction model with the best model evaluation result as the photovoltaic power short-time prediction result. Step S1223B can be performed before step S1222B or after step S1222B, or in parallel with step S1222B.

[0069] The net power consumption short-time prediction result can be calculated based on the photovoltaic power short-time prediction result and the user load power short-time prediction result. Specifically, the net power consumption short-time prediction result is equal to the difference between the user load power short-time prediction result pload(t) and the photovoltaic power short-time prediction result Ppv(t).

[0070] In step S130, a net power consumption prediction curve is constructed based on the net power consumption short-time prediction result. Specifically, the net power consumption short-time prediction result comprises a net power consumption prediction result of each minute or each minute within the future preset time window Correspondingly, the net power consumption prediction curve can be obtained from the net power consumption short-time prediction result. The time series of multiple net power consumption forecast results within the future preset time window is constructed The net power consumption prediction curve intuitively reflects the energy supply and demand status of the system in the future and is the basis for the subsequent positive and negative area matching algorithm.

[0071] In step S140, when the state of charge of the energy storage battery reaches a preset upper limit threshold and there is a negative area that meets the dispatchable conditions in the net power consumption prediction curve, positive and negative area matching is performed on each dispatchable negative area in the net power consumption prediction curve to determine whether to trigger the intelligent control strategy.

[0072] In practice, when a power outage is detected, the intelligent control system of the household photovoltaic energy storage system immediately enters backup power mode, discharging the energy storage battery to power the user's load until the grid is restored or the battery's state of charge (SOC) reaches a preset lower threshold, SOC_min. This power outage backup mode has the highest priority, ensuring that users' basic electricity needs are met during a grid failure.

[0073] When the grid detects power or restores, the intelligent control system of the household PV energy storage device first charges the energy storage battery at maximum power until the state of charge (SOC) reaches the preset upper threshold SOC_max. Incoming calls prioritize charging the energy storage battery, preparing for subsequent "abandoned solar" utilization and backup power mode.

[0074] When the energy storage battery is full (SOC = SOC max) and there is a dispatchable period, positive-negative area matching is performed on each dispatchable negative area in the net power consumption prediction curve to determine whether to trigger the intelligent control strategy. If the negative area meets one of the following conditions, it means that the negative area is dispatchable and can be regarded as a "dispatchable negative area": (1) the negative area area A neg(i) is greater than the preset area threshold A min, the negative area duration (t i_end-t i_start) is greater than the preset time threshold T min, or the absolute value of the minimum net power of the negative area |P net_min| is greater than the preset power threshold P min. The specific values of the preset area threshold A min, the preset time threshold T min and the preset power threshold P min can be optimized and set according to the parameters such as the capacity of the energy storage battery, the charge-discharge rate, and the user's power consumption habit. The "dispatchable period" refers to the period that can be dispatched to charge the energy storage battery. The "dispatchable period" is the time period corresponding to the "dispatchable negative area". Through the definition of the dispatchable period, the intelligent control system of the household photovoltaic energy storage device does not run the positive-negative area matching algorithm at all times, but intelligently starts the control strategy when the energy storage battery is full and it is predicted that there will be "abandoned light" in the future, avoiding unnecessary energy scheduling. This triggering mechanism enables the present application to better balance photovoltaic consumption and uninterrupted power supply (UPS) functions, prioritizing the reliability of user power consumption during power grid failure, while maximizing photovoltaic self-use during normal operation of the power grid, and better meeting the actual needs of users.

[0075] If the energy storage battery is full and the predicted net power consumption curve does not have a negative area that meets the dispatchable condition, the intelligent control of the household photovoltaic energy storage device remains in the monitoring state and does not perform intelligent control strategy, waiting for the arrival of the next control period. When the positive-negative area matching is performed on each dispatchable negative area in the net power consumption prediction curve and a target positive area that meets the preset area matching condition is found, the intelligent control strategy is triggered.

[0076] Correspondingly, as Figure 5As shown, step S140 includes steps of: S141A, extracting the dispatchable negative value region from the net power consumption prediction curve; S142A, calculating the area of each dispatchable negative value region; S143A, searching forward from the starting time of each dispatchable negative value region to find a target positive value region that meets a preset area matching condition; S144A, when the target positive value region is found, triggering an intelligent control strategy, wherein the intelligent control strategy includes controlling the energy storage battery to start discharging at the starting time of the target positive value region and controlling the energy storage battery to start charging at the starting time of the dispatchable negative value region. Accordingly, the present application does not continuously run the positive and negative area matching algorithm, but intelligently starts the control strategy based on the preset triggering condition (for example, finding the target positive value region) to achieve on-demand control and energy-saving operation, while prioritizing the power reliability of users during power grid failure.

[0077] Step S141A extracts the dispatchable negative value region from the net power consumption prediction curve. Specifically, when the area of a single negative value region is too large and multiple positive value regions are needed to match it, the large negative value region can be split into multiple small negative value regions to perform positive and negative area matching to ensure that each split sub-negative value region can find a suitable match; when the area of a single negative value region is too small, multiple small negative value regions can be combined and matched with a positive value region.

[0078] Accordingly, in an embodiment of the present application, as shown in Figure 6 Step S141A includes steps of: S141A1A, extracting an original negative value region from the net power consumption prediction curve, the area of the original negative value region being greater than an ultra-large area threshold, the ultra-large area threshold being greater than the preset area threshold; S141A2A, splitting the original negative value region into multiple original negative value regions; S141A3A, extracting the dispatchable negative value region from the multiple original negative value regions.

[0079] In another embodiment of the present application, as shown in Figure 7 Step S141A includes steps of: S141A1B, extracting multiple original negative value regions from the net power consumption prediction curve, the area of the original negative value region being less than the preset area threshold; S141A2B, merging the multiple original negative value regions to obtain the dispatchable negative value region.

[0080] In step S142A, the area of each dispatchable negative value region is calculated. Specifically, for each dispatchable negative value region its area is calculated , which represents the electric energy available for charging the energy storage battery in the negative value region, wherein the area of the dispatchable negative value region is calculated according to the following formula: ; wherein, denotes the area of the schedulable negative value region, denotes the starting time of the original negative value region; denotes the ending time of the original negative value region; denotes the charging efficiency of the energy storage battery; denotes the function value of the net power consumption prediction curve; denotes the absolute value of . In actual calculation, the integral may be calculated approximately by using numerical integration or numerical summation.

[0081] The schedulable condition is that the area of the negative value region is greater than a preset area threshold, the duration of the negative value region is greater than a preset time threshold, or the absolute value of the minimum net power of the negative value region is greater than a preset power threshold.

[0082] In step S143A, a forward search is performed from the starting time of each of the schedulable negative value regions to find a target positive value region that meets a preset area matching condition. Specifically, in an embodiment of the present application, the preset area matching condition is that the area of the target positive value region is close to or equal to the ratio of the area of the schedulable negative value region to the discharging efficiency of the energy storage battery, i.e., ; wherein, denotes the area of the target positive value region; denotes the discharging efficiency of the energy storage battery, which is less than or equal to 1; wherein, the area of the target positive value region is calculated according to the following formula: ; wherein, denotes the area of the target positive value region, denotes the starting time of the target positive value region; denotes the ending time of the target positive value region. In actual calculation, the integral may be calculated approximately by using numerical integration or numerical summation.

[0083] In an embodiment of the present application, in step S143A, one or more candidate positive value regions that meet the preset area matching condition are found, so that the total area of the positive value regions meets the preset area matching condition; and a dynamic programming or branch and bound algorithm is used to screen the optimal one from the multiple candidate positive value regions as the target positive value region.

[0084] Correspondingly, in an embodiment of the present application, as shown in Figure 8 , step S143A includes steps of: S143A1, finding multiple candidate positive value regions that meet the preset area matching condition; and S143A2, using a dynamic programming or branch and bound algorithm to screen the optimal one from the multiple candidate positive value regions as the target positive value region.

[0085] If no target positive region corresponding to the schedulable negative region is found within the future preset time window, a cross-period matching is considered: try to match the current negative region with a positive region of a future longer time period (beyond the current future preset time window). This requires the use of prediction information of a longer time scale (for example, re-predicting the photovoltaic and load of a longer future time range). If still no match is found, the photovoltaic surplus of the negative region is not stored temporarily and is left for a subsequent rolling control period to attempt matching, and the algorithm continues to search for the next negative region.

[0086] Correspondingly, in an embodiment of the present application, after searching forward from the starting time of each schedulable negative region to find a target positive region meeting the preset area matching condition in step S140, step S145A is further included: when no target positive region corresponding to the schedulable negative region is found within the current future preset time window, performing positive-negative area matching between the schedulable negative region and a positive region of another future preset time window.

[0087] When the target positive region is searched in step S144A, an intelligent control strategy is triggered, wherein the intelligent control strategy includes: controlling the energy storage battery to start discharging at the starting time of the target positive region and controlling the energy storage battery to start charging at the starting time of the schedulable negative region. Specifically, the energy storage battery is controlled to start discharging at the starting time of the target positive region, and the discharging power can be dynamically adjusted according to the matched area and time, so as to try to completely release the power of the target positive region at the starting time of the schedulable negative region, and the energy storage battery is controlled to start charging at the starting time of the schedulable negative region, so as to store the photovoltaic surplus power into the energy storage battery.

[0088] It is worth mentioning that the schedulable negative region can be obtained through splitting and merging operations, and the target positive region can be selected from a plurality of candidate positive regions, that is, the positive region and the negative region are significantly asymmetric in the time sequence direction, and therefore, in order to reduce the asymmetric significant error rate of the positive-negative area matching, a positive-negative area dynamic balance analysis mechanism is introduced.

[0089] First, for the determined schedulable negative region, the sampling of the net power consumption in the net power consumption prediction curve is performed in the time sequence direction, for example, denoted as , and for the searched target positive region, the same number of samples is performed to obtain .

[0090] Then, for the negative sampling time sequence vector and the positive sampling time sequence vector , the time window coincidence phase correlation matrix is calculated as: ; so that the time window coincidence phase correlation matrix represents the dynamic matching relationship between the dispatchable negative value area and the target positive value area in time sequence, which can capture the significant asymmetry fluctuation characteristics of positive and negative asymmetry more than direct area matching.

[0091] Then, the instantaneous load standard deviation is taken as the fluctuation adjustment factor, and the integral is performed along the time sequence path, which can be approximated by summation here, so as to obtain the two-dimensional full-time fluctuation adjustment factor: ; that is, the full-time fluctuation adjustment sensitivity is controlled by the nonlinear fluctuation adjustment factor , and the weighted vector is constructed, so as to adapt to the prediction mutation characteristics of net power consumption.

[0092] Finally, the matching optimization of positive and negative area is: ;

[0093] , and . .

[0094] That is, in the positive-negative area dynamic balance analysis mechanism based on the net power consumption prediction value, for the significant asymmetry characteristics of positive surplus and negative shortage in time sequence window in time sequence, the local over-matching or local under-matching of fluctuation phase is compensated by accurate capture of positive-negative asymmetry fluctuation significant characteristics and full-time fluctuation adjustment correction through time sequence path integral, so as to establish the asymmetric significant time sequence coupling relationship of positive and negative area, and solve the area matching error rate misalignment problem of dispatchable negative value area and target positive value area.

[0095] Correspondingly, in an embodiment of the present application, as shown in Figure 9 , step S140 includes the following steps: S141B, extracting the dispatchable negative value area from the net power consumption prediction curve; S142B, extracting the target positive value area from the net power consumption prediction curve based on the positive-negative area dynamic balance analysis mechanism; S143B, triggering the intelligent control strategy when the target positive value area is searched, wherein the intelligent control strategy includes: controlling the energy storage battery to start discharging at the starting time of the target positive value area and controlling the energy storage battery to start charging at the starting time of the dispatchable negative value area.

[0096] As shown in Figure 10As shown, step S142B includes steps of: S1421B, sampling the positive and negative value regions from the net power consumption curve in the time sequence direction to obtain an initial negative value sampling time sequence vector and an initial positive value sampling time sequence vector; S1422B, calculating a time window coincidence phase correlation matrix based on the initial negative value sampling time sequence vector and the initial positive value sampling time sequence vector; S1423B, calculating a two-dimension full-time fluctuation adjustment factor and a weighting vector based on the initial negative value sampling time sequence vector and the initial positive value sampling time sequence vector; S1424B, calculating an area of an optimized schedulable negative value region and an area of a target positive value region based on the time window coincidence phase correlation matrix, the two-dimension full-time fluctuation adjustment factor, the weighting vector, the initial negative value sampling time sequence vector and the initial positive value sampling time sequence vector to extract a target positive value region from the net power consumption prediction curve based on the area of the target positive value region.

[0097] In step S1421B, the positive and negative value regions are sampled from the net power consumption curve in the time sequence direction to obtain an initial negative value sampling time sequence vector and an initial positive value sampling time sequence vector. Specifically, the schedulable negative value region is sampled from the net power consumption curve in the time sequence direction to obtain an initial negative value sampling time sequence vector; an initial positive value region is extracted from the net power consumption prediction curve; and the initial positive value region is sampled from the net power consumption curve in the time sequence direction to obtain an initial positive value sampling time sequence vector.

[0098] The initial negative value sampling time sequence vector is represented as: , , wherein n represents the number of samples. The initial positive value sampling time sequence vector is represented as: .

[0099] In step S1422B, the time window coincidence phase correlation matrix is calculated by the following formula: ; wherein represents the time window coincidence phase correlation matrix; represents the Kronecker product of the time window coincidence phase correlation matrix and the initial positive value sampling time sequence vector.

[0100] In step S1423B, the two-dimension full-time fluctuation adjustment factor and the weighting vector are calculated by the following formula: ; wherein represents a first-dimension full-time fluctuation adjustment factor in the two-dimension full-time fluctuation adjustment factor; represents a two-dimension full-time fluctuation adjustment factor in the two-dimension full-time fluctuation adjustment factor; represents a weighting vector; denotes the serial number of the sample, and is a positive integer.

[0101] In step S1424B, the area of the schedulable negative value region and the area of the positive value region are calculated by the following formula: ;

[0102] wherein, denotes the optimized negative value sampling timing sequence vector; denotes the optimized positive value sampling timing sequence vector; denotes the area of the schedulable negative value region after optimization; denotes the area of the target positive value region; denotes the reciprocal of the two-dimensional full-time fluctuation adjustment factor; denotes and the "XOR" operation result of denotes and the "XOR" operation result of and the Kronecker product of denotes and the "XOR" operation result of denotes and the "XOR" operation result of , and .

[0103] It is worth mentioning that in order to adapt to the dynamic changes of photovoltaic power generation and user load, the application adopts a rolling control strategy. Taking the time period as the control period , the positive and negative area matching of step S140 is executed in a loop, that is, every time period, according to the latest electricity consumption prediction curve, the control strategy of the future preset time window is regenerated, but only the control strategy of the first time period is executed each time. For example, when the control period is 1 hour, the future preset time window At every hour, the intelligent control system of the household photovoltaic energy storage device re-forecasts the photovoltaic power generation, user load power, and net power consumption for the next two hours. Based on the latest short-term forecast results of photovoltaic power generation, user load power, and net power consumption, it generates a control strategy for the next two hours. However, it only executes the control strategy for the next hour, leaving the remaining control strategy for the next control cycle. Through rolling control, the intelligent control system of the household photovoltaic energy storage device can promptly adjust the control strategy based on the latest forecast information, improving the real-time and flexibility of control, more effectively responding to fluctuations in photovoltaic power generation and user load power, and reducing the adverse effects of forecast errors.

[0104] The reason for using rolling control strategy is that only the first The time period control strategy is because for the prediction of time series data, the prediction accuracy of closer data points is usually higher, that is, predicting data in the near future is usually more accurate than predicting data in the far future. In this application, the predicted data points can be weighted according to their distance from the current moment. The closer the predicted value is to the current moment, the greater the weight of its error in the loss function of the candidate prediction model. Therefore, the candidate prediction model will pay more attention to the accuracy of recent predictions. This strategy ensures the real-time and accuracy of control, and can adjust the control behavior in time according to the latest prediction results, thereby more effectively responding to fluctuations in photovoltaic power generation and load.

[0105] In order to more clearly illustrate the execution process of the intelligent control strategy in a control cycle, this section uses the control cycle Hours, predicted duration Taking hours as an example, the execution steps and logic of the intelligent control strategy in one control cycle are described in detail.

[0106] Step 1: Obtain weather forecast information and collect real-time operating data of household photovoltaic energy storage equipment, wherein the operating data includes photovoltaic power generation, user load power and energy storage battery charge state.

[0107] Step 2: At each hour (e.g., 9:00, 10:00, 11:00, ...), the intelligent control system of the household photovoltaic energy storage device calculates the future preset time window based on the operating data and the weather forecast information. The photovoltaic power generation, user load power, and net power consumption within the hour are predicted to obtain short-term photovoltaic power generation, user load power, and net power consumption forecast results. Specifically, different types of candidate prediction models such as Prophet, XGBoost, and TCN can be adaptively selected based on the data volume of the operating data and weather forecast information, or other data, and online updates and candidate prediction model switching can be performed to ensure prediction accuracy.

[0108] Step 3, constructing a future preset time window based on the future preset time window Step 3, constructing a future preset time window based on the future preset time window Step 3, constructing a future preset time window based on the future preset time window The short-term prediction result of the net electricity consumption in an hour includes the prediction result of the net electricity consumption in each minute (or other time interval).

[0109] Step 4, when the state of charge of the energy storage battery reaches the preset upper threshold and there is a negative value region in the net electricity consumption prediction curve that meets the dispatchable condition, the positive-negative area matching algorithm is executed by the method of steps S141A-S144A or the method of steps S141B-S143B, and a detailed energy storage battery charging and discharging intelligent control strategy is generated for the next 2 hours. The core idea of the positive-negative area matching algorithm is to find a suitable positive value region to match each dispatchable negative value region within the future prediction time window, so that the total area of the positive value region is as close as possible to or equal to the ratio of the area of the negative value region to the discharging efficiency of the energy storage battery (considering the energy conversion efficiency). The control strategy is in minutes, indicating the charging and discharging state (charging, discharging or standby) and power of the energy storage battery at each time in the next 2 hours. Special case handling includes cases where the negative value region area is too large or too small, and the strategies of negative value region splitting and negative value region merging are used for optimization processing.

[0110] In the process of executing the energy storage battery charging and discharging intelligent control strategy, although a detailed energy storage battery charging and discharging intelligent control strategy is generated for the next 2 hours, only the intelligent control strategy for the next 1 hour is executed. For example, at 9:00, the intelligent control strategy for 9:00-11:00 is generated, but only the intelligent control strategy for 9:00-10:00 is actually executed. After the end of the 9:00-10:00 time period, even if the control strategy for 10:00-11:00 generated at 9:00 has not been executed, a new control strategy for the next 2 hours will be re-predicted and re-generated at 10:00, and the new strategy for 10:00-11:00 will be executed, and so on. This rolling control method can ensure the real-time and flexibility of the control strategy, and adjust in time according to the latest prediction information.

[0111] Steps 1-4 are repeated at each whole point to realize the rolling update and execution of the intelligent control strategy. If the future preset time window If the function value of the predicted net power consumption curve is always positive (P_net(t)≥0), i.e. the future will not occur "abandoned light" phenomenon, the monitoring will continue, and no control will be performed, and the arrival of the next control period will be waited. Only when the future "abandoned light" is predicted, the positive and negative area matching algorithm will be triggered, and the intelligent control strategy will be generated and executed.

[0112] Based on the mechanism of the intelligent control method of the household photovoltaic energy storage equipment, the application proposes an intelligent control system of the household photovoltaic energy storage equipment. Hereinafter, the intelligent control system of the household photovoltaic energy storage equipment of the embodiment of the application will be described with reference to Figure 11 the accompanying drawings.

[0113] Figure 11 Fig. 1 shows a block diagram of the intelligent control system of the household photovoltaic energy storage equipment according to the embodiment of the application.

[0114] As shown in Figure 11 , the intelligent control system of the household photovoltaic energy storage equipment includes a processor 100.

[0115] The processor 100 can be a central processing unit (CPU) or other forms of processing units with data processing capability and / or instruction execution capability, and can control other components in the intelligent control system of the household photovoltaic energy storage equipment to perform desired functions.

[0116] The processor 100 includes a data acquisition module 11, a prediction module 12, a net power consumption curve construction module 13, and a control strategy triggering module 14.

[0117] The data acquisition module 11 is configured to acquire weather forecast information and real-time acquisition of operation data of the household photovoltaic energy storage equipment, wherein the operation data includes photovoltaic power generation power, user load power, and energy storage battery state of charge; the prediction module 12 is configured to predict the photovoltaic power generation power, the user load power, and the net power consumption in a future preset time window based on the operation data and the weather forecast information to obtain a photovoltaic power generation power short-time prediction result, a user load power short-time prediction result, and a net power consumption short-time prediction result; the net power consumption curve construction module 13 is configured to construct a net power consumption prediction curve based on the net power consumption short-time prediction result; and the control strategy triggering module 14 is configured to, when the energy storage battery state of charge reaches a preset upper limit threshold and there is a negative value region in the net power consumption prediction curve that meets a dispatchable condition, perform positive and negative area matching on each dispatchable negative value region in the net power consumption prediction curve to determine whether to trigger an intelligent control strategy, wherein the dispatchable condition is that the negative value region area is greater than a preset area threshold, or the negative value region duration is greater than a preset time threshold, or the absolute value of the minimum net power of the negative value region is greater than a preset power threshold.

[0118] In an embodiment of the present application, the data collection module is further configured to collect power grid state.

[0119] In an embodiment of the present application, the prediction module 12 comprises a multi-model preliminary prediction module and a prediction result determination module. The multi-model preliminary prediction module is configured to process the operation data and the weather forecast information using a plurality of candidate prediction models to obtain a set of candidate short-time photovoltaic power generation prediction results, a set of candidate short-time user load power prediction results and a set of candidate short-time net power consumption prediction results; and the prediction result determination module is configured to calculate the short-time photovoltaic power generation prediction result, the short-time user load power prediction result and the short-time net power consumption prediction result based on the set of candidate short-time photovoltaic power generation prediction results, the set of candidate short-time user load power prediction results and the set of candidate short-time net power consumption prediction results.

[0120] In an example of the present application, the plurality of candidate prediction models comprises a statistical prediction model, a machine learning model and a deep learning model.

[0121] In an embodiment of the present application, the prediction result determination module comprises a model evaluation unit, a weight value determination unit and a result calculation unit. The model evaluation unit is configured to evaluate the plurality of candidate prediction models to obtain a plurality of model evaluation results; the weight value determination unit is configured to determine a weight value of each candidate short-time photovoltaic power generation prediction result in the set of candidate short-time photovoltaic power generation prediction results based on the plurality of model evaluation results to obtain a first set of weight values; and the result calculation unit is configured to calculate a weighted sum of the set of candidate short-time photovoltaic power generation prediction results based on the first set of weight values to obtain the short-time photovoltaic power generation prediction result.

[0122] In an embodiment of the present application, the weight value determination unit is further configured to determine a weight value of each candidate short-time user load power prediction result in the set of candidate short-time user load power prediction results based on the plurality of model evaluation results to obtain a second set of weight values; and the result calculation unit is further configured to calculate a weighted sum of the set of candidate short-time user load power prediction results based on the second set of weight values to obtain the short-time user load power prediction result.

[0123] In an embodiment of the present application, the prediction result determination module comprises a model evaluation unit and a result calculation unit. The model evaluation unit is configured to evaluate the plurality of candidate prediction models to obtain a plurality of model evaluation results; and the result calculation unit is configured to select a candidate short-time photovoltaic power generation prediction result of a candidate prediction model corresponding to a model evaluation result with the best evaluation result from the set of candidate short-time photovoltaic power generation prediction results as the short-time photovoltaic power generation prediction result.

[0124] In an embodiment of the present application, the result calculation unit is further configured to select, as the short-term photovoltaic power prediction result, a candidate short-term user load power prediction result corresponding to a candidate prediction model with the optimal model evaluation result from the set of candidate short-term user load power prediction results.

[0125] In an embodiment of the present application, the control strategy triggering module 14 comprises a dispatchable negative value region extraction unit, a dispatchable negative value region area calculation unit, a positive-negative area matching unit, and a strategy execution unit. The dispatchable negative value region extraction unit is configured to extract the dispatchable negative value region from the net power consumption prediction curve; the dispatchable negative value region area calculation unit is configured to calculate the area of each dispatchable negative value region; the positive-negative area matching unit is configured to search forward from the starting time of each dispatchable negative value region to find a target positive value region meeting a preset area matching condition; and the strategy execution unit is configured to trigger an intelligent control strategy when the target positive value region is found, wherein the intelligent control strategy comprises controlling the energy storage battery to start discharging at the starting time of the target positive value region and controlling the energy storage battery to start charging at the starting time of the dispatchable negative value region.

[0126] In an embodiment of the present application, the dispatchable negative value region extraction unit is further configured to extract an original negative value region from the net power consumption prediction curve, the area of the original negative value region being greater than an extra-large area threshold, the extra-large area threshold being greater than the preset area threshold; split the original negative value region into a plurality of original negative value regions; and extract the dispatchable negative value region from the plurality of original negative value regions.

[0127] In another embodiment of the present application, the dispatchable negative value region extraction unit is further configured to extract a plurality of original negative value regions from the net power consumption prediction curve, the area of each original negative value region being less than the preset area threshold; and perform a merging operation on the plurality of original negative value regions to obtain the dispatchable negative value region.

[0128] In an embodiment of the present application, the positive-negative area matching unit is configured to find a plurality of candidate positive value regions meeting the preset area matching condition; and select an optimal one from the plurality of candidate positive value regions as the target positive value region by using a dynamic programming or branch and bound algorithm.

[0129] In an embodiment of the present application, the control strategy triggering module 14 further comprises a cross-time period matching unit. The cross-time period matching unit is configured to perform positive-negative area matching between a dispatchable negative value region and a positive value region of another future preset time window when no target positive value region corresponding to the dispatchable negative value region is found within a current future preset time window.

[0130] In an embodiment of the present application, the control strategy triggering module 14 comprises a dispatchable negative area extraction unit, a positive-negative area matching unit and a strategy execution unit. The dispatchable negative area extraction unit is configured to extract the dispatchable negative area from the net power consumption prediction curve; the positive-negative area matching unit is configured to extract a target positive area from the net power consumption prediction curve based on a positive-negative area dynamic balance analysis mechanism; and the strategy execution unit is configured to trigger an intelligent control strategy when the target positive area is searched, wherein the intelligent control strategy comprises: controlling the energy storage battery to start discharging at the starting time of the target positive area and controlling the energy storage battery to start charging at the starting time of the dispatchable negative area.

[0131] The positive-negative area matching unit comprises a dispatchable negative area sampling unit, an initial positive area extraction unit, an initial positive area sampling unit, a correlation matrix calculation unit, an adjustment parameter unit and a target positive area extraction unit. The dispatchable negative area sampling unit is configured to sample the dispatchable negative area from the net power consumption curve in a time sequence direction to obtain an initial negative sampling time sequence vector, wherein the initial negative sampling time sequence vector is represented as: , The initial positive area extraction unit is configured to extract an initial positive area from the net power consumption prediction curve; the initial positive area sampling unit is configured to sample the initial positive area from the net power consumption curve in a time sequence direction to obtain an initial positive sampling time sequence vector, wherein the initial positive sampling time sequence vector is represented as: ; the correlation matrix calculation unit is configured to calculate a time window coincidence phase correlation matrix based on the initial negative sampling time sequence vector and the initial positive sampling time sequence vector; the adjustment parameter unit is configured to calculate a two-dimensional full-time fluctuation adjustment factor and a weighting vector based on the initial negative sampling time sequence vector and the initial positive sampling time sequence vector; and the target positive area extraction unit is configured to calculate the area of an optimized dispatchable negative area and the area of a target positive area based on the two-dimensional full-time fluctuation adjustment factor, the weighting vector, the time window coincidence phase correlation matrix, the initial negative sampling time sequence vector and the initial positive sampling time sequence vector to extract the target positive area from the net power consumption prediction curve based on the area of the target positive area.

[0132] In an embodiment of the present application, the processor 100 further comprises a rolling control module. The rolling control unit is configured to cycle the steps S110 to S140 of the intelligent control method of the household photovoltaic energy storage device with a preset control period as the control period.

[0133] In an embodiment of the present application, the intelligent control system of the household photovoltaic energy storage device comprises a photovoltaic array 10, an energy storage battery 20, a hybrid inverter 30, an edge computing device 40 and a cloud server 70. The energy storage battery 20 is connected between the photovoltaic array 10 and the hybrid inverter 30. The hybrid inverter 30 is connected between the energy storage battery 20 and a power grid 50. The power grid 50 is connected to a user load 60. The edge computing device 40 is connected between the photovoltaic array 10 and the energy storage battery 20. A processor 100 is arranged in the edge computing device 40. The cloud server 70 is connected to the edge computing device 40, the energy storage battery 20 and the hybrid inverter 30.

[0134] The photovoltaic array 10 is used to convert light energy into electrical energy by using the photovoltaic effect of solar panels, thereby providing power for the energy storage battery 20. The hybrid inverter 30 is used to convert direct current into alternating current, so that the energy source of the energy storage battery 20 can provide power for the user load 60 through the power grid 50. The edge computing device 40 switches and adjusts the working state of the energy storage battery 20 through the processor 100: the energy of the energy storage battery 20 is supplied to the user load 60, and the excess energy is sent to the energy storage battery 20 for storage. The cloud server 70 is used to monitor the photovoltaic array 10, the energy storage battery 20 and the hybrid inverter 30 through a remote network. In an embodiment of the present application, the edge computing device 40 is a low-cost embedded computing platform, which can be conveniently integrated into the intelligent control system of the existing household photovoltaic energy storage device without replacing the original hardware device. The edge computing device 40 serves as the intelligent control core, and performs data interaction and instruction transmission with the hybrid inverter 30 and the cloud server 70 through communication connection, thereby realizing local intelligent control and cloud collaborative optimization. This architecture embodies the characteristics of software and hardware decoupling, easy upgrading and modification of the present application.

[0135] In summary, the intelligent control method and system of the household photovoltaic energy storage device are illustrated. The intelligent control method of the household photovoltaic energy storage device does not use traditional rules or threshold control, but accurately predicts future photovoltaic power generation and user load, predicts the risk of "abandoned light" in advance, and uses an innovative positive and negative area matching algorithm to accurately match the photovoltaic power generation surplus period (negative area) with the user electricity demand period (positive area), realizing fine intelligent scheduling of the charging and discharging behavior of the energy storage battery, maximizing the storage of surplus photovoltaic power generation in the battery, and formulating the optimal charging and discharging control strategy to release at the user electricity peak period, thereby significantly reducing the "abandoned light" phenomenon, improving the self-use rate of photovoltaic power generation, electricity reliability, and ensuring power supply. The introduction of prediction makes the control strategy more forward-looking and predictive, better able to cope with the volatility of photovoltaic and load, and achieve more refined energy scheduling. The intelligent control method of the household photovoltaic energy storage device can seamlessly switch to a backup power supply mode when the power grid is powered off, providing reliable power supply for users and ensuring the continuous operation of critical loads.

[0136] It is worth emphasizing that the present application maximizes photovoltaic self-use under the premise of prioritizing user electricity reliability, which is more in line with the actual needs of users. Specifically, the present application innovatively proposes an intelligent triggering mechanism based on a dispatchable period, which only starts the positive and negative area matching algorithm when the battery is fully charged (SOC=SOC_max) and it is predicted that there may be "abandoned light" in the future. This triggering mechanism enables the present application to maximize photovoltaic self-use under the premise of prioritizing user electricity reliability, avoiding unnecessary energy scheduling.

[0137] To address the prediction challenges in different data volumes and application scenarios, and improve the environmental adaptability and robustness of the intelligent control system of the household photovoltaic energy storage device, the present application innovatively adopts an adaptive prediction model fusion strategy and a rolling control mechanism. The intelligent control system of the household photovoltaic energy storage device can adaptively select the best prediction model according to the data volume, and use rolling control to adjust the control strategy in a timely manner based on the latest prediction information, ensuring that the intelligent control system of the household photovoltaic energy storage device maintains optimal performance under various complex and uncertain working conditions.

[0138] The intelligent control method of the household photovoltaic energy storage device uses a fully automatic intelligent control strategy, which can intelligently determine whether to start the positive and negative area matching algorithm based on the grid state, the state of charge of the battery, and the net electricity consumption prediction curve, and automatically generate the optimal charging and discharging control strategy without human intervention, reducing the user usage threshold and improving the ease of use and intelligence level of the system.

[0139] The positive and negative area matching algorithm adopted in the application has relatively low calculation complexity, and the computing power requirement of the edge computing device is not high, so it is easy to deploy and apply in a household energy storage system with limited computing resources, for example, in the test process, the intelligent control method of the household photovoltaic energy storage device perfectly runs in the edge computing chip RK3566.

[0140] The intelligent control system of the household photovoltaic energy storage device adopts a flexible system architecture, which is easy to extend and upgrade. The processor can be conveniently integrated into an existing household energy storage system without replacing the original hardware device. Only a set of low-cost edge computing device and the corresponding software system need to be deployed to realize intelligent upgrading and reconstruction, which greatly reduces the upgrading cost and deployment difficulty, and has outstanding engineering application value.

[0141] The above describes the application and its embodiments, which are not restrictive, and the drawings shown are only one of the embodiments of the application, and the actual structure is not limited thereto. In summary, if a person skilled in the art is inspired by it, without departing from the purpose of the application, similar structural modes and embodiments can be designed without creativity, which should belong to the protection scope of the application.

Claims

1. An intelligent control method for household photovoltaic energy storage equipment, characterized in that: Including steps: S110: Acquire weather forecast information and collect real-time operating data of household photovoltaic energy storage equipment, wherein the operating data includes photovoltaic power generation, user load power, and state of charge of the energy storage battery; S120: Predicting the photovoltaic power generation power, user load power, and net power consumption within a future preset time window based on the operating data and the weather forecast information to obtain a short-term photovoltaic power generation power prediction result, a short-term user load power prediction result, and a short-term net power consumption prediction result; S130: Constructing a net power consumption prediction curve based on the short-term net power consumption prediction result; S140: When the state of charge of the energy storage battery reaches a preset upper threshold and there is a negative area in the net power consumption forecast curve that meets the dispatchable condition, performing positive and negative area matching on each dispatchable negative area in the net power consumption forecast curve to determine whether to trigger an intelligent control strategy, wherein the dispatchable condition is that the area of ​​the negative area is greater than a preset area threshold, or the duration of the negative area is greater than a preset time threshold, or the absolute value of the minimum net power in the negative area is greater than a preset power threshold, and the intelligent control strategy includes: controlling the energy storage battery to start discharging at the start time of the target positive area and controlling the energy storage battery to start charging at the start time of the dispatchable negative area; The method of extracting the target positive area from the net power consumption forecast curve based on the positive-negative area dynamic balance analysis mechanism includes the following steps: Sampling the positive and negative regions of the net power consumption curve along the time series direction to obtain an initial negative sampling time series vector and an initial positive sampling time series vector; The time window is calculated based on the initial negative sampling time series vector and the initial positive sampling time series vector, which conforms to the phase correlation matrix, expressed as: ;in, Represents the initial negative sampling time series vector, represents the initial positive sampling time series vector, Indicates that the time window conforms to the phase correlation matrix, Indicates that the time window conforms to the Kronecker product of the phase correlation matrix and the initial positive sampling time series vector; The two-dimensional full-time fluctuation adjustment factor and weighting vector are calculated based on the initial negative sampling time series vector and the initial positive sampling time series vector, which can be expressed as: ;in, It represents the first dimension full-time fluctuation adjustment factor in the two-dimensional full-time fluctuation adjustment factor; represents the two-dimensional full-time fluctuation adjustment factor in the two-dimensional full-time fluctuation adjustment factor; represents the weight vector; Indicates the sampling sequence number, which is a positive integer; Based on the two-dimensional full-time fluctuation adjustment factor, the weighted vector, the time window phase correlation matrix, the initial negative sampling time series vector and the initial positive sampling time series vector, the area of ​​the optimized dispatchable negative area and the area of ​​the target positive area are calculated to extract the target positive area from the net power consumption forecast curve based on the area of ​​the target positive area.

2. The intelligent control method for household photovoltaic energy storage equipment according to claim 1, characterized in that: Step S140 includes the following steps: Extracting the dispatchable negative value area from the net power consumption forecast curve; Calculating the area of ​​each of the schedulable negative value regions; Searching forward from the starting time of each of the schedulable negative value areas to find a target positive value area that meets a preset area matching condition; When the target positive area is found, the intelligent control strategy is triggered.

3. The intelligent control method for household photovoltaic energy storage equipment according to claim 2, characterized in that: Searching forward from the starting time of each of the schedulable negative value areas to find a target positive value area that meets the preset area matching condition, including the steps of: Find multiple candidate positive areas that meet the preset area matching conditions; Dynamic programming or branch-and-bound algorithm is used to select the best one from multiple candidate positive value regions as the target positive value region.

4. The intelligent control method for household photovoltaic energy storage equipment according to claim 2, characterized in that: Step S140, after searching forward from the starting time of each of the schedulable negative value areas to find a target positive value area that meets the preset area matching condition, includes the following steps: When no target positive region corresponding to the schedulable negative region is found in the current future preset time window, the schedulable negative region is matched with the positive region of another future preset time window for positive and negative areas.

5. The intelligent control method for household photovoltaic energy storage equipment according to claim 1, characterized in that: Extracting the dispatchable negative value area from the net power consumption forecast curve includes the following steps: Extracting an original negative area from the net power consumption prediction curve, wherein the area of ​​the original negative area is greater than an extra-large area threshold, and the extra-large area threshold is greater than the preset area threshold; Splitting the original negative value region into multiple original negative value regions; The schedulable negative value region is extracted from the plurality of original negative value regions.

6. The intelligent control method for household photovoltaic energy storage equipment according to claim 1, characterized in that: Extracting the dispatchable negative value area from the net power consumption forecast curve includes the following steps: Extracting a plurality of original negative value regions from the net power consumption prediction curve, wherein the area of ​​the original negative value region is smaller than the preset area threshold; A merging operation is performed on the multiple original negative value regions to obtain the schedulable negative value region.

7. The intelligent control method for household photovoltaic energy storage equipment according to claim 1, characterized in that: Step S140 includes the following steps: Extracting the dispatchable negative value area from the net power consumption forecast curve; Extracting a target positive area from the net power consumption prediction curve based on a positive-negative area dynamic balance analysis mechanism; When the target positive area is found, the intelligent control strategy is triggered.

8. The intelligent control method for household photovoltaic energy storage equipment according to claim 1, characterized in that: Sampling positive and negative regions from the net power consumption curve along the time sequence direction to obtain an initial negative sampling time sequence vector and an initial positive sampling time sequence vector includes the following steps: Sampling the schedulable negative value region from the net power consumption curve along the time sequence direction to obtain an initial negative value sampling time sequence vector; extracting an initial positive value region from the net power consumption prediction curve; The initial positive region is sampled from the net power consumption curve along a timing direction to obtain an initial positive sampling timing sequence vector, wherein the number of samples in the initial positive region is the same as the number of samples in the schedulable negative region.

9. An intelligent control system for household photovoltaic energy storage equipment, characterized in that: comprising a processor, the processor comprising: A data acquisition module, used to obtain weather forecast information and real-time collection of operating data of household photovoltaic energy storage equipment, wherein the operating data includes photovoltaic power generation, user load power and energy storage battery state of charge; A prediction module is used to predict the photovoltaic power generation power, user load power and net power consumption within a preset time window in the future based on the operating data and the weather forecast information to obtain a short-term prediction result of the photovoltaic power generation power, a short-term prediction result of the user load power and a short-term prediction result of the net power consumption; A net power consumption curve construction module, configured to construct a net power consumption prediction curve based on the short-term net power consumption prediction result; a control strategy triggering module, configured to, when the state of charge of the energy storage battery reaches a preset upper threshold and a negative region that satisfies a dispatchable condition exists in the net power consumption forecast curve, perform positive and negative area matching on each dispatchable negative region in the net power consumption forecast curve to determine whether to trigger an intelligent control strategy, wherein the dispatchable condition is that the area of ​​the negative region is greater than a preset area threshold, or the duration of the negative region is greater than a preset time threshold, or the absolute value of the minimum net power in the negative region is greater than a preset power threshold, and the intelligent control strategy includes: controlling the energy storage battery to start discharging at the start time of the target positive region and controlling the energy storage battery to start charging at the start time of the dispatchable negative region; The method of extracting the target positive area from the net power consumption forecast curve based on the positive-negative area dynamic balance analysis mechanism includes the following steps: Sampling the positive and negative regions of the net power consumption curve along the time series direction to obtain an initial negative sampling time series vector and an initial positive sampling time series vector; The time window is calculated based on the initial negative sampling time series vector and the initial positive sampling time series vector, which conforms to the phase correlation matrix, expressed as: ;in, Represents the initial negative sampling time series vector, represents the initial positive sampling time series vector, Indicates that the time window conforms to the phase correlation matrix, Indicates that the time window conforms to the Kronecker product of the phase correlation matrix and the initial positive sampling time series vector; The two-dimensional full-time fluctuation adjustment factor and weighting vector are calculated based on the initial negative sampling time series vector and the initial positive sampling time series vector, which can be expressed as: ;in, It represents the first dimension full-time fluctuation adjustment factor in the two-dimensional full-time fluctuation adjustment factor; represents the two-dimensional full-time fluctuation adjustment factor in the two-dimensional full-time fluctuation adjustment factor; represents the weight vector; Indicates the sampling sequence number, which is a positive integer; Based on the two-dimensional full-time fluctuation adjustment factor, the weighted vector, the time window phase correlation matrix, the initial negative sampling time series vector and the initial positive sampling time series vector, the area of ​​the optimized dispatchable negative area and the area of ​​the target positive area are calculated to extract the target positive area from the net power consumption forecast curve based on the area of ​​the target positive area.

Images