Control method of optical storage direct charging system module based on power prediction

Abstract

The invention relates to the technical field of new energy, hybrid energy storage systems and power distribution network optimization, and provides a control method of an optical storage direct charging system module based on power prediction, which is used for establishing a comprehensive target domain meeting the requirements of compensating prediction errors and stabilizing photovoltaic fluctuation on the basis of photovoltaic actual power and predicted power. Photovoltaic power is divided into an inner part and an outer part of a comprehensive target domain by adopting ICEEMDAN to be decomposed respectively, then high-frequency and low-frequency components are reconstructed through a TOPSIS method, the high-frequency power fluctuation component is smoothly output by a super capacitor, and the low-frequency power fluctuation component is smoothly output by a storage battery. According to different working conditions, grouped batteries are adopted to carry out primary control for compensating prediction errors and secondary control for stabilizing fluctuations, so that the energy storage device gives consideration to the photovoltaic power generation fluctuation and the prediction errors and has a rapid compensation effect, and irregularity, instability and uncertainty caused by large-scale photovoltaic power generation grid connection are further reduced; the electric energy quality and stability of a power grid are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of new energy, hybrid energy storage system and distribution network optimization, and in particular to a control method for a photovoltaic storage direct charging system module based on power prediction. Background Art

[0002] The present invention relates to the technical fields of new energy, hybrid energy storage system and distribution network optimization, and in particular to a control method for a photovoltaic storage direct charging system module based on power prediction. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present invention provides a control method for a photovoltaic direct charging system module based on power prediction. Through three levels of data analysis, target domain setting, charge / discharge reference power selection and energy storage battery group allocation, a complete data analysis model and power allocation mechanism are established, which solves the problem that the energy storage device in the prior art cannot effectively play its role in smoothing the volatility of photovoltaic power generation and compensating for prediction errors. Technical solution: The present invention provides a control method for a photovoltaic storage direct charging system module based on power prediction, which specifically includes the following steps: S1. Development of target domain: Based on the measured and predicted PV power, a probabilistic statistical analysis is performed on the PV output power prediction error and PV output power fluctuation, and a comprehensive target domain is established that satisfies both the compensation of prediction error and the suppression of PV fluctuation. S2 Selection of charge / discharge reference power: Prioritize compensation by using energy storage inside and outside the comprehensive target domain. Outside the comprehensive target domain, the upper and lower limits of the comprehensive target domain are used as the charge / discharge reference power; inside the comprehensive target domain, the predicted power is used as the charge / discharge reference power. The supercapacitor scope and battery scope are obtained by using the adaptive noise ensemble empirical mode decomposition (ICEEMDAN) and the topology of nearest ideal solutions (TOPSIS). The supercapacitor takes on the part with a larger rate of change, and the battery takes on the smooth part. S3 energy storage battery group allocation: the hybrid energy storage battery group acting on the target domain is divided into 4 groups, denoted as A1, B1, and A2, B2, and the charge / discharge state is switched between the groups. The 4 hybrid energy storage battery groups are allocated according to different working conditions. Preferably, in the formulation of the S1 target domain, a Gaussian mixture model distribution is used to analyze the photovoltaic power prediction error and power fluctuation. The probability density function expression of the Gaussian mixture model distribution is: Where: α k is the weight of the kth Gaussian distribution, Ν(x|μ k ,∑ k) is the mean value μ k , the covariance matrix is ​​∑ k Gaussian distribution, and Preferably, in the formulation of the S1 target domain, the photovoltaic predicted power error allowable bandwidth is set to: Where: D yc E is the photovoltaic power prediction error bandwidth; pk is the prediction error value. The allowed bandwidth of photovoltaic power fluctuation is: Where D bd is the photovoltaic power fluctuation bandwidth; P p Predict power fluctuations for photovoltaics. The allowed bandwidth of the comprehensive target domain is: Preferably, in the S2 charge / discharge reference power selection, outside the comprehensive target domain, the upper and lower limits of the comprehensive target domain are used as the charge / discharge reference power. When the actual power exceeds the upper limit of the target domain, the energy storage is charged; when the actual power exceeds the lower limit of the target domain, the energy storage is discharged. Preferably, in the selection of the charge / discharge reference power in S2, within the comprehensive target domain, the predicted power is used as the charge / discharge reference power. When the actual power is greater than the predicted power, the energy storage is charged; otherwise, the energy storage is discharged. Preferably, in the selection of S2 charge / discharge reference power, ICEEMDAN decomposition is adopted, which adds the IMF component of the original noise in each round of obtaining the low-frequency intrinsic mode function IMF component, so that all the IMF components finally generated can be added to reconstitute the original signal. Preferably, in the selection of S2 charge / discharge reference power, ICEEMDAN decomposition is adopted, which adds the IMF component of the original noise in each round of obtaining the low-frequency intrinsic mode function IMF component, so that all the IMF components finally generated can be added to reconstitute the original signal. Preferably, in S2, ICEEMDAN is used to decompose the photovoltaic power, and the decomposition steps are as follows: (1) Construct the actual photovoltaic power P (i) The decomposition formula of (t) is: P (i) (t) = P (t) + α i E i (ω (i) (t)) Where: P(t) is the actual photovoltaic power; α iis the signal-to-noise ratio of the residual component and the added noise; E i is the i-th IMF component generated by EMD decomposition; ω (i) (t) is the i-th white noise. (2) Calculate P (i) The local mean M(P (i) (t)), we get the first residual component r1(t)=M(P (i) (t)). (3) Calculate the modal component I b (t), when the number of components b = 1, the value of the first modal component I1(t) is: I1(t)=P(t)-r1(t) (4) Calculate the value of the second modal component I2(t): I2(t)=r1(t)-r2(t) Where: r2(t) is the second-order residual component r2(t) = M(r1(t) + α1E2[ω (i) (t)]). (5) Calculate the bth residual component r b (t) and the bth mode component I b for: r b (t) = M (r b-1 (t)+α b-1 E b [ω (i) (t)]) I b (t) = r b-1 (t)-r b (t) (6) Repeat step S2.5 until all modal components are obtained and the final ICEEMDAN decomposition result is obtained, that is: Where P imf-u (t) is the IMF component of each order after ICEEMDAN decomposition; t=1,2,3,…,u; P r (t) is the afterwave component. Preferably, in the S3 energy storage battery group allocation, 4 groups of hybrid energy storage battery groups are designed for the comprehensive target domain, each group contains 1 set of supercapacitors and batteries, and the power and capacity of each battery group are equal. The working conditions of the 4 groups of batteries are as follows: 1) Outside the comprehensive target domain, when the actual photovoltaic power is greater than the upper limit of the comprehensive target domain, the energy storage battery group A1 is charged. 2) Outside the comprehensive target domain, when the actual photovoltaic power is less than the lower limit of the comprehensive target domain, the energy storage battery group B1 is discharged. 3) Inside the comprehensive target domain, when the actual photovoltaic power is greater than the predicted power, the energy storage battery group A2 is charged. 4) Inside the comprehensive target domain, when the actual photovoltaic power is less than the predicted power, the energy storage battery group B2 is discharged. Preferably, in the S3 energy storage battery group allocation, the working conditions of the four hybrid energy storage battery groups are as follows: (1) Outside the comprehensive target domain, when the actual photovoltaic power is greater than the upper limit of the comprehensive target domain, the energy storage battery group A1 is charged; (2) Outside the comprehensive target domain, when the actual photovoltaic power is less than the lower limit of the comprehensive target domain, the energy storage battery group B1 discharges; (3) Within the comprehensive target domain, when the actual photovoltaic power is greater than the predicted power, the energy storage battery group A2 is charged. 4) Within the comprehensive target domain, when the actual photovoltaic power is less than the predicted power, the energy storage battery group B2 is discharged; (4) Within the comprehensive target domain, when the actual photovoltaic power is less than the predicted power, the energy storage battery group B2 discharges. Preferably, in the S3 energy storage battery group allocation, the hybrid energy storage battery group is charged and discharged alternately according to the state of charge, and the specific process is as follows: Outside the comprehensive target domain, when the energy storage battery group A1 is charged, the energy storage battery group B1 is to be discharged; when the energy storage battery group B1 is to be discharged, the energy storage battery group A1 is to be charged; when the state of charge of the energy storage battery group A1 reaches the upper limit or the state of charge of B1 reaches the lower limit, A1 and B1 are switched; Within the comprehensive target domain, when the energy storage battery group A2 is charging, the energy storage battery group B2 is to be discharged; when the energy storage battery group B2 is discharging, the energy storage battery group A2 is to be charged; when the charge state of the energy storage battery group A2 reaches the upper limit or the charge state of the energy storage battery group B2 reaches the lower limit, A2 and B2 switch. Beneficial effects: 1. The present invention establishes a model based on the distribution of prediction error and power fluctuation-related characteristics, and establishes a comprehensive target domain that satisfies both compensation for prediction error and smoothing of photovoltaic fluctuation based on actual photovoltaic power and predicted power. ICEEMDAN is used to divide photovoltaic power into two parts inside and outside the comprehensive target domain, and then the parts are decomposed separately. Then, the TOPSIS method is used to reconstruct the high-frequency and low-frequency components, in which the high-frequency power fluctuation component is smoothed by the supercapacitor, and the low-frequency power fluctuation component is smoothed by the battery. According to different working conditions, grouped batteries are used for primary control to compensate for prediction errors and secondary control to smooth fluctuations, so that the charge / discharge state of group A and group B batteries is switched, thereby realizing the rapid compensation effect of energy storage devices on photovoltaic power generation volatility and prediction errors, further reducing the irregularity, instability and uncertainty brought about by large-scale photovoltaic power generation grid connection, so as to improve the power quality and stability of the power grid. 2. The hybrid energy storage battery pack of the present invention switches the charge / discharge state of the grouped batteries according to the charge / discharge reference power, thereby avoiding the problem of overcharge or power loss of the energy storage battery pack. 3. The hybrid energy storage battery pack of the present invention adopts a combination of batteries and supercapacitors to distribute power into high-frequency components and low-frequency components. The high-frequency components are charged / discharged by the supercapacitors, and the low-frequency components are charged / discharged by the batteries, thereby improving the response speed and service life of the hybrid energy storage battery pack. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 It is an overall block diagram of an embodiment of the present invention. Figure 2 Flow chart of the photovoltaic hybrid energy storage priority charge / discharge control strategy in an embodiment of the present invention. DETAILED DESCRIPTION Next, this article will refer to the drawings in the embodiments of the present invention to elaborate on the technical solution of the present invention in detail and clearly. It should be pointed out that the embodiments mentioned only represent some applications of the present invention, not all implementation cases. Without involving creative work, any other implementation methods that can be derived by any technician in this field based on the embodiments of the present invention should be regarded as the protection scope of the present invention. Please refer to Figure 1-2 The present invention provides a control method for a photovoltaic storage direct charging system module based on power prediction, and the specific process is as follows: S1 Setting of target domain: Based on the measured and predicted PV power, a probabilistic statistical analysis is performed on the PV output power prediction error and PV output power fluctuation, and a comprehensive target domain is established that satisfies both the compensation of prediction error and the suppression of PV fluctuation. Specifically, through the observation and analysis of the typical day prediction error and power fluctuation data of a photovoltaic power station, the prediction error and power fluctuation are mainly concentrated in a single model distribution. Therefore, the Gaussian mixture model (Gaussian MixtureModel) distribution is used to analyze the photovoltaic power prediction error and power fluctuation. The probability density function expression of the Gaussian mixture model (Gaussian MixtureModel) distribution is: Where: α k is the weight of the kth Gaussian distribution, Ν(x|μ k ,∑ k ) is the mean value μ k , the covariance matrix is ​​∑ k Gaussian distribution, and In the formulation of the S1 target domain, the allowed bandwidth of the photovoltaic prediction power error is set as: Where: D yc E is the photovoltaic power prediction error bandwidth; pk is the prediction error value. The allowed bandwidth of photovoltaic power fluctuation is: Where D bd is the photovoltaic power fluctuation bandwidth; P p Predict power fluctuations for photovoltaics. The allowed bandwidth for the comprehensive target domain is: S2 charging / discharging reference power selection: Different from the traditional compensation method, establishing a comprehensive target domain can divide the actual power into two parts, which are prioritized by using energy storage inside and outside the comprehensive target domain. Specifically, outside the comprehensive target domain, the upper and lower limits of the comprehensive target domain are used as the reference power for charging / discharging. When the actual power exceeds the upper limit of the target domain, the energy storage is charged; when the actual power exceeds the lower limit of the target domain, the energy storage is discharged. Charging / discharging area D outside the target domain out (t) is: Where: P is the actual photovoltaic power; t is the time series; P up is the upper limit of the comprehensive target domain; P down is the lower limit of the comprehensive target domain. In the target domain, the predicted power is used as the reference power for charging / discharging. When the actual power is greater than the predicted power, the energy storage is charged; otherwise, the energy storage is discharged. The charging / discharging area Din(t) in the target domain is: Where: Predict power for photovoltaics. ICEEMDAN is used to decompose photovoltaic power. The decomposition steps are as follows: (1) Constructing the actual photovoltaic power P (i) The decomposition formula of (t) is: P (i) (t) = P (t) + α i E i (ω (i) (t)) Where: P(t) is the actual photovoltaic power; α i is the signal-to-noise ratio of the residual component and the added noise; E i is the i-th IMF component generated by EMD decomposition; ω (i) (t) is the i-th white noise. (2) Calculate P (i) The local mean M(P (i) (t)), we get the first residual component r1(t)=M(P (i) (t)). (3) Calculation of modal component I b (t), when the number of components b = 1, the value of the first modal component I1(t) is: I1(t)=P(t)-r1(t) (4) The value of the second modal component I2(t) is calculated as: I2(t)=r1(t)-r2(t) Where: r2(t) is the second-order residual component r2(t) = M(r1(t) + α1E2[ω (i) (t)]). (5) Similarly, calculate the bth residual component r b (t) and the bth mode component I b for: r b (t) = M (r b-1 (t)+α b-1 E b [ω (i) (t)]) I b (t) = r b-1 (t)-r b (t) (6) Repeat step (5) until all modal components are obtained and the final ICEEMDAN decomposition result is obtained, that is: Where Pimf-u (t) is the IMF component of each order after ICEEMDAN decomposition; t=1,2,3,…,u; P r (t) is the afterwave component. The TOPSIS method is used to accurately extract the component signals after ICEEMDAN decomposition. The reconstruction steps are as follows: (1) Determine the target sequence and comparison sequence of each order IMF component, and form a data matrix x. According to its elements x ij The matrix is ​​normalized as follows: Where: x' kj is the per-unit value of each sequence; a is the IMF component of each order; k and j are the number of rows and columns where the data is located. (2) The undecomposed data is The minimum value of the IMF component at each moment is Where: n is the total number of data rows. (3) The kth data in each column and The distance is The distance between the kth data in each column and the minimum value of the IMF component at each moment is (4) Obtain the degree of relationship between each comparison sequence and the target sequence, that is: Where: g k For the degree of relationship. (5) According to the degree of relationship between each sequence and the target sequence, the IMF components are reconstructed, the components with high degree of relationship are classified as high frequency, and the remaining components are classified as low frequency, and the final high-frequency and low-frequency smoothed output power instructions are obtained. The high-frequency smoothed output power instruction is: The low-frequency smooth output power command is: Using the sliding average filtering method, the sliding average filtering target reference power at time t is obtained: Where P ref (t) is the sliding average filtering target reference power at time t; N is the data window length; w = 0, 1, 2, ..., N-1. The charge / discharge power instructions for the supercapacitor are: Psc (t) = P high (t)-P ref (t) Where P sc (t) is the supercapacitor charge / discharge power instruction at time t. sc (t)>0, supercapacitor is charged; P sc (t)<0, the supercapacitor is discharged. The battery charge / discharge power command is: P b (t) = P low (t)-P ref (t) Where P b (t) is the supercapacitor charge / discharge power instruction at time t. b (t)>0, the battery is charged; P b (t)<0, the battery is discharged. S3 Energy storage battery group allocation: Set up 4 groups of hybrid energy storage battery groups, each group contains 1 set of supercapacitors and batteries, and the power and capacity of each battery group are equal. The 4 groups of hybrid energy storage battery groups are recorded as A1, B1 and A2, B2 respectively, and work in the target domain of compensation prediction error and the target domain of smoothing fluctuations. The charge / discharge state is switched between groups A and B, and the 4 groups of hybrid energy storage battery groups are allocated according to different working conditions. Specifically, the working conditions of the 4 groups of batteries are as follows: 1) Outside the comprehensive target domain, when the actual photovoltaic power is greater than the upper limit of the comprehensive target domain, the energy storage battery group A1 is charged. 2) Outside the comprehensive target domain, when the actual photovoltaic power is less than the lower limit of the comprehensive target domain, the energy storage battery group B1 discharges. 3) Within the comprehensive target domain, when the actual photovoltaic power is greater than the predicted power, the energy storage battery group A2 is charged. 4) Within the comprehensive target domain, when the actual photovoltaic power is less than the predicted power, the energy storage battery group B2 discharges. The switching of the charging / discharging working state of the hybrid energy storage battery group is specifically as follows: since the charge state of the hybrid energy storage battery has an upper and lower limit, charging and discharging need to be alternated during the charging / discharging process, so at the same time, only one of the charging group and the discharging group is in the working state, that is, when one group of batteries is in the charging state, the other group is in the state to be discharged; conversely, if the battery group is in the discharging state, the other group is in the state to be charged. If one group of batteries is fully charged or fully discharged, the two groups of batteries will switch working states. In addition, four operating conditions are set to determine whether the power fluctuations in the target domain of compensation prediction error and smoothing fluctuations meet the grid connection standards. Operating condition 1: The prediction error is not satisfied, the power fluctuation is not satisfied, and the primary control of compensation prediction error is performed first, and the A1 and B1 groups work. If the upper limit of the prediction allowable error is exceeded, the A1 battery group works; if the lower limit of the prediction allowable error is exceeded, the B1 battery group works; the two groups of batteries charge / discharge alternately. After completion, the secondary control of smoothing fluctuations is performed, and the A2 and B2 groups work; if the power is greater than the first-order low-pass filter smooth curve, the smoothing charging group works; if the power is less than the first-order low-pass filter smooth curve, the smoothing discharge group works; the two groups of batteries charge / discharge alternately. Operating condition 2: The prediction error is not satisfied, the power fluctuation is satisfied, and the primary control of compensation prediction error is performed directly, and the secondary control of smoothing fluctuations does not work. The A1 and B1 groups work, the two groups of batteries charge / discharge alternately, and the A2 and B2 groups standby. Condition 3: The prediction error is satisfied, the power fluctuation is also satisfied, and the grid is directly connected. The primary control for compensating the prediction error and the secondary control for smoothing the fluctuation are not activated, and groups A1, B1 and A2, B2 are on standby. Condition 4: The prediction error is satisfied, the power fluctuation is not satisfied, and the primary control for compensating the prediction error is directly skipped, and the secondary control for smoothing the fluctuation is performed. Groups A1 and B1 are on standby, and groups A2 and B2 are working. Furthermore, the four groups of hybrid energy storage battery packs are all composed of batteries and supercapacitors. The volatility is used as an evaluation index of 5 minutes, and a hybrid energy storage battery pack represented by supercapacitors and batteries is used for charging / discharging control strategy. The smoothness is used as an evaluation index of 1 minute, and a single energy storage represented by supercapacitors is used for charging / discharging control strategy to realize energy storage system priority charging / discharging control strategies of different time scales. The specific control strategies are as follows. (1) Taking the photovoltaic grid-connected standards as the target, determine whether the photovoltaic active power meets the grid-connected standards, that is, Where P ch (t) is the change value of photovoltaic active power in 1 minute; P max (t) is the maximum active power value of 1 min; P min (t) is the minimum active power value for 1 min; C pv is the total installed capacity of photovoltaic power. (2) The photovoltaic active power meets the photovoltaic grid-connected standard, the energy storage battery group is in the shutdown state, and the supercapacitor battery group enters the working state. A smoothing process is performed on a time scale of 1 minute. The expression is: Where: P o (t) is the photovoltaic active power smoothness curve; L is the total length of photovoltaic output grid-connected power time. sc(t) is the supercapacitor charge / discharge power instruction at time t. sc (t)>P o (t), supercapacitor charging; P sc (t)<P o (t), supercapacitor discharge. (3) When the photovoltaic active power exceeds the upper limit specified by the grid-connected standard and enters the smoothing fluctuation area, the battery and supercapacitor enter the working state, and the output fluctuation is smoothed on a 5-minute time scale. The smoothness follows the smooth output fluctuation. After the output fluctuation is smoothed, the smoothness is smoothed on a 1-minute time scale. The main energy storage expression is: P hess (t) = P pv (t)-P ref (t) Where: P hess (t) is the charging and discharging power of the hybrid energy storage system, P ref (t) is the sliding average filtering target reference power at time t. hess (t)>P ref (t), hybrid energy storage charging; P hess (t)<P ref (t) Hybrid energy storage discharge. In order to prevent the energy storage system from being overcharged or over-discharged during operation, it is necessary to constrain the SOC of the supercapacitor and the battery. The constraints are: Where: SOC sc (t), SOC b (t) are the state of charge of the supercapacitor and the battery at time t; SOC sc-min (t), SOC b-min (t) are the minimum allowable range of the state of charge of the supercapacitor and the battery at time t; SOC sc-max (t), SOC b-max (t) are the maximum allowable range of the charge state of the supercapacitor and the battery at time t. Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A control method for a photovoltaic direct charging system module based on power prediction, characterized in that: The method comprises the following steps: S1: Target domain formulation: Based on the actual PV power and predicted power, a comprehensive target domain is established to compensate for the prediction error and smooth the PV fluctuation; S2: Selection of charging / discharging reference power: The actual photovoltaic power is divided into two parts, one within the comprehensive target domain and one within the comprehensive target domain. ICEEMDAN is used to decompose the photovoltaic power into two parts, one inside the comprehensive target domain and one outside the comprehensive target domain, and then the TOPSIS method is used to reconstruct the high-frequency power fluctuation component and the low-frequency power fluctuation component. S3: Energy storage battery group allocation: set 4 groups of hybrid energy storage battery groups, namely, charging group A1 outside the target domain, discharging group B1 outside the target domain, charging group A2 inside the target domain, and discharging group B2 inside the target domain. Each group of hybrid energy storage battery groups is composed of batteries and supercapacitors. The charging group A1 outside the target area and the discharging group B1 outside the target area are used to compensate for the power outside the target area; The charging group A2 in the target area and the discharging group B2 in the target area are used to compensate the power in the target area; The high-frequency power fluctuation component is smoothed out by the supercapacitor, and the low-frequency power fluctuation component is smoothed out by the battery.

2. According to the control method of the photovoltaic direct charging system module based on power prediction in claim 1, it is characterized in that: In the formulation of the S1 target domain, the Gaussian mixture model distribution is used to analyze the photovoltaic power prediction error and power fluctuation. The probability density function expression of the Gaussian mixture model distribution is: Where: α k is the weight of the kth Gaussian distribution, Ν(x|μ k ,∑ k ) is the mean value μ k , the covariance matrix is ​​∑ k Gaussian distribution, and 3. According to the control method of the photovoltaic direct charging system module based on power prediction in claim 1, it is characterized in that: In the formulation of the S1 target domain, the formulation process of the comprehensive target domain is as follows: (1) The allowable bandwidth of photovoltaic power prediction error is set to: Where: D yc E is the photovoltaic power prediction error bandwidth; pk is the prediction error value; (2) The allowable bandwidth of photovoltaic power fluctuation is: Where D bd is the photovoltaic power fluctuation bandwidth; P p Predicting power fluctuations for photovoltaics; (3) The allowed bandwidth of the comprehensive target domain is:

4. According to the control method of the photovoltaic direct charging system module based on power prediction in claim 1, it is characterized in that: In the S2 charge / discharge reference power selection, outside the comprehensive target domain, the upper and lower limits of the comprehensive target domain are used as the charge / discharge reference power. When the actual power exceeds the upper limit of the target domain, the energy storage is charged; when the actual power exceeds the lower limit of the target domain, the energy storage is discharged.

5. According to the control method of the photovoltaic direct charging system module based on power prediction in claim 1, it is characterized in that: In the S2 charge / discharge reference power selection, within the comprehensive target domain, the predicted power is used as the charge / discharge reference power. When the actual power is greater than the predicted power, the energy storage is charged; otherwise, the energy storage is discharged.

6. The control method of a photovoltaic direct charging system module based on power prediction according to claim 1 is characterized in that: In the selection of S2 charge / discharge reference power, ICEEMDAN decomposition is used. In each round of calculating the low-frequency intrinsic mode function IMF component, the IMF component of the original noise is added, so that all the IMF components finally generated can be added together to reconstruct the original signal.

7. The control method of a photovoltaic direct charging system module based on power prediction according to claim 5 is characterized in that: In S2, ICEEMDAN is used to decompose the photovoltaic power. The decomposition steps are as follows: (1) Construct the actual photovoltaic power P (i) The decomposition formula of (t) is: P (i) (t)=P(t)+α i E i (oh (i) (t)) Where: P(t) is the actual photovoltaic power; α i is the signal-to-noise ratio of the residual component and the added noise; E i is the i-th IMF component generated by EMD decomposition; ω (i) (t) is the i-th white noise; (2) Calculate P (i) The local mean M(P (i) (t)), we get the first residual component r1(t)=M(P (i) (t)); (3) Calculate the modal component I b (t), when the number of components b = 1, the value of the first modal component I1(t) is: I1(t) = P(t) - r1(t), (4) Calculate the value of the second modal component I2(t): I2(t)=r1(t)-r2(t), Where: r2(t) is the second-order residual component r2(t) = M(r1(t) + α1E2[ω (i) (t)]); (5) Calculate the bth residual component r b (t) and the bth mode component I b for: r b (t)=M(r b-1 (t)+α b-1 E b [ω (i) (t)]), I b (t)=r b-1 (t)-r b (t), (6) Repeat step S2.5 until all modal components are obtained and the final ICEEMDAN decomposition result is obtained, that is: Where P imf-u (t) is the IMF component of each order after ICEEMDAN decomposition; t=1,2,3,…,u; P r (t) is the afterwave component.

8. The control method of a photovoltaic direct charging system module based on power prediction according to claim 5 is characterized in that: In S2, the TOPSIS method is used to accurately extract the component signals after ICEEMDAN decomposition. The reconstruction steps are as follows: (1) Determine the target sequence and comparison sequence of each order IMF component, and form a data matrix x. According to its elements x ij The matrix is ​​normalized as follows: Where: x' kj is the per-unit value of each sequence; a is the IMF component of each order; k and j are the number of rows and columns where the data is located; (2) The undecomposed data is The minimum value of the IMF component at each moment is Where: n is the total number of data rows; (3) The kth data in each column and The distance is The distance between the kth data in each column and the minimum value of the IMF component at each moment is Where: g k For the degree of relationship, (5) According to the degree of relationship between each sequence and the target sequence, the IMF components are reconstructed, the components with high degree of relationship are classified as high frequency, and the remaining components are classified as low frequency, and the final high-frequency and low-frequency smoothed output power instructions are obtained. The high-frequency smoothed output power instruction is: The low-frequency smooth output power command is: Using the sliding average filtering method, the sliding average filtering target reference power at time t is obtained: Where P ref (t) is the sliding average filtering target reference power at time t; N is the data window length; w= 0,1,2,…,N-1, the charge / discharge power command of the supercapacitor is: P sc (t)=P high (t)-P ref (t) Where P sc (t) is the supercapacitor charging / discharging power instruction at time t, P sc (t)>0, supercapacitor is charged; P sc (t)<0, the supercapacitor discharges, and the battery charge / discharge power command is: P b (t)=P low (t)-P ref (t) Where P b (t) is the supercapacitor charging / discharging power instruction at time t, P b (t)>0, the battery is charged; P b (t)<0, the battery is discharged.

9. The control method of a photovoltaic direct charging system module based on power prediction according to claim 1 is characterized in that: In the S3 energy storage battery group allocation, the working conditions of the four hybrid energy storage battery groups are as follows: (1) Outside the comprehensive target domain, when the actual photovoltaic power is greater than the upper limit of the comprehensive target domain, the energy storage battery group A1 is charged; (2) Outside the comprehensive target domain, when the actual photovoltaic power is less than the lower limit of the comprehensive target domain, the energy storage battery group B1 discharges; (3) Within the comprehensive target domain, when the actual photovoltaic power is greater than the predicted power, the energy storage battery group A2 is charged. (4) Within the comprehensive target domain, when the actual photovoltaic power is less than the predicted power, the energy storage battery group B2 discharges.

10. The control method of a photovoltaic direct charging system module based on power prediction according to claim 1, characterized in that: In the S3 energy storage battery group allocation, the hybrid energy storage battery group is charged and discharged alternately according to the state of charge. The specific process is as follows: Outside the comprehensive target domain, when the energy storage battery group A1 is charged, the energy storage battery group B1 is to be discharged; when the energy storage battery group B1 is to be discharged, the energy storage battery group A1 is to be charged; when the state of charge of the energy storage battery group A1 reaches the upper limit or the state of charge of B1 reaches the lower limit, A1 and B1 are switched; Within the comprehensive target domain, when the energy storage battery group A2 is charging, the energy storage battery group B2 is to be discharged; when the energy storage battery group B2 is discharging, the energy storage battery group A2 is to be charged; when the charge state of the energy storage battery group A2 reaches the upper limit or the charge state of the energy storage battery group B2 reaches the lower limit, A2 and B2 switch.

Images