Micro inverter and method for energy storage battery system

Abstract

The present application relates to the field of intelligent control, and specifically discloses a micro-inverter and method for an energy storage battery system, which collects the energy storage voltage of the energy storage battery in real time through a first voltage sensor, collects the grid voltage in real time through a second voltage sensor, and collects the grid-connected current in real time through a current sensor, and uses a data processing and analysis algorithm based on deep learning to analyze and process the timing correlation characteristics of the energy storage voltage, the grid voltage, and the grid-connected current, so as to adaptively generate a control result of the recommended duty ratio of the first duty ratio signal of the first power tube in the flyback converter according to the coordinated changes and correlations of the real-time operating parameters of the micro-inverter. In this way, the real-time operating parameters of the micro-inverter can be understood, so as to intelligently recommend the duty ratio of the first power tube in the flyback converter, thereby reducing energy loss, improving the energy utilization efficiency of the entire energy storage battery system, and improving the overall performance and efficiency of the system.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to a micro inverter and method for an energy storage battery system. Background Art

[0002] A micro inverter is an electronic device used in energy storage battery systems. It consists of an MCU controller, energy storage capacitors connected between photovoltaic panels and the power grid, a flyback converter, an industrial frequency inverter bridge, and an EMI filter. It can convert direct current (DC) into alternating current (AC), so that the energy in the battery can be used to power the power grid or other electrical equipment. Micro inverters can improve the efficiency and reliability of energy storage battery systems and are widely used in energy storage battery systems, including photovoltaic power generation systems, wind power generation systems, electric vehicle charging systems, etc. Among them, the flyback converter is a commonly used switching power supply topology used to convert direct current into alternating current or change the frequency of alternating current to ensure stable and reliable operation of the circuit.

[0003] However, traditional micro inverters have high power loss, which leads to lower overall system efficiency and poor stability. In addition, the dynamic response of traditional micro inverters is limited by control algorithms and hardware design, and cannot respond in real time and meet certain requirements of high dynamic performance or different working conditions.

[0004] Therefore, an optimized microinverter for an energy storage battery system is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a micro-inverter and method for an energy storage battery system, which collects the energy storage voltage of the energy storage battery in real time through a first voltage sensor, collects the grid voltage in real time through a second voltage sensor, and collects the grid-connected current in real time through a current sensor, and uses a data processing and analysis algorithm based on deep learning to analyze and process the timing correlation characteristics of the energy storage voltage, the grid voltage, and the grid-connected current, so as to adaptively generate a control result of the recommended duty ratio of the first duty ratio signal of the first power tube in the flyback converter according to the coordinated changes and correlations of the real-time operating parameters of the micro-inverter. In this way, the real-time operating parameters of the micro-inverter can be understood, so as to intelligently recommend the duty ratio of the first power tube in the flyback converter, thereby reducing energy loss, improving the energy utilization efficiency of the entire energy storage battery system, and improving the overall performance and efficiency of the system.

[0006] According to one aspect of the present application, a micro inverter for an energy storage battery system is provided, comprising:

[0007] An energy storage voltage acquisition module, used to acquire a time series of the energy storage voltage of the energy storage battery acquired by the first voltage sensor;

[0008] A power grid voltage acquisition module, used to acquire a time series of the power grid voltage acquired by the second voltage sensor;

[0009] A grid-connected current acquisition module, used to acquire the time series of the grid-connected current acquired by the current sensor;

[0010] A time dimension arrangement module, used to arrange the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid-connected current into an energy storage voltage timing input vector, a grid voltage timing input vector, and a grid-connected current timing input vector according to the time dimension;

[0011] A timing correlation feature module is used to extract timing correlation features from the energy storage voltage timing input vector, the grid voltage timing input vector and the grid-connected current timing input vector to obtain an energy storage voltage timing correlation feature vector, a grid voltage timing correlation feature vector and a grid-connected current timing correlation feature vector;

[0012] A feature expression enhancement module is used to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid-connected current timing association feature vector;

[0013] A feature fusion module, used for processing the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector using a quasi-Bayesian probability network to obtain an empty ratio a posteriori control semantic feature vector as an empty ratio a posteriori control semantic feature;

[0014] The control result module is used to determine the control result based on the posterior control semantic feature of the empty ratio.

[0015] According to another aspect of the present application, a micro-inverter method for an energy storage battery system is provided, comprising:

[0016] Acquire a time series of the energy storage voltage of the energy storage battery collected by the first voltage sensor;

[0017] Acquire a time series of a grid voltage collected by a second voltage sensor;

[0018] Obtaining a time series of grid-connected current collected by a current sensor;

[0019] Arrange the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid-connected current into an energy storage voltage timing input vector, a grid voltage timing input vector, and a grid-connected current timing input vector, respectively, according to the time dimension;

[0020] Extracting time series correlation features from the energy storage voltage time series input vector, the grid voltage time series input vector and the grid-connected current time series input vector to obtain an energy storage voltage time series correlation feature vector, a grid voltage time series correlation feature vector and a grid-connected current time series correlation feature vector;

[0021] Performing feature expression enhancement on the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid-connected current timing association feature vector;

[0022] Using a quasi-Bayesian probability network, the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector are processed to obtain an empty ratio a posteriori control semantic feature vector as an empty ratio a posteriori control semantic feature;

[0023] Based on the posterior control semantic feature of the empty ratio, a control result is determined.

[0024] Compared with the prior art, the present application provides a micro-inverter and method for an energy storage battery system, which collects the energy storage voltage of the energy storage battery in real time through a first voltage sensor, collects the grid voltage in real time through a second voltage sensor, and collects the grid-connected current in real time through a current sensor, and uses a data processing and analysis algorithm based on deep learning to analyze and process the timing correlation characteristics of the energy storage voltage, the grid voltage, and the grid-connected current, so as to adaptively generate a control result of the recommended duty ratio of the first duty ratio signal of the first power tube in the flyback converter according to the coordinated changes and correlations of the real-time operating parameters of the micro-inverter. In this way, the real-time operating parameters of the micro-inverter can be understood, so as to intelligently recommend the duty ratio of the first power tube in the flyback converter, thereby reducing energy loss, improving the energy utilization efficiency of the entire energy storage battery system, and improving the overall performance and efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 4 is a block diagram of a micro inverter for an energy storage battery system according to an embodiment of the present application.

[0027] Figure 2 Schematic diagram of the architecture of a micro inverter for an energy storage battery system according to an embodiment of the present application.

[0028] Figure 3 4 is a block diagram of a training module in a micro inverter for an energy storage battery system according to an embodiment of the present application.

[0029] Figure 4 The present invention is a flowchart of a micro-inversion method for an energy storage battery system according to an embodiment of the present application. DETAILED DESCRIPTION

[0030] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0031] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0032] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0033] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0034] A micro inverter is an electronic device used in energy storage battery systems. It is usually installed near the battery pack and connected to the battery and the power grid through wires. The micro inverter consists of an MCU controller, energy storage capacitors connected between the photovoltaic panel and the power grid, a flyback converter, an industrial frequency inverter bridge, and an EMI filter. It can convert direct current (DC) into alternating current (AC) so that the electrical energy in the battery can be used to power the power grid or other electrical equipment. Micro inverters can improve the efficiency and reliability of energy storage battery systems and are widely used in energy storage battery systems, including photovoltaic power generation systems, wind power generation systems, electric vehicle charging systems, etc. Among them, the flyback converter is a commonly used switching power supply topology used to convert direct current to alternating current or change the frequency of alternating current to ensure stable and reliable operation of the circuit.

[0035] However, traditional micro inverters have high power loss, which leads to lower overall system efficiency and poor stability. In addition, the dynamic response of traditional micro inverters is limited by control algorithms and hardware design, and cannot respond in real time and meet certain requirements of high dynamic performance or different working conditions.

[0036] Therefore, in view of the above technical problems, the technical concept of the present application is to collect the energy storage voltage of the energy storage battery in real time through the first voltage sensor, collect the grid voltage in real time through the second voltage sensor, and collect the grid-connected current in real time through the current sensor, and use the data processing and analysis algorithm based on deep learning to analyze and process the timing correlation characteristics of the energy storage voltage, the grid voltage and the grid-connected current, so as to adaptively generate the control result of the recommended duty ratio of the first duty ratio signal of the first power tube in the flyback converter according to the coordinated changes and correlations of the real-time operating parameters of the micro-inverter. In this way, the real-time operating parameters of the micro-inverter can be understood, so as to intelligently recommend the duty ratio of the first power tube in the flyback converter, thereby reducing energy loss, improving the energy utilization efficiency of the entire energy storage battery system, and improving the overall performance and efficiency of the system.

[0037] Figure 1 4 is a block diagram of a micro inverter for an energy storage battery system according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of a micro inverter for an energy storage battery system according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to an embodiment of the present application, a micro inverter 100 for an energy storage battery system includes: an energy storage voltage acquisition module 110, which is used to obtain the time series of the energy storage voltage of the energy storage battery acquired by a first voltage sensor; a grid voltage acquisition module 120, which is used to obtain the time series of the grid voltage acquired by a second voltage sensor; a grid current acquisition module 130, which is used to obtain the time series of the grid current acquired by the current sensor; a time dimension arrangement module 140, which is used to arrange the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid current according to the time dimension into an energy storage voltage time series input vector, a grid voltage time series input vector, and a grid current time series input vector; a time series correlation feature module 150, which is used to perform time series correlation feature extraction on the energy storage voltage time series input vector, the grid voltage time series input vector, and the grid current time series input vector The energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector are obtained; a feature expression enhancement module 160 is used to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid current timing association feature vector; a feature fusion module 170 is used to use a Bayesian probability network to process the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid current timing association feature vector to obtain an empty ratio a posteriori control semantic feature vector as an empty ratio a posteriori control semantic feature; and a control result module 180 is used to determine a control result based on the empty ratio a posteriori control semantic feature.

[0038] In an embodiment of the present application, the energy storage voltage acquisition module 110 is used to obtain the time series of the energy storage voltage of the energy storage battery collected by the first voltage sensor. The grid voltage acquisition module 120 is used to obtain the time series of the grid voltage collected by the second voltage sensor. The grid current acquisition module 130 is used to obtain the time series of the grid current collected by the current sensor. It should be understood that the energy storage voltage, the grid voltage and the grid current are important parameters commonly found in energy storage battery systems. In detail, the energy storage voltage refers to the voltage value corresponding to the electric energy stored in the energy storage device (such as a battery, a super capacitor, etc.). The stability of the energy storage voltage directly affects the operating stability and performance of the energy storage battery system. The grid voltage refers to the power supply voltage in the power system, usually the voltage of the AC power grid. A stable grid voltage is an important condition for ensuring normal power consumption by users. The grid current refers to the current connected to the grid from the energy storage system or other power sources, which can be used to adjust the power balance of the grid and help maintain the stable operation of the grid. Therefore, in the technical solution of the present application, the time series of the energy storage voltage of the energy storage battery collected by the first voltage sensor is obtained, the time series of the grid voltage collected by the second voltage sensor is obtained, and the time series of the grid-connected current collected by the current sensor is obtained, and the timing characteristics of the energy storage voltage, the grid voltage and the grid-connected current are analyzed and coordinated. The control strategy of the first power tube in the flyback converter can be adjusted according to actual conditions to achieve precise control of the energy storage battery system, thereby ensuring the stable operation of the energy storage battery system and improving the overall efficiency.

[0039] In the embodiment of the present application, the time dimension arrangement module 140 is used to arrange the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid-connected current into energy storage voltage timing input vectors, grid voltage timing input vectors, and grid-connected current timing input vectors according to the time dimension. Accordingly, considering that the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid-connected current respectively contain the changes and fluctuation trends of the energy storage voltage, the grid voltage, and the grid-connected current in the time dimension, this timing feature is crucial for the control of the first power tube in the flyback converter. Therefore, in order to better understand the periodic trends and change characteristic information of the energy storage voltage, the grid voltage and the grid-connected current in timing, so as to better control the output of the first power tube in the flyback converter, in the technical solution of the present application, the time series of the energy storage voltage, the time series of the grid voltage and the time series of the grid-connected current are arranged respectively according to the time dimension, so as to respectively integrate the timing information of the energy storage voltage, the grid voltage and the grid-connected current in the timing distribution, so as to better capture the periodic changes and coordination laws of these data in the future, so as to obtain more representative energy storage voltage timing input vector, grid voltage timing input vector and grid current timing input vector, which is conducive to better understanding and analyzing the operation of the energy storage battery system, so as to more accurately control the first power tube in the flyback converter in the future.

[0040] In an embodiment of the present application, the timing correlation feature module 150 is used to extract the timing correlation features of the energy storage voltage timing input vector, the grid voltage timing input vector and the grid current timing input vector to obtain the energy storage voltage timing correlation feature vector, the grid voltage timing correlation feature vector and the grid current timing correlation feature vector. Specifically, in an embodiment of the present application, the timing correlation feature module is used to: pass the energy storage voltage timing input vector, the grid voltage timing input vector and the grid current timing input vector through a timing correlation feature extractor based on a one-dimensional convolution layer to obtain the energy storage voltage timing correlation feature vector, the grid voltage timing correlation feature vector and the grid current timing correlation feature vector. It should be understood that, considering that the energy storage voltage, the grid voltage and the grid-connected current are constantly changing in the time dimension, they not only have the dynamic characteristics of the time series in the entire time domain, but also have a time series change pattern in a short time, such as a fluctuation trend in the short term, a periodic change, etc., and considering that the one-dimensional convolution layer can effectively capture local time series characteristics in time series data, based on this, in the technical solution of the present application, the energy storage voltage time series input vector, the grid voltage time series input vector and the grid current time series input vector are passed through a time series correlation feature extractor based on a one-dimensional convolution layer to respectively extract the local time series feature patterns and structures of the energy storage voltage, the grid voltage and the grid-connected current, thereby obtaining a more expressive energy storage voltage time series correlation feature vector, a grid voltage time series correlation feature vector and a grid current time series correlation feature vector.

[0041] In an embodiment of the present application, the feature expression enhancement module 160 is used to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid current timing association feature vector. Specifically, in an embodiment of the present application, the feature expression enhancement module is used to: use a feature enhancer based on a reparameterization layer to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector to obtain the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid current timing association feature vector. It should be understood that in order to improve the expression ability and flexibility of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector, in the technical solution of the present application, a feature enhancer based on a reparameterization layer is used to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid current timing association feature vector. It is worth mentioning that the reparameterization layer enhances the expression ability of the feature by introducing randomness, thereby enriching the diversity and robustness of the feature, so that the model can better learn and represent complex features in the data. That is to say, the re-parameter layer calculates the mean and variance of the energy storage voltage timing-related feature vector, the grid voltage timing-related feature vector and the grid-connected current timing-related feature vector respectively, and introduces additional parameters by randomly sampling the Gaussian distribution of the energy storage voltage timing-related feature vector, the grid voltage timing-related feature vector and the grid-connected current timing-related feature vector, and then reconstructs the features of each position of the energy storage voltage timing-related feature vector, the grid voltage timing-related feature vector and the grid-connected current timing-related feature vector in the form of re-parameterization, so as to generate multiple different feature expressions, thereby obtaining the reconstructed enhanced energy storage voltage timing-related feature vector, the enhanced grid voltage timing-related feature vector and the enhanced grid-connected current timing-related feature vector with richer feature capabilities. In this way, the timing feature information of the energy storage voltage, the grid voltage and the grid-connected current can be more accurately integrated, so as to improve the model's understanding of the operation of the energy storage battery system and the control capability of the first power tube in the flyback converter.

[0042] More specifically, in an embodiment of the present application, the feature expression enhancement module is used to: use the feature enhancer based on the reparameterization layer to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid current timing association feature vector with the following reparameter enhancement formula to obtain the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid current timing association feature vector; wherein the reparameter enhancement formula is:

[0043]

[0044] in, represents the mean value of the energy storage voltage time series associated characteristic vector, represents the variance of the energy storage voltage time series associated characteristic vector, ∈ represents the θth value obtained by randomly sampling the Gaussian distribution of the energy storage voltage time series associated characteristic vector, ⊙ represents the multiplication by position point, Represents the eigenvalues ​​of each position in the enhanced energy storage voltage time series associated eigenvector.

[0045] In an embodiment of the present application, the feature fusion module 170 is used to use a Bayesian probability network to process the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector to obtain an empty ratio posterior control semantic feature vector as an empty ratio posterior control semantic feature. Accordingly, considering that there are many uncertain factors in the energy storage battery system, such as weather changes, load fluctuations, etc., these factors will affect the system's control of the working state of the first power tube in the flyback converter. Therefore, in order to more accurately control the first power tube in the flyback converter, in the technical solution of the present application, a Bayesian probability network is used to process the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector to obtain an empty ratio posterior control semantic feature vector. It is worth mentioning that the Bayesian probability network can handle uncertainty and noise between data, and is used to infer the relationship between variables, and obtain the posterior probability distribution of other variables under given observation data by inference, discover the hidden correlation and regularity in the data, and help understand the dependency between variables. Specifically, the posterior probability is the prior probability multiplied by the event probability divided by the evidence probability. That is, based on the prior information and real-time observation data, the posterior semantic distribution information of the duty cycle of the first power tube in the flyback converter is obtained, and real-time control and recommendation are performed, which helps to improve the model's understanding and prediction capabilities of the operation of the energy storage battery system.

[0046] Specifically, in the embodiment of the present application, the feature fusion module is used to: use the Bayesian probability network to process the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector using the following fusion formula to obtain the idle ratio posterior control semantic feature vector; wherein the fusion formula is:

[0047]

[0048] in is the value of each position in the enhanced energy storage voltage time series associated characteristic vector, is the value of each position in the enhanced grid voltage timing correlation feature vector, is the value of each position in the enhanced grid-connected current timing correlation characteristic vector, is the value of each position in the posterior control semantic feature vector of the empty proportion.

[0049] In the embodiment of the present application, the control result module 180 is used to determine the control result based on the a posteriori control semantic feature of the duty cycle. Specifically, in the embodiment of the present application, the control result module is used to: pass the a posteriori control semantic feature vector of the duty cycle through the duty cycle controller based on the decoder to obtain the control result, and the control result is used to represent the recommended duty cycle of the first duty cycle signal of the first power tube in the flyback converter. That is, the a posteriori control semantic feature of the duty cycle obtained by the a posteriori expression of each operating parameter in the micro-inverter is used for decoding processing, so as to recommend the first duty cycle signal of the first power tube in the flyback converter according to the real-time changes of multiple parameters and the timing coordinated correlation relationship during the operation of the micro-inverter. In this way, the operation effect of the micro-inverter can be made more intelligent, efficient and stable, while reducing the energy consumption cost and improving the overall performance and reliability of the energy storage battery system.

[0050] It is worth mentioning that those skilled in the art should be aware that before applying a deep neural network model for inference, the deep neural network model must be trained so that the deep neural network can implement specific functional capabilities.

[0051] That is, in the technical solution of the present application, the micro inverter for the energy storage battery system also includes a training module for training the one-dimensional convolutional layer-based timing correlation feature extractor, the re-parameterized layer-based feature enhancer, the Bayesian-like probability network and the decoder-based duty cycle controller.

[0052] Figure 3 is a block diagram of a training module in a micro inverter for an energy storage battery system according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 3 As shown, the training module 200 includes: a training data acquisition unit 210, which is used to acquire training data, wherein the training data includes a time series of a training energy storage voltage of the energy storage battery collected by a first voltage sensor, a time series of a training grid voltage collected by a second voltage sensor, a time series of a training grid-connected current collected by a current sensor, and a real value of a recommended duty cycle of a first duty cycle signal of a first power tube in the flyback converter; a training time dimension arrangement unit 220, which is used to arrange the time series of the training energy storage voltage, the time series of the training grid voltage, and the time series of the training grid-connected current into training data according to the time dimension. Training energy storage voltage timing input vector, training grid voltage timing input vector and training grid current timing input vector; training timing association feature extraction unit 230, used to pass the training energy storage voltage timing input vector, the training grid voltage timing input vector and the training grid current timing input vector through the timing association feature extractor based on the one-dimensional convolution layer to obtain training energy storage voltage timing association feature vector, training grid voltage timing association feature vector and training grid current timing association feature vector; training re-parameterization feature enhancement unit 240, used to use the feature enhancer based on the re-parameterization layer to enhance the training energy storage voltage timing association feature vector, The training grid voltage timing association feature vector and the training grid current timing association feature vector are subjected to feature expression enhancement to obtain the training enhanced energy storage voltage timing association feature vector, the training enhanced grid voltage timing association feature vector and the training enhanced grid current timing association feature vector; a training class Bayesian fusion unit 250 is used to use the class Bayesian probability network to process the training enhanced energy storage voltage timing association feature vector, the training enhanced grid voltage timing association feature vector and the training enhanced grid current timing association feature vector to obtain the training empty ratio posterior control semantic feature vector; a decoding loss calculation unit 260 is used to convert the training empty ratio into a posterior control semantic feature vector. The posterior control semantic feature vector of the ratio is passed through the decoder-based space ratio controller to obtain a decoding loss function value; and a model training unit 270 is used to train the one-dimensional convolutional layer-based temporal correlation feature extractor, the re-parameterized layer-based feature enhancer, the Bayesian-like probability network and the decoder-based space ratio controller based on the decoding loss function value and through gradient descent back propagation, wherein each time the training space ratio posterior control semantic feature vector is subjected to decoding regression iterative training through the decoder-based space ratio controller, the training space ratio posterior control semantic feature vector is subjected to feature aggregation optimization.

[0053] Specifically, the model training unit 270 is used to train the one-dimensional convolutional layer-based temporal correlation feature extractor, the re-parameterized layer-based feature enhancer, the Bayesian-like probability network and the decoder-based space ratio controller based on the decoding loss function value and through back propagation of gradient descent, wherein each time the training space ratio posterior control semantic feature vector is subjected to decoding regression iterative training through the decoder-based space ratio controller, the training space ratio posterior control semantic feature vector is subjected to feature aggregation optimization. It should be understood that in the above technical solution, the training enhanced energy storage voltage timing association feature vector, the training enhanced grid voltage timing association feature vector and the training enhanced grid-connected current timing association feature vector respectively express the enhanced local timing association features of the energy storage voltage of the training energy storage battery, the training grid voltage and the training grid-connected current. However, considering that the source data of the energy storage voltage of the training energy storage battery, the training grid voltage and the training grid-connected current are distributed inconsistently along the timing direction, the characteristic distribution representations of the local timing association features of the training enhanced energy storage voltage timing association feature vector, the training enhanced grid voltage timing association feature vector and the training enhanced grid-connected current timing association feature vector also have large distribution differences. Therefore, after passing through a Bayesian probability network that performs Bayesian posterior probability calculation based on the eigenvalue granularity, the training empty ratio posterior control semantic feature vector will also have the problem of insufficient feature distribution posterior aggregation, which affects the decoding regression effect of the training empty ratio posterior control semantic feature vector. Based on this, in the technical solution of the present application, each time the training empty ratio posterior control semantic feature vector is subjected to decoding regression iterative training through the decoder-based empty ratio controller, feature aggregation optimization is performed on the training empty ratio posterior control semantic feature vector.

[0054] Specifically, in an embodiment of the present application, the model training unit is used to: perform feature aggregation optimization on the training empty ratio a posteriori control semantic feature vector each time the training empty ratio controller based on the decoder performs decoding regression iterative training, using the following optimization formula to obtain an optimized training empty ratio a posteriori control semantic feature vector; wherein the optimization formula is:

[0055]

[0056] Among them, v i is the feature value of the ith position in the training space ratio posterior control semantic feature vector, is the square of the 1-norm of the training space ratio posterior control semantic feature vector V, ||V|| 2 -1 / 2is the inverse of the square root of the 2-norm of the training vacancy ratio posterior control semantic feature vector V, L is the length of the training vacancy ratio posterior control semantic feature vector V, and ε is a scaling hyperparameter, v′ i The optimization training empty ratio posterior controls the feature value of the ith position in the semantic feature vector, and log is a logarithmic function value with 2 as the base.

[0057] In particular, here, a voting cluster of the feature value set aggregation of the training duty ratio a posteriori control semantic feature vector V is represented by a low-rank structure based on the norm of the training duty ratio a posteriori control semantic feature vector V, and each feature value of the training duty ratio a posteriori control semantic feature vector V is subjected to canonical voting relative to the information perception framework, so as to map the feature values ​​belonging to the same regression class to a similar local canonical coordinate (LCC: Local Canonical Coordinates) set by a regression representation of the direction and scale of the aggregated feature distribution, thereby improving the aggregation effect of the feature set of the training duty ratio a posteriori control semantic feature vector V, and improving the decoding regression effect of the training duty ratio a posteriori control semantic feature vector V through the decoder-based duty ratio controller, that is, improving the training efficiency of the decoder-based duty ratio controller and the accuracy of the decoding value. In this way, the real-time operating parameters of the micro inverter can be understood, so as to intelligently recommend the duty ratio of the first power tube in the flyback converter, thereby reducing energy loss, improving the energy utilization efficiency of the entire energy storage battery system, and improving the overall performance and efficiency of the system.

[0058] In summary, the micro-inverter 100 for the energy storage battery system based on the embodiment of the present application is explained, which collects the energy storage voltage of the energy storage battery in real time through the first voltage sensor, collects the grid voltage in real time through the second voltage sensor, and collects the grid-connected current in real time through the current sensor, and uses the data processing and analysis algorithm based on deep learning to analyze and process the timing correlation characteristics of the energy storage voltage, the grid voltage and the grid-connected current, so as to adaptively generate the control result of the recommended duty ratio of the first duty ratio signal of the first power tube in the flyback converter according to the coordinated changes and correlations of the real-time operating parameters of the micro-inverter. In this way, the real-time operating parameters of the micro-inverter can be understood, so as to intelligently recommend the duty ratio of the first power tube in the flyback converter, thereby reducing energy loss, improving the energy utilization efficiency of the entire energy storage battery system, and improving the overall performance and efficiency of the system.

[0059] As described above, the micro-inverter 100 for the energy storage battery system according to the embodiment of the present application can be implemented in various wireless terminals, such as a server having a micro-inverter algorithm for the energy storage battery system. In one possible implementation, the micro-inverter 100 for the energy storage battery system according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the micro-inverter 100 for the energy storage battery system can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the micro-inverter 100 for the energy storage battery system can also be one of the many hardware modules of the wireless terminal.

[0060] Alternatively, in another example, the micro-inverter 100 for the energy storage battery system and the wireless terminal may also be separate devices, and the micro-inverter 100 for the energy storage battery system may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0061] Figure 4 Flow chart of a micro-inverter method for an energy storage battery system according to an embodiment of the present application. Figure 4 As shown, according to the embodiment of the present application, the micro-inverting method for an energy storage battery system includes: S110, obtaining the time series of the energy storage voltage of the energy storage battery collected by the first voltage sensor; S120, obtaining the time series of the grid voltage collected by the second voltage sensor; S130, obtaining the time series of the grid current collected by the current sensor; S140, arranging the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid current respectively according to the time dimension into an energy storage voltage timing input vector, a grid voltage timing input vector, and a grid current timing input vector; S150, extracting the timing correlation features of the energy storage voltage timing input vector, the grid voltage timing input vector, and the grid current timing input vector to obtain the energy storage voltage timing correlation characteristic vector, grid voltage timing association characteristic vector and grid current timing association characteristic vector; S160, perform feature expression enhancement on the energy storage voltage timing association characteristic vector, the grid voltage timing association characteristic vector and the grid current timing association characteristic vector to obtain enhanced energy storage voltage timing association characteristic vector, enhanced grid voltage timing association characteristic vector and enhanced grid current timing association characteristic vector; S170, use a quasi-Bayesian probability network to process the enhanced energy storage voltage timing association characteristic vector, the enhanced grid voltage timing association characteristic vector and the enhanced grid current timing association characteristic vector to obtain an empty ratio a posteriori control semantic feature vector as an empty ratio a posteriori control semantic feature; and, S180, determine a control result based on the empty ratio a posteriori control semantic feature.

[0062] Here, those skilled in the art can understand that the specific operations of each step in the micro-inverter method for the energy storage battery system have been referred to above. Figures 1 to 3 The invention has been described in detail in the description of the micro inverter for the energy storage battery system, and therefore, its repeated description will be omitted.

[0063] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A micro inverter for an energy storage battery system, characterized in that: include: An energy storage voltage acquisition module, used to acquire a time series of the energy storage voltage of the energy storage battery acquired by the first voltage sensor; A power grid voltage acquisition module, used to acquire a time series of the power grid voltage acquired by the second voltage sensor; A grid-connected current acquisition module, used to acquire the time series of the grid-connected current acquired by the current sensor; A time dimension arrangement module, used to arrange the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid-connected current into an energy storage voltage timing input vector, a grid voltage timing input vector, and a grid-connected current timing input vector according to the time dimension; A timing correlation feature module is used to extract timing correlation features from the energy storage voltage timing input vector, the grid voltage timing input vector and the grid-connected current timing input vector to obtain an energy storage voltage timing correlation feature vector, a grid voltage timing correlation feature vector and a grid-connected current timing correlation feature vector; A feature expression enhancement module is used to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid-connected current timing association feature vector; A feature fusion module, used for processing the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector using a quasi-Bayesian probability network to obtain an empty ratio a posteriori control semantic feature vector as an empty ratio a posteriori control semantic feature; A control result module, used for determining a control result based on the posterior control semantic feature of the empty ratio; The feature expression enhancement module is used to: use a feature enhancer based on a reparameterization layer to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector to obtain the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector; The feature expression enhancement module is used to: use the feature enhancer based on the re-parameterization layer to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector using the following re-parameter enhancement formula to obtain the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector; Among them, the heavy parameter strengthening formula is: in, represents the mean value of the energy storage voltage time series associated characteristic vector, represents the variance of the energy storage voltage time series associated characteristic vector, ∈ represents the θth value obtained by randomly sampling the Gaussian distribution of the energy storage voltage time series associated characteristic vector, ⊙ represents the multiplication by position point, Representing the characteristic value of each position in the enhanced energy storage voltage time series associated characteristic vector; The feature fusion module is used to: use the Bayesian probability network to process the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector using the following fusion formula to obtain the idle ratio posterior control semantic feature vector; Among them, the fusion formula is: in is the value of each position in the enhanced energy storage voltage time series associated characteristic vector, is the value of each position in the enhanced grid voltage timing correlation feature vector, is the value of each position in the enhanced grid-connected current timing correlation characteristic vector, is the value of each position in the posterior control semantic feature vector of the empty proportion; The control result module is used to: pass the duty cycle a posteriori control semantic feature vector through a decoder-based duty cycle controller to obtain a control result, and the control result is used to represent the recommended duty cycle of the first duty cycle signal of the first power tube in the flyback converter.

2. The micro inverter for energy storage battery system according to claim 1, characterized in that: The timing association feature module is used to: pass the energy storage voltage timing input vector, the grid voltage timing input vector and the grid-connected current timing input vector through a timing association feature extractor based on a one-dimensional convolution layer to obtain the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector.

3. The micro inverter for energy storage battery system according to claim 2, characterized in that: It also includes a training module for training the temporal correlation feature extractor based on the one-dimensional convolutional layer, the feature enhancer based on the re-parameterization layer, the Bayesian probability network and the decoder-based space ratio controller.

4. The micro inverter for energy storage battery system according to claim 3, characterized in that: The training module comprises: A training data acquisition unit, used to acquire training data, wherein the training data includes a time series of a training energy storage voltage of the energy storage battery acquired by a first voltage sensor, a time series of a training grid voltage acquired by a second voltage sensor, a time series of a training grid-connected current acquired by a current sensor, and a true value of a recommended duty cycle of a first duty cycle signal of a first power tube in the flyback converter; A training time dimension arrangement unit, used for arranging the time series of the training energy storage voltage, the time series of the training grid voltage, and the time series of the training grid-connected current into a training energy storage voltage timing input vector, a training grid voltage timing input vector, and a training grid-connected current timing input vector, respectively, according to the time dimension; A training timing association feature extraction unit, used for passing the training energy storage voltage timing input vector, the training grid voltage timing input vector and the training grid-connected current timing input vector through the timing association feature extractor based on the one-dimensional convolution layer to obtain a training energy storage voltage timing association feature vector, a training grid voltage timing association feature vector and a training grid-connected current timing association feature vector; A training re-parameterized feature enhancement unit is used to use the feature enhancer based on the re-parameterized layer to enhance the feature expression of the training energy storage voltage timing association feature vector, the training grid voltage timing association feature vector and the training grid-connected current timing association feature vector to obtain a training enhanced energy storage voltage timing association feature vector, a training enhanced grid voltage timing association feature vector and a training enhanced grid-connected current timing association feature vector; A training-like Bayesian fusion unit is used to use the Bayesian-like probability network to process the training-enhanced energy storage voltage timing-related feature vector, the training-enhanced grid voltage timing-related feature vector and the training-enhanced grid-connected current timing-related feature vector to obtain a training empty ratio posterior control semantic feature vector; A decoding loss calculation unit, used for passing the training space ratio posterior control semantic feature vector through the decoder-based space ratio controller to obtain a decoding loss function value; A model training unit is used to train the temporal correlation feature extractor based on the one-dimensional convolutional layer, the feature enhancer based on the reparameterization layer, the Bayesian probability network and the decoder-based space ratio controller based on the decoding loss function value and through back propagation of gradient descent, wherein each time the training space ratio posterior control semantic feature vector is subjected to decoding regression iterative training through the decoder-based space ratio controller, the training space ratio posterior control semantic feature vector is subjected to feature aggregation optimization.

5. A micro-inversion method for an energy storage battery system, characterized in that: include: Acquire a time series of the energy storage voltage of the energy storage battery collected by the first voltage sensor; Acquire a time series of a grid voltage collected by a second voltage sensor; Acquire the time series of the grid-connected current collected by the current sensor; Arrange the time series of the energy storage voltage, the time series of the grid voltage, and the time series of the grid-connected current into an energy storage voltage timing input vector, a grid voltage timing input vector, and a grid-connected current timing input vector, respectively, according to the time dimension; Extracting time series correlation features from the energy storage voltage time series input vector, the grid voltage time series input vector and the grid-connected current time series input vector to obtain an energy storage voltage time series correlation feature vector, a grid voltage time series correlation feature vector and a grid-connected current time series correlation feature vector; Performing feature expression enhancement on the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector to obtain an enhanced energy storage voltage timing association feature vector, an enhanced grid voltage timing association feature vector and an enhanced grid-connected current timing association feature vector; Using a quasi-Bayesian probability network, the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector are processed to obtain an empty ratio a posteriori control semantic feature vector as an empty ratio a posteriori control semantic feature; Determining a control result based on the posterior control semantic feature of the empty ratio; Using a feature enhancer based on a reparameterization layer to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector to obtain the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector; Use the feature enhancer based on the reparameterization layer to enhance the feature expression of the energy storage voltage timing association feature vector, the grid voltage timing association feature vector and the grid-connected current timing association feature vector using the following reparameter enhancement formula to obtain the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector; Among them, the heavy parameter strengthening formula is: in, represents the mean value of the energy storage voltage time series associated characteristic vector, represents the variance of the energy storage voltage time series associated characteristic vector, ∈ represents the θth value obtained by randomly sampling the Gaussian distribution of the energy storage voltage time series associated characteristic vector, ⊙ represents the multiplication by position point, Representing the characteristic value of each position in the enhanced energy storage voltage time series associated characteristic vector; Using the Bayesian probability network to process the enhanced energy storage voltage timing association feature vector, the enhanced grid voltage timing association feature vector and the enhanced grid-connected current timing association feature vector using the following fusion formula to obtain the idle ratio posterior control semantic feature vector; Among them, the fusion formula is: in is the value of each position in the enhanced energy storage voltage time series associated characteristic vector, is the value of each position in the enhanced grid voltage timing correlation feature vector, is the value of each position in the enhanced grid-connected current timing correlation characteristic vector, is the value of each position in the posterior control semantic feature vector of the empty proportion; The duty cycle a posteriori control semantic feature vector is passed through a duty cycle controller based on a decoder to obtain a control result, and the control result is used to represent the recommended duty cycle of the first duty cycle signal of the first power tube in the flyback converter.

6. The micro-inversion method for energy storage battery system according to claim 5, characterized in that: The energy storage voltage timing input vector, the grid voltage timing input vector and the grid current timing input vector are subjected to timing correlation feature extraction to obtain the energy storage voltage timing correlation feature vector, the grid voltage timing correlation feature vector and the grid current timing correlation feature vector, including: the energy storage voltage timing input vector, the grid voltage timing input vector and the grid current timing input vector are subjected to a timing correlation feature extractor based on a one-dimensional convolutional layer to obtain the energy storage voltage timing correlation feature vector, the grid voltage timing correlation feature vector and the grid current timing correlation feature vector.

Images