Array type wave energy flow battery energy storage system and power prediction method

Abstract

The invention discloses an array type wave energy flow battery energy storage system and a power prediction method in the technical field of ocean renewable energy sources. The flow battery energy storage system of the array type wave energy comprises a double-ring array ocean wave energy acquisition module, a deep learning prediction module MMD-CNN-BiLSTM-Attention, an all-vanadium flow battery energy storage module and a comprehensive system control module. The four modules realize information interaction and cooperative regulation and control through a high-speed data network. According to the array type flow battery energy storage system for wave energy, the input voltage and power are greatly improved, the energy consumption of power electronic conversion is reduced, the problems of power fluctuation and energy storage regulation and control in wave energy power generation are effectively solved through the energy storage system with accurate prediction and quick response, and obvious technical advantages and application prospects are shown.

Description

Technical Field

[0001] The present invention relates to the technical field of marine renewable energy, and in particular to an array-type wave energy flow battery energy storage system and a power prediction method. Background Art

[0002] Currently, marine renewable energy systems face the following major challenges in practical applications: low collection efficiency and high conversion losses, specifically:

[0003] 1. Low wave energy capture efficiency: Hybrid wave energy and solar energy harvesting methods often use a single structure or simple parallel configuration. These designs struggle to adequately address the variability of ocean conditions, such as waves and wind speeds. This results in a persistently low energy capture rate, coupled with significant energy losses during mechanical and conversion processes. Specifically, the energy capture efficiency of conventional systems generally does not exceed 40%, severely limiting their potential for large-scale engineering applications.

[0004] 2. Inadequate power forecasting accuracy: The nonlinear and dynamically changing nature of ocean energy input poses a significant challenge to traditional power forecasting methods. Forecasting methods using empirical models or simple mathematical methods cannot accurately capture the instantaneous fluctuations and non-stationary characteristics of the input signal, making it difficult for system control strategies to achieve real-time and accurate responses, impacting the overall energy balance and the stable operation of energy storage equipment.

[0005] 3. Poor synergy of flow battery energy storage systems: Current flow battery energy storage systems, when operating in conjunction with wave power and solar hybrid power generation units, generally suffer from single control strategies, response lags, and weak adaptability. This uncoordinated coupling control makes it difficult for the energy storage system to release or absorb power in a timely manner during periods of high winds and waves, with large fluctuations in energy output, thus affecting the stability and reliability of the entire system.

[0006] The closest existing technical solution may be presented as: combining some form of wave energy array (which may not have been deeply optimized for hydrodynamic coupling effects, such as a dual-ring layout), adopting some kind of power prediction model based on deep learning (such as the similar solution mentioned in the above patent, but may lack effective signal decomposition preprocessing and targeted network structure optimization), and equipped with a liquid flow battery energy storage system (whose control strategy is mainly based on real-time feedback or simple threshold management). The main shortcomings of this type of solution are: the efficiency of wave energy capture is limited by the array layout; the power prediction accuracy and robustness are limited due to the failure to effectively process the non-stationary and multi-scale characteristics of the signal; the liquid flow battery fails to perform forward-looking and refined control based on accurate predictions, resulting in poor power smoothing effect and low energy storage utilization efficiency; the entire system lacks deep integration and collaborative optimization, and fails to give full play to the comprehensive advantages of each component. Summary of the Invention

[0007] The purpose of the present invention is to provide an array-type wave energy liquid flow battery energy storage system and a power prediction method to solve the technical problems in the existing technology of hybrid power generation, power prediction and energy storage coupling control in practical applications of marine renewable energy systems. It utilizes a dual-loop wave-solar hybrid array structure to optimize equipment layout, improve energy capture efficiency, and effectively respond to nonlinear and dynamic changes in the marine environment; introduces a deep learning prediction model, and through multi-scale and multi-domain signal processing, realizes high-precision prediction of ocean energy input, thereby providing a basis for real-time system regulation; combines all-vanadium liquid flow battery energy storage technology, adopts advanced adaptive control strategies, and realizes coordinated management of the entire process of energy collection, conversion, storage and output.

[0008] In order to achieve the above object, the technical solution of the present invention is as follows:

[0009] An array-type wave energy flow battery energy storage system, comprising:

[0010] The dual-ring array ocean wave energy harvesting module uses a matrix-type wave energy converter (WEC) array with an inner and outer dual-ring layout. Through hydrodynamic optimization design, it can achieve synchronous capture of waves from multiple directions and improve energy harvesting efficiency.

[0011] The deep learning prediction module, MMD-CNN-BiLSTM-Attention, uses multi-scale pattern decomposition (MMD) to pre-process wave signals into multiple components to reduce non-stationarity. It then combines a convolutional neural network (CNN), a bidirectional long short-term memory (BiLSTM) network, and an attention mechanism (Attention) to build a deep learning model for high-precision power prediction.

[0012] The all-vanadium flow battery energy storage module, based on the output of the deep learning prediction module, adopts an integrated design and applies a voltage and current dual-loop control strategy to dynamically adjust charge and discharge instructions to achieve forward-looking energy management;

[0013] The integrated system control module integrates sensor data, prediction results and energy storage status, coordinates the operation of the above modules, and ensures the efficiency and stability of the liquid flow battery energy storage system.

[0014] Furthermore, the dual-ring array ocean wave energy collection module, the deep learning prediction module, the all-vanadium liquid flow battery energy storage module and the integrated system control module realize information interaction and coordinated regulation through a high-speed data network.

[0015] Furthermore, the dual-ring array of the dual-ring array ocean wave energy collection module is based on a ring-type composite double-ring staggered ring layout and is composed of a number of power conversion units. Each of the power conversion units contains a mechanical energy conversion device, which drives the power generation equipment through the kinetic energy of the waves to efficiently convert mechanical energy into electrical energy.

[0016] Furthermore, the deep learning prediction module decomposes the wave signal into multiple representative components through MMD, and then fuses the time series features through the CNN-BiLSTM-Attention model.

[0017] Furthermore, the control strategy of the all-vanadium redox flow battery energy storage module includes voltage-current dual-loop feedback and feedforward compensation.

[0018] A wave energy power prediction method comprises the following steps:

[0019] S1. Data preprocessing: The raw wave signal is input through the multi-scale pattern decomposition module (MMD) to preprocess the wave data and decompose the complex signal into multiple representative components.

[0020] S2. Feature extraction: We use the CNN-BiLSTM-Attention model to fuse temporal features and extract features from ocean wave data.

[0021] S3. Sequence Prediction: Predict the power output of the waves and generate preliminary prediction results of the wave power;

[0022] S4. Feedback correction: Adjust the model based on the actual output and prediction error, perform parameter optimization, and output accurate power prediction results after optimization.

[0023] Furthermore, the S2 includes the following steps:

[0024] S21. Extract local features of representative components through convolutional neural network (CNN);

[0025] S22. Capture the sequential information and dynamic features of the signal using a bidirectional long short-term memory (BiLSTM) network.

[0026] S23. Improve prediction accuracy by focusing on input features at key moments through the “Attention” mechanism.

[0027] By adopting the above technical solution, the present invention has the following advantages:

[0028] This invention provides an array-type wave energy flow battery energy storage system and power prediction method. It utilizes a dual-ring array ocean wave energy harvesting module to optimize equipment layout, improve energy capture efficiency, and effectively address the nonlinear and dynamic changes in the ocean environment. It introduces the "MMD-CNN-BiLSTM-Attention" deep learning prediction model, which achieves high-precision prediction of ocean energy input through multi-scale and multi-domain signal processing, thus providing a basis for real-time system control. Combined with all-vanadium liquid flow battery energy storage technology, it adopts advanced control strategies to achieve coordinated management of the entire energy harvesting, conversion, storage, and output process. By organically combining a dual-ring array matrix WEC layout, deep learning prediction, and all-vanadium liquid flow battery energy storage, this invention achieves efficient coordination of wave energy capture, energy prediction, and energy storage control. Compared with existing technologies, this technical solution not only addresses the shortcomings of traditional hybrid power generation, power prediction, and energy storage coupled control, but also provides a solid technical guarantee for achieving overall performance optimization in practical applications of marine renewable energy systems. This technical solution not only significantly improves the input voltage and power and reduces the energy consumption of power electronic conversion, but also effectively solves the power fluctuation and energy storage regulation problems in wave power generation through the accurate prediction and fast response energy storage system, showing obvious technical advantages and application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a schematic structural diagram of the array-type wave energy flow battery energy storage system of the present invention;

[0030] Figure 2 Schematic diagram of the double-ring array of the present invention based on the ring-type composite double-ring staggered ring layout;

[0031] Figure 3 Verify the overall scheme for integrated voltage stabilization and energy storage control of wave power generation system;

[0032] Figure 4 is a flow chart of the wave energy power prediction method of the present invention;

[0033] Figure 5 This is a diagram showing the working principle of the all-vanadium redox flow battery energy storage module of the present invention;

[0034] Figure 6 This is the dynamic load response characteristics of the all-vanadium redox flow battery energy storage module before the control strategy is implemented;

[0035] Figure 7 Dynamic load response characteristics of the all-vanadium redox flow battery energy storage module after the control strategy is implemented. DETAILED DESCRIPTION

[0036] The technical solution of the present invention is described in detail below with reference to the accompanying drawings. It should be noted that, in this document, relational terms such as first and second, etc., are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include," "comprise," or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or apparatus.

[0037] An array of wave energy flow battery energy storage system is as follows Figure 1 As shown, the system comprises a dual-ring array ocean wave energy harvesting module, a deep learning prediction module (MMD-CNN-BiLSTM-Attention), an all-vanadium redox flow battery energy storage module, and an integrated system control module. These four modules interact and coordinate via a high-speed data network, forming a complete system for real-time monitoring, feedback regulation, and energy balance.

[0038] The dual-ring array ocean wave energy harvesting module utilizes a dual-ring array design to overcome the low efficiency of traditional single-ring arrays in capturing waves from multiple directions. Specifically, the dual-ring array consists of several power conversion units, each containing a mechanical energy conversion device. This unit uses the kinetic energy of the waves to drive a power generation device, efficiently converting mechanical energy into electrical energy. This structure leverages the advantages of a two-ring arrangement, not only achieving simultaneous energy capture from different directions but also facilitating expansion and maintenance through its modular design. Figure 2 The structure of the double ring array and the connection between each unit are demonstrated. Each key node and control interface are marked in detail in Chinese to ensure full disclosure of the technology. The double ring array of the double ring array ocean wave energy collection module is based on a ring-type composite double ring staggered ring layout. Figure 2 As shown, Figure 3 The overall scheme for the verification of the integrated voltage stabilization and energy storage control of the wave energy power generation system is shown. The matrix wave energy converter WEC array with an inner and outer double ring layout is designed through hydrodynamic optimization to achieve synchronous capture of waves in multiple directions and improve energy collection efficiency. This layout can effectively capture wave energy from multiple directions and optimize the continuous utilization of wave energy through structural design to achieve multiple energy captures for each wave. Figure 3 As shown. Figure 3As shown at point A in the center, a single wave impact can be utilized in multiple ways by different WEC devices within the circle. A matrix-like WEC layout, comprised of a dual-ring array, ensures even distribution of wave energy during capture, significantly improving input voltage and power levels. This layout optimizes energy collection efficiency and reduces subsequent losses caused by low input voltage during power electronics conversion and transmission, facilitating efficient voltage regulation and energy storage across the entire system. The flexible solar panel arrangement allows for inward or outward adjustment, maximizing sunlight reception and enabling 360-degree power generation, optimizing and balancing solar energy and wave energy synergy. Furthermore, this layout maintains system stability and efficient energy capture in areas with high wave energy. The ring layout's high dynamic adaptability and structural integrity, combined with the dispersed and staggered arrangement between devices to minimize interference, enhances overall system stability. This layout's design combines strong environmental adaptability and flexibility, making it ideal for navigating changing marine environments.

[0039] The deep learning prediction module, MMD-CNN-BiLSTM-Attention, addresses the non-stationarity and complexity of wave energy input signals. It preprocesses the wave signals using multi-scale pattern decomposition (MMD), decomposing them into multiple components to reduce non-stationarity. Combining a convolutional neural network (CNN), a bidirectional long short-term memory (BiLSTM) network, and an attention mechanism (Attention), this deep learning model achieves high-precision power prediction. "MMD" refers to the multi-scale pattern decomposition module, which preprocesses wave data and decomposes the complex signal into multiple representative components. "CNN" refers to the convolutional neural network, which primarily extracts local features. "BiLSTM" stands for the bidirectional long short-term memory (BiLSTM) network, which captures long-term dependencies and bidirectional dynamic features in the signal. The "Attention" mechanism enhances the model's focus on input features at critical moments, improving prediction accuracy. To address the instability of output power during wave energy generation, the present invention incorporates the MDD-CNN-BiLSTM-Attention deep learning algorithm, which achieves accurate power generation prediction through a front-end and back-end feedback mechanism. This prediction system enables real-time control of power generation and storage processes, improving the overall system's responsiveness and stability to environmental changes and ensuring efficient operation even in complex sea conditions.

[0040] A wave energy power prediction method includes the following steps: Figure 4 As shown:

[0041] S1. Data preprocessing: The raw wave signal is input through the multi-scale pattern decomposition module (MMD) to preprocess the wave data and decompose the complex signal into multiple representative components.

[0042] S2. Feature extraction: We use the CNN-BiLSTM-Attention model to fuse temporal features and extract features from ocean wave data.

[0043] Among them, S2 includes the following specific steps:

[0044] S21. Extract local features of representative components through convolutional neural network (CNN);

[0045] S22. Capture the sequential information and dynamic features of the signal using a bidirectional long short-term memory (BiLSTM) network.

[0046] S23. Improve prediction accuracy by focusing on input features at key moments through the “Attention” mechanism.

[0047] S3. Sequence Prediction: Predict the power output of the waves and generate preliminary prediction results of the wave power;

[0048] That is, the non-stationary wave power signal is first effectively decomposed using MMD, and then the decomposed multi-scale components are input into a specific deep learning network containing convolutional neural network (CNN), bidirectional long short-term memory (BiLSTM) and attention mechanism for prediction.

[0049] S4. Feedback correction: Adjust the model based on the actual output and prediction error, perform parameter optimization, and output accurate power prediction results after optimization.

[0050] To achieve the stabilization of ocean energy output and the continuity of the overall energy supply of the system, an all-vanadium liquid flow battery energy storage module is integrated in the embodiment of the present application. Specifically, the all-vanadium liquid flow battery energy storage module adopts high-efficiency energy conversion, storage and discharge management technology, applies a voltage and current dual-loop control strategy, dynamically adjusts the charge and discharge instructions, and realizes forward-looking energy management; it can quickly absorb and store excess energy when the wave energy output is unstable or exceeds the immediate needs of the system, and releases the stored energy to ensure system stability when the energy supply is insufficient. It integrates sensor data acquisition, real-time prediction and control strategies, and realizes coordinated operation between modules through closed-loop control. Figure 5 (a) is the integrated schematic diagram and circuit connection diagram of the all-vanadium redox flow battery energy storage module. Figure 5 (b) is the working flow diagram of the all-vanadium liquid flow battery energy storage module. Figure 5 (a) Figure 5 As shown in Figure (b), all signal interfaces, control units, and data transmission paths of the vanadium redox flow battery energy storage module are labeled in Chinese, ensuring the integrity and feasibility of the disclosed content. Furthermore, the control strategy of the vanadium redox flow battery energy storage module includes dual-loop voltage and current feedback and feedforward compensation.

[0051] Figure 6 The raw response characteristics of an all-vanadium redox flow battery energy storage module under highly dynamic load conditions are demonstrated without applying a specific control strategy. Over a 50-time-unit observation period, the battery state of charge (SOC) decreases nonlinearly from approximately 0.5 to approximately 0.479, with the downward trajectory reflecting dramatic fluctuations in current and voltage. The current curve exhibits irregular, high-amplitude pulses, with peak positive discharge currents reaching approximately 2000A, accompanied by significant negative current peaks (approximately -800A), and plateaus near zero current, indicating rapid changes in load demand and potential energy recovery events. Correspondingly, the battery terminal voltage exhibits significant instability, initially dropping from approximately 580V to approximately 400V during the discharge pulse, rebounding to approximately 600V during the negative current period, and remaining in the 500V to 520V range during the low-current plateau period. This uncontrolled behavior reveals the significant electrochemical and electrical stresses that the battery undergoes when responding to rapidly changing power demands.

[0052] Compared to Figure 6 As shown, Figure 7 It shows that after applying the control strategy of the all-vanadium liquid flow battery energy storage module (voltage and current dual-loop feedback and feedforward compensation), the same battery system has significantly improved response characteristics under similar initial conditions (initial SOC is about 0.5, and it drops to about 0.479 after 50 time units). Although in the initial stage (about t < 6 time units), the system still undergoes a short period of dynamic adjustment, the current has positive and negative pulses (the peak value can still reach 2000A) and the voltage fluctuates between 400V and 600V, reflecting the response process of the control system to the initial transient. However, after t≈6, the effect of the control strategy becomes apparent: the current fluctuation is effectively suppressed and stabilized in the forward discharge range of about 800-900A, eliminating the Figure 6 The sharp negative pulses and long zero-current plateaus observed in the original image are now eliminated. Accordingly, the voltage stabilizes at around 500V, significantly reducing the previous wide oscillations. This results in a smoother, nearly linear decrease in SOC after the initial adjustment. Figure 7 The effectiveness of the implemented control strategy of the all-vanadium liquid flow battery energy storage module in smoothing current, stabilizing voltage and reducing battery operating stress was clearly demonstrated, significantly improving the operating stability of the all-vanadium liquid flow battery energy storage module under dynamic load.

[0053] In order to solve the intermittent problem of wave energy generation, the present invention adopts all-vanadium liquid flow battery as energy storage unit. Figure 5 As shown in the figure, the dynamic load response performance of the all-vanadium redox flow battery has been significantly improved before and after the application of the control strategy. Figure 6 、 Figure 7This enables the all-vanadium flow battery energy storage module to exhibit faster response speed and higher stability during energy storage and energy allocation, thereby further improving the overall energy conversion efficiency.

[0054] In the specific embodiments, this section lists a variety of alternative solutions, ranging from process parameters and method steps to complete technical solutions. Any alternative solutions that can achieve the same or similar technical effects as the original solution by adjusting or replacing the layout, algorithm, energy storage system, and its control strategy without deviating from the technical spirit and basic objectives of the present invention (i.e., efficient wave energy capture, precise predictive control, and stable energy storage management) shall be considered within the scope of protection of the present invention.

[0055] (1) Alternative layout options for wave energy converters (WECs)

[0056] 1. Annular and multi-layer concentric array deformation:

[0057] In addition to the double-ring array layout mentioned in the present invention, a single-ring or multi-layer concentric ring array can also be used to distribute the wave energy conversion units at different radii or levels. By adjusting the radius, number and arrangement angle of the units, efficient energy capture and collection can still be achieved.

[0058] 2. Distributed or grid layout:

[0059] WEC units can be dispersed into several clusters and laid out in a grid, staggered or other irregular forms. Combined with adaptive control strategies, they can not only make full use of marine resources, but also produce a local wave energy concentration effect, thereby increasing the input voltage and power and improving the overall power generation efficiency.

[0060] 3. Optimize linear or reconfigurable arrays:

[0061] Using linear arrays or small clusters, the system dynamically adjusts the position, connection topology, and operating status of the floats based on real-time wave conditions, while still achieving stable energy capture and output power. Once changes in sea conditions are detected, the array layout can be reconfigured within a certain range to maintain robust power generation performance.

[0062] (2) Alternatives to wave energy prediction

[0063] 1. Multiple deep learning model replacements:

[0064] In addition to the MDD-CNN-BiLSTM-Attention deep learning architecture of the present invention, various time series prediction models such as Transformer, CNN-GRU, TCN (temporal convolutional network) or standard LSTM / GRU can also be used. As long as they can effectively capture the nonlinear and periodic characteristics of wave energy, they can achieve prediction accuracy and operation efficiency similar to the original solution.

[0065] 2. Traditional Machine Learning and Time Series Models:

[0066] If combined with appropriate feature engineering and parameter tuning, statistical models such as support vector regression, random forest, and ARIMA, SARIMA, and Prophet can also achieve considerable forecast accuracy. For scenarios with significant cyclical fluctuations, physical wave models can be combined with data-driven machine learning models.

[0067] 3. Multi-model integration or adaptive fusion:

[0068] By weighting or nonlinearly fusing the results of multiple prediction methods, the robustness and accuracy of the prediction can be further improved; in a real-time environment, it is also possible to combine online learning, rolling prediction updates, or error-based adaptive adjustments to allow the prediction model to be continuously optimized as wave conditions change.

[0069] (3) Energy storage system alternatives

[0070] 1. Different types of red oxygen flow batteries:

[0071] While all-vanadium redox flow batteries are used in this invention, zinc-bromine, iron-chromium, or all-organic redox flow batteries can also be used if energy and power decoupling requirements are met. These systems have slight differences in chemical properties and safety performance, but by tailoring charge-discharge and thermal management strategies, they can achieve similar energy storage and dispatch performance to all-vanadium redox flow batteries.

[0072] 2. Hybrid energy storage or other chemical energy storage:

[0073] All-vanadium redox flow batteries can also be combined with lithium-ion batteries and supercapacitors to form a "fast response + large capacity" hybrid energy storage architecture. Alternatively, using a single lithium-ion battery (such as lithium iron phosphate or sodium-ion battery) combined with refined control methods can also achieve the goal of smoothing wave power generation fluctuations.

[0074] 3. Non-battery energy storage:

[0075] In some specific scenarios, other modes such as flywheel energy storage, compressed air energy storage, or gravitational potential energy (pumped storage) can also be used for energy buffering. As long as they can achieve the same peak-shaving and valley-filling, voltage stabilization and current limiting effects as the original solution, they can be considered as equivalent alternatives to the present invention.

[0076] (IV) Control strategies and alternative methods and procedures

[0077] 1. Adaptive or hierarchical control:

[0078] By dividing the wave energy system into several levels, managing energy through sub-control loops at each level, and then transmitting information to the higher-level aggregation or grid-connected control, the same real-time regulation and power balancing goals as the third part can be achieved within the hardware's permitted range.

[0079] 2. Other types of prediction and control algorithms:

[0080] Intelligent control methods such as model predictive control (MPC), sliding mode control, fuzzy control or expert systems can effectively manage energy storage units, thereby ensuring output voltage stability and reducing battery cycle load.

[0081] 3. Parameter optimization and reconfigurable process:

[0082] In specific implementations, corresponding modifications and adjustments can be made from structural parameters such as WEC size, arrangement radius, inclination angle, to network depth, convolution kernel size, learning rate and attention mechanism configuration in the prediction algorithm, to the charging and discharging thresholds, temperature control strategy, grid connection point power limit, etc. of the energy storage system. As long as the basic functions and effects still conform to the technical gist of the present invention, they fall within the scope of protection of the present invention.

[0083] (V) Complete technical solution alternative and comprehensive description

[0084] 1. Overall replacement from layout to energy storage:

[0085] Besides replacing specific components or algorithms, a comprehensive system solution with similar principles but different technical approaches can also be adopted. As long as the three aspects of wave energy capture, energy prediction, and energy storage regulation can be coordinated to achieve high-efficiency, low-fluctuation power generation output and dispatchable power utilization similar to the present invention, it will constitute an equivalent technical solution to the present invention.

[0086] 2. Addition, deletion or combination of functional modules:

[0087] In specific application scenarios, integrating a deep prediction module might not be necessary, or the energy storage component could be reduced. Conversely, it might be possible to add a backup power source or even upgrade to a multi-energy complementary system. As long as the core objectives of the present invention—dynamic prediction, smooth regulation, and efficient output—are achieved, any modifications are considered equivalent extensions.

[0088] In summary, the above-mentioned alternative approaches encompass multiple aspects, including WEC layout, prediction models, energy storage systems, and control strategies. These alternative approaches can be appropriately adjusted in terms of process parameters and method steps, or substituted or recombined into complete solutions to achieve the same or similar technical effects described in Section 3. Any variations or improvements made using different means within the technical principles of this invention should be considered within the scope of protection of this invention.

[0089] Finally, it should be pointed out that although the present invention has been described with reference to the current specific embodiments, ordinary technicians in this technical field should realize that the above embodiments are only used to illustrate the present invention and are not used to limit the present invention. Various equivalent changes or substitutions can be made without departing from the concept of the present invention. Therefore, as long as the changes and modifications to the above embodiments are within the scope of the essential spirit of the present invention, they will fall within the scope of the claims of the present invention.

Claims

1. An array-type wave energy flow battery energy storage system, characterized in that: include: The dual-ring array ocean wave energy harvesting module uses a matrix-type wave energy converter (WEC) array with an inner and outer dual-ring layout. Through hydrodynamic optimization design, it can achieve synchronous capture of waves from multiple directions and improve energy harvesting efficiency. The deep learning prediction module, MMD-CNN-BiLSTM-Attention, uses multi-scale pattern decomposition (MMD) to pre-process wave signals into multiple components to reduce non-stationarity. It then combines a convolutional neural network (CNN), a bidirectional long short-term memory (BiLSTM) network, and an attention mechanism (Attention) to build a deep learning model for high-precision power prediction. The all-vanadium flow battery energy storage module, based on the output of the deep learning prediction module, adopts an integrated design and applies a voltage and current dual-loop control strategy to dynamically adjust charge and discharge instructions to achieve forward-looking energy management; The integrated system control module integrates sensor data, prediction results and energy storage status, coordinates the operation of the above modules, and ensures the efficiency and stability of the liquid flow battery energy storage system.

2. The array-type wave energy flow battery energy storage system according to claim 1, characterized in that: The dual-ring array ocean wave energy collection module, the deep learning prediction module, the all-vanadium liquid flow battery energy storage module and the integrated system control module realize information interaction and coordinated regulation through a high-speed data network.

3. The array-type wave energy flow battery energy storage system according to claim 2, characterized in that: The dual-ring array of the dual-ring array ocean wave energy collection module is based on a ring-type composite dual-ring staggered ring layout and consists of several power conversion units. Each of the power conversion units contains a mechanical energy conversion device, which drives the power generation equipment through the kinetic energy of the waves to efficiently convert mechanical energy into electrical energy.

4. The array-type wave energy flow battery energy storage system according to claim 1, characterized in that: The deep learning prediction module decomposes the wave signal into multiple representative components through MMD, and then integrates the time series features through the CNN-BiLSTM-Attention model.

5. The array-type wave energy flow battery energy storage system according to claim 1, characterized in that: The control strategy of the all-vanadium redox flow battery energy storage module includes voltage-current dual-loop feedback and feedforward compensation.

6. A wave energy power prediction method, characterized in that: The following steps are involved: S1. Data preprocessing: The raw wave signal is input through the multi-scale pattern decomposition module (MMD) to preprocess the wave data and decompose the complex signal into multiple representative components. S2. Feature extraction: We use the CNN-BiLSTM-Attention model to fuse temporal features and extract features from ocean wave data. S3. Sequence Prediction: Predict the power output of the waves and generate preliminary prediction results of the wave power; S4. Feedback correction: Adjust the model based on the actual output and prediction error, perform parameter optimization, and output accurate power prediction results after optimization.

7. The array-type wave energy flow battery energy storage system according to claim 6, characterized in that: The S2 comprises the following steps: S21. Extract local features of representative components through convolutional neural network (CNN); S22. Capture the sequential information and dynamic features of the signal using a bidirectional long short-term memory (BiLSTM) network. S23. Improve prediction accuracy by focusing on input features at key moments through the “Attention” mechanism.

Images