Miniature photovoltaic inverter high-efficiency energy conversion device based on double MPPT (Maximum Power Point Tracking) topology

Abstract

The invention discloses a miniature photovoltaic inverter high-efficiency energy conversion device based on double MPPT topology, and relates to the technical field of photovoltaic inversion, and the key point of the technical scheme is that the miniature photovoltaic inverter high-efficiency energy conversion device comprises a photovoltaic input end, a double MPPT control module, a self-adaptive impedance matching network, a power conversion circuit, an intelligent filter network and a control processing unit of a dual-core architecture; the double MPPT control module dynamically switches an algorithm to track the maximum power point; the self-adaptive impedance matching network realizes wide-range closed-loop impedance matching through an adjustable inductor and a capacitor assembly in combination with high-frequency small-signal injection type impedance detection; the power conversion circuit is integrated with double Boost, full-bridge inversion and LLC resonance type soft switching auxiliary circuits, so that the switching loss is reduced; and the intelligent filter network adaptively adjusts parameters to suppress harmonic waves. Dynamic working condition adaptation is achieved, the energy conversion efficiency and the power density are remarkably improved, low-harmonic-wave high-quality alternating current is output, and stability is high.

Description

A high-efficiency energy conversion device for micro photovoltaic inverters based on dual MPPT topology Technical Field

[0001] This invention relates to the field of photovoltaic inverter technology, and more specifically, to a high-efficiency energy conversion device for a micro photovoltaic inverter based on a dual MPPT topology. Background Technology

[0002] With the rapid popularization of distributed photovoltaic (PV) and residential PV systems, micro PV inverters have become the mainstream development direction in the industry due to their advantages such as module-level maximum power point tracking (MPPT), flexible installation, and high reliability. Their core requirement is to achieve efficient energy conversion under complex and ever-changing actual operating conditions. However, the output characteristics of PV modules are highly susceptible to factors such as fluctuations in light intensity, changes in ambient temperature, and local shading (e.g., building shadows, foliage obstruction), exhibiting strong nonlinear characteristics. Especially under local shading, the power-voltage (PV) curve of the PV array will show multiple local maximum power points (LMPPs), with only one global maximum power point (GMPP). This places stringent requirements on the inverter's adaptability to operating conditions, making the coordinated operation of impedance dynamic matching and precise MPPT tracking crucial to determining energy conversion efficiency.

[0003] Existing micro-photovoltaic inverter technologies have significant shortcomings, making it difficult to meet the high-efficiency conversion requirements under complex operating conditions: First, impedance matching often employs fixed-parameter inductor-capacitor (LC) networks. The parameters of these networks are designed only for rated illumination, temperature, and load conditions. When actual operating conditions deviate from the design values, the input impedance of the photovoltaic module and the inverter load impedance cannot maintain a conjugate match, leading to power reflection and transmission losses. Second, MPPT control often relies on a single algorithm (such as the traditional perturbation-observation method or the incremental conductance method). While these algorithms can generally track the maximum power point under uniform illumination, they are prone to getting trapped in low-power LMPP scenarios with multi-peak PV curves caused by partial shading. First, traditional algorithms cannot lock GMPP and are difficult to balance dynamic response speed and steady-state tracking accuracy. Second, although a few improved solutions attempt to introduce impedance regulation or intelligent MPPT algorithms, they have technical bottlenecks. Impedance detection often uses single-frequency disturbance point-by-point measurement or wideband disturbance measurement. The former is time-consuming, and the latter is easily affected by converter frequency coupling interference, resulting in insufficient detection accuracy. Intelligent MPPT algorithms (such as neural networks and fuzzy control) have problems such as large computational load, slow convergence speed, and complex parameter tuning, making them difficult to apply in engineering. In addition, existing technologies generally lack a deep collaborative mechanism between impedance matching and MPPT control. The independent operation of the two leads to adaptation lag, further reducing energy conversion efficiency.

[0004] The aforementioned technical defects directly lead to a series of serious problems: Under fluctuating light or localized shading, the traditional MPPT algorithm gets stuck in LMPP (Limited-Low Power Profit), and combined with the mismatch effect of fixed impedance matching, this causes a significant decrease in the output power of photovoltaic modules, with energy conversion efficiency dropping by 10%–30% compared to ideal operating conditions. Impedance mismatch also exacerbates the voltage and current stress on power switching transistors, causing a surge in switching losses and inducing current harmonic distortion. This not only increases the heat load on devices but may also lead to broadband oscillations between the inverter and the grid, threatening the safety of system operation. Inefficient impedance detection methods and lagging matching algorithms prevent the inverter from responding to changes in operating conditions in real time. Even under uniform illumination, continuous power loss occurs due to untimely impedance adaptation. These problems combined significantly reduce the power generation revenue of photovoltaic systems, shorten the lifespan of inverters, and severely restrict the promotion and application of micro photovoltaic inverters in complex environments, becoming a core technical pain point that the industry urgently needs to address. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a high-efficiency energy conversion device for a micro photovoltaic inverter based on a dual MPPT topology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a high-efficiency energy conversion device for a micro photovoltaic inverter based on a dual MPPT topology, comprising: a photovoltaic input terminal connected to two independent photovoltaic modules; a dual MPPT control module, including a first MPPT controller and a second MPPT controller, wherein the first MPPT controller and the second MPPT controller are respectively connected to the two photovoltaic modules through the photovoltaic input terminal; an adaptive impedance matching network, which, through adjustable inductor components and adjustable capacitor components, combined with a machine learning algorithm module and high-frequency small-signal injection impedance detection, performs wide-range impedance adjustment on the DC power input to the photovoltaic input terminal, forming a closed-loop impedance matching link; a power conversion circuit, employing a dual Boost boost module and a full-bridge inverter module, integrating an LLC resonant soft-switching auxiliary circuit, wherein the dual Boost boost module includes two independent boost circuits, respectively connected to the first MPPT controller and the second MPPT controller, and the power conversion circuit integrates an LLC resonant soft-switching auxiliary circuit; The power conversion circuit achieves zero-voltage turn-on and zero-current turn-off of the main switch through precise timing control, and sequentially performs secondary boosting and inversion processing on the DC power after impedance matching and boosting. The intelligent filter network adaptively adjusts the filter parameters of the inverted AC power to suppress harmonics. The control processing unit adopts a dual-core processor architecture, whose fast tracking processing core collects the power generation signals of the two photovoltaic modules to calculate the rate of change of light, and controls the first MPPT controller and the second MPPT controller to dynamically switch the MPPT algorithm and output PWM control signals to drive the corresponding boost circuit. The system monitoring processing core of the control processing unit performs system monitoring, and the two cores achieve data sharing through a high-speed bus. The DC power output from the photovoltaic modules is input through the photovoltaic input terminal, and sequentially passes through the maximum power point tracking of the first MPPT controller and the second MPPT controller, impedance matching of the adaptive impedance matching network, boosting of the boost circuit, secondary boosting and inversion processing of the power conversion circuit, and harmonic suppression of the intelligent filter network, before outputting high-quality AC power through the inverter output terminal.

[0007] Furthermore, the first MPPT controller and the second MPPT controller are each configured with an independent MCU and an independent Boost converter circuit, and each MPPT is equipped with a high-precision ADC. The MPPT algorithm includes the perturbation-observation method and the incremental conductance method. The illuminance variation rate of the photovoltaic module output power is calculated by multiple consecutive samplings, and the perturbation-observation method and the incremental conductance method are automatically switched according to the magnitude of the illuminance variation rate. When switching, the latest iteration result of the previous algorithm is retained as the initial value. After the algorithm converges, it outputs a PWM control signal with an adjustable duty cycle to drive the switching transistor of the corresponding Boost converter circuit.

[0008] Furthermore, the adaptive impedance matching network includes: an adjustable inductor component, employing a ferrite core structure, with the core air gap adjusted via a core air gap adjustment mechanism to achieve wide-range continuous adjustment; an adjustable capacitor component, composed of a varactor diode array, with wide-range continuous adjustment achieved through reverse bias voltage adjustment; an impedance detection unit, employing a high-frequency small-signal injection method, simultaneously acquiring voltage and current signals before and after the injection point, separating the high-frequency detection signal from the main power signal to achieve interference-free impedance detection; and a matching algorithm processor, integrating a machine learning algorithm module, which, based on light intensity, ambient temperature, photovoltaic module internal resistance, and load impedance parameters, outputs optimal matching parameters using an offline training + online fine-tuning mode to achieve rapid convergence.

[0009] Furthermore, the LLC resonant soft-switching auxiliary circuit includes a resonant inductor, a resonant capacitor, an auxiliary switching transistor, and a control timing circuit. The resonant inductor is connected in series to the main power path, and the resonant capacitor and the auxiliary switching transistor are connected in parallel and then connected across the drain and source of the main switching transistor. The control timing circuit captures the voltage zero-crossing point and current zero-crossing point of the main switching transistor through a detection element, generates timing commands, and controls the complementary conduction of the auxiliary switching transistor and the main switching transistor to achieve zero-voltage turn-on and zero-current turn-off of the main switching transistor.

[0010] Furthermore, the intelligent filtering network includes a multi-stage filtering circuit and a filtering parameter adjustment unit; the multi-stage filtering circuit includes an LC filtering stage and an active filtering stage, the LC filtering stage is composed of inductors and capacitors, and the active filtering stage is composed of operational amplifiers forming a band-stop filtering structure to compensate for specific harmonics; the filtering parameter adjustment unit adopts a programmable adjustment element, which automatically adjusts the damping parameters and gain of the filtering circuit by collecting the load impedance characteristics, thereby achieving adaptive matching of the filtering parameters.

[0011] Furthermore, the dual-core processor architecture of the control processing unit includes a fast tracking processing core and a system monitoring processing core, which are dedicated to performing MPPT algorithm calculations and system monitoring functions, respectively. Dual-core data sharing is achieved through a high-speed bus. The system monitoring functions include temperature monitoring, fault diagnosis, power grid synchronization, and communication interaction.

[0012] Furthermore, it also includes a phase change material heat dissipation system, which includes: a phase change material heat storage unit filled with paraffin-based composite phase change material and an internal fin structure to enhance heat transfer efficiency; a liquid level detection device for monitoring the melting degree of the phase change material; and an active air cooling unit including a temperature-controlled fan and air ducts; when the temperature of the power device reaches a set threshold, the phase change material begins to melt, and after the liquid level detection device detects that the phase change material has completely melted, the control processing unit starts the active air cooling unit.

[0013] Furthermore, it also includes a multi-stage overvoltage protection circuit, which includes: a fast-response protection stage, consisting of multiple sets of TVS diode arrays with different clamping voltages connected in parallel, adapted to the photovoltaic input terminal, DC bus, and full-bridge inverter module input side respectively, to achieve transient overvoltage energy absorption; a slow-recovery protection stage, including a relay protection circuit, which consists of multiple sets of double-pole double-throw normally open relays connected in series in the key power circuit to achieve complete electrical isolation; and a voltage sampling circuit, which consists of a voltage divider network composed of high-precision resistors to monitor the voltage of each key node in real time. When a continuous overvoltage is detected, the relay protection circuit is triggered to disconnect, and it automatically resets after the voltage recovers and stabilizes.

[0014] Further, the process includes the following steps: The photovoltaic module is connected to the device via a waterproof connector; the first MPPT controller and the second MPPT controller are activated, and a dynamic switching algorithm based on the rate of change of light is used to output a PWM signal to drive the Boost circuit; the impedance detection unit collects the internal resistance of the photovoltaic module and the load impedance, and combines this with data on light intensity and ambient temperature, outputting optimal matching parameter instructions through a machine learning algorithm module to adjust the adjustable inductor and adjustable capacitor components; the DC power is input to the full-bridge inverter module after passing through the dual Boost modules, and the soft-switching auxiliary circuit achieves zero-voltage turn-on and zero-current turn-off of the main switch tube through precise timing control; the inverted AC power is subjected to harmonic suppression by an intelligent filter network, and the filter parameters are adaptively adjusted according to the load impedance; during operation, if a phase change material heat dissipation system is configured, heat dissipation is activated based on the device temperature and the melting state of the phase change material; if a multi-level overvoltage protection circuit is configured, the voltage is monitored in real time to achieve transient clamping and continuous overvoltage cutoff; the dual-core processor architecture of the control processing unit works in tandem to quickly track and process the real-time optimization algorithm parameters, and the system monitoring processing core monitors the system status, ultimately outputting a low-harmonic-distortion sinusoidal AC power.

[0015] Compared with existing technologies, this invention has the following advantages: By deeply collaborating with a dual MPPT control module and an adaptive impedance matching network, this invention solves the core problem of fixed impedance matching in existing technologies being unable to adapt to changes in operating conditions. The dual MPPT controller accurately tracks the maximum power point based on the dynamic switching disturbance observation method of illumination change rate and the incremental conductance method. The adaptive impedance matching network acquires impedance parameters in real time through high-frequency small-signal injection detection technology, and quickly outputs the optimal matching command by combining machine learning algorithms. It dynamically adjusts the air gap of the adjustable inductor core and the bias voltage of the adjustable capacitor to achieve wide-range closed-loop impedance matching, significantly reducing power transmission loss and significantly improving energy conversion efficiency and overall power density. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and constitute a part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 is a schematic diagram of the operation flow of the invention; Figure 2 is a block diagram of the workflow and control guarantee of the invention; Figure 3 is a schematic diagram of the internal structure of the conversion device in the invention; Figure 4 is a circuit diagram of the adaptive impedance matching network in the invention.

[0017] 1. Photovoltaic input terminal; 2. First MPPT controller; 3. Second MPPT controller; 4. Control processing unit; 5. Power conversion circuit; 6. Inverter output terminal. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0020] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0021] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0022] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0023] As shown in Figures 1-4, this invention provides a high-efficiency energy conversion device for a micro photovoltaic inverter based on a dual MPPT topology. It adopts a compact integrated shell structure, with the outer shell being injection molded from high-strength engineering plastic. The interior is divided into a lower power area and an upper control area by an aluminum alloy frame, following the energy flow logic of "left input - right output". The core components include a photovoltaic input terminal 1, a first MPPT controller, a second MPPT controller 3, a power conversion circuit 5, an inverter output terminal 6, a control processing unit 4, an adaptive impedance matching network, a soft-switching auxiliary circuit, an intelligent filtering network, a phase change material heat dissipation system, and a multi-level overvoltage protection circuit. Each module works collaboratively through a high-speed bus or hard connection.

[0024] Photovoltaic input terminal 1 is located on the left side of the housing and is equipped with two parallel MC4 waterproof connectors. It supports a maximum input current of 14A and a starting voltage of 22V, and is compatible with single or multiple photovoltaic modules in series. An impedance detection unit is fixed nearby to connect two independent photovoltaic modules and provide DC input to the device. The first MPPT controller 2 and the second MPPT controller 3 are arranged in parallel in the lower left part of the power area, and are connected to the two connectors of the photovoltaic input terminal 1 (i.e., the first MPPT controller 2 is connected to the first photovoltaic module, and the second MPPT controller 3 is connected to the second photovoltaic module). A dual Boost boost module front-end design is adopted, specifically, two independent Boost boost circuits are connected one-to-one with the dual MPPT control module. The output of the first MPPT controller 2 is directly connected to the first Boost boost circuit, and the output of the second MPPT controller 3 is directly connected to the second Boost boost circuit. This dual Boost boost module serves as a front-end stage, and its output is connected to the input of the full-bridge inverter module of the subsequent power conversion circuit 5 after being combined through the bus, forming a fixed power flow direction of photovoltaic input, MPPT tracking, Boost boost, and full-bridge inverter. Each Boost boost circuit is independently configured with a boost inductor, SiC MOSFET switch and freewheeling diode. Its control signal is output separately by the fast tracking processing core of the control processing unit 4 to realize independent adjustment of the two boost processes.The dual MPPT control module is built on an STM32H745ZI dual-core MCU (Cortex-M7 480MHz + Cortex-M4 240MHz). Each MPPT is independently configured with a 12-bit ADC (3.45MSPS sampling rate). This ADC is specifically designed to collect the real-time voltage and current signals of the corresponding photovoltaic module and transmits them to the fast tracking processing core (Cortex-M7) through the on-chip DMA channel, providing data support for algorithm calculation. The disturbance observation method and conductance increment method supported by the dual MPPT control module are built-in software algorithms, requiring no additional hardware configuration. The core execution logic is as follows: Disturbance observation method: adopts a closed-loop logic of "voltage disturbance - power detection - direction adjustment". Each time a disturbance occurs, the output voltage of the photovoltaic module is finely adjusted in steps of 0.05V. The power change before and after the disturbance is calculated by the ADC through two consecutive samplings (1ms interval). If the power increases, the current disturbance direction is maintained; if the power decreases, the disturbance is reversed until the power fluctuation is ≤0.5%. Conductance increment method: based on the principle that "the maximum power point is reached when dP / dV=0", the ADC is used to calculate the power change before and after the disturbance. The ADC sampling data is used to calculate the power P, conductance G=I / V, and conductance increment dG / dV in real time. When dG / dV+G>0, the adjustment is made towards increasing voltage; when dG / dV+G<0, the adjustment is made towards decreasing voltage, with an adjustment accuracy of ±0.1V. The algorithm's dynamic switching is determined by calculations based on the ADC sampling data: the rate of change of photovoltaic module output power (ΔP / Δt) is calculated through 5 consecutive samples (sampling interval 1ms). When the rate of change is ≤5% / s, the algorithm automatically switches to the conductance increment method (higher steady-state accuracy); when the rate of change is >5%, the algorithm switches to the conductance increment method. When the speed is 5%, the algorithm switches to the perturbation and observation method (which has a faster dynamic response). The algorithm converges when the error is ≤0.5% after 5 iterations. During the switching process, the latest iteration result of the previous algorithm is retained as the initial value to avoid power surges caused by voltage changes. After the algorithm is calculated, it outputs a PWM control signal with an adjustable duty cycle to drive the SiCMOSFET switch of the corresponding Boost circuit independently, realizing dual-path independent maximum power point tracking. This adapts to the differences in light and temperature of different photovoltaic modules, avoids the "barrel effect" caused by module shading and temperature differences, and improves power generation.

[0025] The power conversion circuit 5, as the core energy conversion unit, is located in the middle of the power region. It adopts a dual-Boost boost + full-bridge inverter topology and integrates an adaptive impedance matching network and soft-switching auxiliary circuit. The adaptive impedance matching network is located between the photovoltaic input terminal 1 and the boost inductor. It mainly consists of an adjustable inductor component, an adjustable capacitor component, an impedance detection unit, and a matching algorithm processor. The structure, adjustment principle, and collaborative logic of each component are as follows: The adjustable inductor component adopts an E-shaped ferrite core structure, paired with a stepper motor. It is rigidly connected to the movable block of the magnetic core through a precision threaded push rod, forming a linkage adjustment mechanism of motor-push rod-magnetic core. The stepper motor receives the pulse signal output by the control processing unit 4. When rotating forward, it pushes the movable magnetic core closer to the fixed magnetic core, reducing the air gap of the magnetic core (minimum 0.1mm) and increasing the inductance value. When rotating in reverse, it pulls the movable magnetic core away, increasing the air gap (maximum 2.5mm) and decreasing the inductance value. Finally, it achieves continuous adjustment from 50μH to 350μH with an adjustment accuracy of ±0.01mm. Its core principle is that "the size of the air gap is negatively correlated with the inductance value". It achieves precise matching of the inductance value through micron-level air gap control.

[0026] The adjustable capacitor assembly consists of eight BBY65 silicon varactor diodes connected in parallel to form a varactor diode array. Each varactor diode unit is equipped with an independent digital-to-analog conversion channel (provided by the DAC module of the control processing unit 4). The junction capacitance is adjusted by outputting a continuously adjustable reverse bias voltage of 0-30V. According to the physical characteristics of varactor diodes, when the reverse bias voltage increases, the depletion layer of the PN junction widens and the junction capacitance decreases. When the reverse bias voltage decreases, the depletion layer narrows and the junction capacitance increases. The junction capacitance of a single unit can dynamically change between 2pF and 60pF. After the eight units are connected in parallel, a continuously adjustable capacitance range of 15pF to 450pF is formed, and the junction capacitance of each unit is ≤10pF, so as to avoid parasitic interference to the main power transmission.

[0027] The impedance detection unit employs a high-frequency small-signal injection method to achieve interference-free impedance measurement. The specific workflow is as follows: A sinusoidal detection signal with a frequency of 10kHz and an amplitude of only 0.5% of the normal power signal is superimposed onto the main power path through a high-frequency coupling transformer connected in series. This signal, superimposed with the main power DC signal, is transmitted to the photovoltaic input terminal 1 and the load terminal. A dual-channel sampling circuit synchronously acquires the detection signal voltage values ​​(U1, U2) and current values ​​(I1, I2) before and after the injection point. An on-chip bandpass filter separates the high-frequency detection signal from the main power signal, avoiding interference from the main signal. Based on the complex definition of impedance Z=U / I, the internal resistance of the photovoltaic module (Z_in=U1 / I1) and the equivalent impedance of the load terminal (Z_load=U2 / I2) are derived by calculating the voltage-current amplitude ratio and phase difference of the detection signal. The detection accuracy reaches ±1%, and the data sampling period is 1ms, ensuring real-time performance.

[0028] The matching algorithm processor is integrated into the fast tracking processing core (Cortex-M7 480MHz) of the control processing unit 4. It incorporates a three-layer feedforward neural network structure for machine learning algorithms. The acquisition paths for the four node parameters in the input layer are clearly defined: light intensity is collected by a photosensitive sensor mounted on the device casing; ambient temperature is collected by an NTC temperature sensor attached to the power device; and the photovoltaic module's internal resistance and load impedance are output in real time by the impedance detection unit. This module adopts an "offline training + online fine-tuning" working mode. In the offline stage, a test platform is first built to simulate the full operating conditions of the photovoltaic system, covering all combinations of light intensity (50-1000W / m², step size 50W / m²), ambient temperature (-10℃-60℃, step size 5℃), photovoltaic module internal resistance (5Ω-50Ω, step size 5Ω), and load impedance (10Ω-100Ω, step size 10Ω), generating a total of 100,000 sets of operating condition data. For each set of operating conditions... The optimal matching parameters were measured using the "impedance scanning method"—with fixed illumination, temperature, and load, the core air gap (0.1mm-2.5mm, step size 0.01mm) and varactor diode bias voltage (0-30V, step size 0.1V) were adjusted sequentially. The output power of the photovoltaic module and the transmission efficiency of the power conversion circuit 5 were monitored in real time. The inductor air gap value and capacitor bias voltage at the point of "maximum output power + highest transmission efficiency" were labeled as the "optimal label" for that set of operating conditions. Then, a neural network was trained based on these 100,000 sets of labeled data. The hidden layer of the network used the ReLU activation function, the mean squared error (MSE) was used as the loss function, and the Adam optimizer was selected (learning rate 0.001, 500 iterations). Iterative training was performed using a Mini-Batch (batch size 32) method until the loss function value was ≤0.001 (corresponding to inductor air gap error ≤0.01mm and capacitor bias voltage error ≤0.If the loss value does not decrease for 20 consecutive iterations (1V), the network weight parameters are saved, and an initial mapping relationship between the environment-load parameters and the matching parameters is established. During online operation, the impedance detection unit collects the internal resistance and load impedance data of the photovoltaic module every 1ms. The photosensitive sensor and NTC temperature sensor simultaneously collect the light intensity and ambient temperature. All data are transmitted to the fast tracking processing core via a high-speed bus. The similarity between the real-time collected 4D input parameters and the offline training dataset is calculated (using the Euclidean distance algorithm). If there is an offline working condition with a similarity ≥95%, the optimal parameter corresponding to that working condition is directly called as the initial value. If the similarity <95%, the 30 consecutive sets of real-time collected data are used as the initial value. The data (time window 30ms) consists of small batches of samples. The hidden layer weights are fixed, with only the output layer weights fine-tuned. The mapping model is updated after 10 iterations. Subsequently, the real-time input parameters are normalized to the 0-1 range (using the normalized parameters from offline training). This normalized parameter is then passed from the input layer to the hidden layer. After linear operations of "input × weight + bias" and processing using the ReLU activation function, a hidden layer feature vector is output. This feature vector is then passed to the output layer for linear operations to obtain the normalized output value. Finally, the normalized output value is denormalized to restore the actual physical quantities, namely the optimal inductor air gap command and capacitor bias voltage command. The entire convergence time is ≤3ms, ensuring rapid response under dynamic operating conditions.

[0029] The closed-loop adjustment process of the adaptive impedance matching network is as follows: 1. Data acquisition stage: The impedance detection unit collects the photovoltaic internal resistance and load impedance in real time, while the photosensitive sensor and NTC sensor collect the light intensity and ambient temperature respectively. All parameters are synchronously transmitted to the matching algorithm processor via a high-speed bus; 2. Algorithm calculation stage: The processor calls the machine learning module to deduce the optimal inductor air gap value and capacitor bias voltage value based on the input parameters; 3. Command output stage: The control processing unit 4 converts the air gap command into a pulse drive signal for the stepper motor (including forward and reverse direction and pulse number), and converts the capacitor bias voltage command into 8 independent 0-30V DC bias voltages through a digital-to-analog converter circuit; 4. Execution adjustment stage: The stepper motor drives the threaded push rod according to the pulse signal to precisely adjust the air gap of the adjustable inductor component's magnetic core and the variable capacitance. Each unit of the diode array adjusts its junction capacitance under independent bias voltage, achieving synchronous dynamic adjustment of inductance and capacitance values; 5. Feedback verification stage: After adjustment, the impedance detection unit collects the updated internal resistance and load impedance again, and the matching algorithm processor verifies whether the current inductance and capacitance parameters make the input impedance and load impedance achieve conjugate matching (matching degree ≥98% is optimal). If the optimal matching degree is not achieved, the current matching degree is used as a feedback parameter and input to the machine learning module to fine-tune the output layer weights (learning rate 0.0001, 5 iterations), and re-output the optimized instructions. The above process is repeated until the optimal power transmission is achieved, forming a complete closed-loop adjustment link to ensure that the photovoltaic module always works near the maximum power point, while making the input impedance of the power conversion circuit 5 dynamically matched with the load, effectively improving the overall energy conversion efficiency.

[0030] The soft-switching auxiliary circuit adopts an LLC resonant quasi-resonant structure, specifically composed of a resonant inductor, a resonant capacitor, an auxiliary switching transistor, and a control timing circuit (the resonant cavity is the functional circuit composed of the above components). The resonant inductor has a value of 3.3μH (DCR≤0.1Ω), the resonant capacitor has a value of 22nF (withstanding voltage 1kV), and the auxiliary switching transistor is a SiCMOSFET of the same model as the main switching transistor. An auxiliary switching transistor is connected in parallel across each main switching transistor. The resonant inductor is connected in series to the main power path (the main power path refers to the core energy transmission path of "DC bus after dual-Boost boost, full-bridge inverter module, and intelligent filter network" in power conversion circuit 5). The resonant capacitor is connected in parallel across the main switching transistor, forming a complete resonant circuit together with the main switching transistor and the auxiliary switching transistor. The specific connection logic is as follows: DC bus positive terminal, resonant inductor, main switch of the full-bridge inverter module bridge arm, DC bus negative terminal, resonant capacitor is connected across the drain and source of the main switch, auxiliary switch is connected in parallel with the resonant capacitor and then in parallel with the two ends of the main switch, forming an LLC resonant topology with inductor in series and capacitor in parallel with auxiliary switch.

[0031] It should be clarified that the main switch is derived from the full-bridge inverter module of the power conversion circuit 5. It is the core switch of the bridge arm of this unit (650V CoolMOS is selected, Qg≤20nC). It is mainly responsible for converting the high-voltage DC power after the double boost to AC power. The soft-switching auxiliary circuit reduces switching losses by working in conjunction with the main switch.

[0032] The control timing circuit uses an STM32H745 dual-core MCU as its core. Through the chip's built-in high-speed comparator (with a reference voltage set to 0.1V to ensure zero-crossing detection accuracy), it acquires the drain-source voltage and drain current signals of the main switch in real time. When the drain-source voltage drops to near the reference voltage, it is determined to be a voltage zero-crossing (the core criterion for zero-voltage turn-on, ZVS); when the drain current drops to the current threshold corresponding to the reference voltage, it is determined to be a current zero-crossing (the core criterion for zero-current turn-off, ZCS). The comparator transmits the zero-crossing detection signal to the MCU's advanced timer module. The timer generates precise timing instructions based on preset timing logic (based on the matching relationship between the LLC resonant frequency and the main switch's switching frequency). The MCU's PWM module then outputs a preliminary drive signal, which, after being enhanced by a gate driver chip (IR2110), is transmitted to the gate of the auxiliary switch, achieving complementary conduction between the auxiliary and main switches.

[0033] The specific coordination process is as follows: 30ns before the main switch is about to turn on, the auxiliary switch is turned off. At this time, the resonant inductor and resonant capacitor in the resonant circuit have completed 1 / 4 cycle resonance. The resonant capacitor discharges to a low potential through the resonant inductor, causing the drain-source voltage of the main switch to drop to near zero potential, satisfying the zero-voltage turn-on (ZVS) condition and avoiding turn-on losses caused by voltage and current overlap during turn-on. 50ns before the main switch is about to turn off, the auxiliary switch is turned on, and the resonant circuit begins to absorb the drain current of the main switch, causing the current to smoothly drop to zero. At this time, the main switch is turned off in a zero-current state (ZCS), avoiding turn-off losses caused by sudden current changes during turn-off. In this coordination process, the resonant circuit adjusts the voltage and current waveforms of the main switch in real time through electromagnetic resonance characteristics, fundamentally avoiding the overlap of voltage and current during switching, significantly reducing switching losses, and ensuring that the overall device maintains high efficiency and stability in conversion efficiency across the entire load range.

[0034] The control processing unit 4 is located in the upper control area and adopts the STM32H745 series dual-core processor architecture, including a fast tracking processing core and a system monitoring processing core. The fast tracking processing core is a Cortex-M7 core with a running frequency of 480MHz, which is dedicated to executing the MPPT algorithm that combines the perturbation observation method and the incremental conductance method, and completes the submaximum power point tracking calculation every 10ms. The system monitoring processing core is a Cortex-M4 core with a running frequency of 240MHz, which is responsible for temperature monitoring, fault diagnosis, power grid synchronization, and Modbus communication with the host computer (baud rate 9600bps). The dual cores share data through the on-chip high-speed AXI interconnect bus (transmission rate ≥500Mbps). The on-chip configuration includes 2MB Flash + 1MB RAM (including 192KBTCM zero-wait cache), and is equipped with 16 high-resolution PWM channels (151ps precision), 7 comparators, and 3 12-bit ADCs, supporting synchronous acquisition of 31 external channels to ensure that fast tracking and system safety do not interfere with each other. The computational efficiency is improved compared to the single-core solution, and sub-millisecond dynamic response is achieved.

[0035] The inverter output terminal 6 is located on the right side of the casing and serves as a waterproof AC socket or terminal block. It outputs high-quality sinusoidal AC power to the AC grid or load. An intelligent filtering network connects it to the power conversion circuit 5. This network includes a two-stage π-type LC filter stage and an active filter stage. The LC filter stage uses a 15μH ferrite inductor (saturation current ≥20A) and a 4.7μF FX7R capacitor (450V withstand voltage) to filter fundamental harmonics. The active filter stage uses a high-speed operational amplifier OPA564 to form a dual quadratic band-stop filter, capable of filtering specific harmonics such as the 5th, 7th, and 11th harmonics (150Hz / 210Hz / 330Hz). The intelligent filter network also includes a filter parameter adjustment unit, which uses a programmable digital potentiometer AD5235 (1024-bit resolution, 10kΩ-250kΩ adjustable). The unit acquires the load impedance through the STM32H745's ADC (sampling period 2ms) and automatically adjusts the damping resistor of the LC filter stage and the gain of the active filter stage. This allows the cutoff frequency of the filter circuit to adaptively change between 1.8kHz and 3.2kHz, maintaining a quality factor of 0.707±0.05. Ultimately, this results in a total harmonic distortion (THD) of less than 2% for the output current, meeting the requirements for grid connection and precision load applications.

[0036] The phase change material heat dissipation system is arranged close to the heat dissipation surface of the power devices in the power conversion circuit 5. Its outer shell is extruded from high thermal conductivity aluminum alloy 6063, and the interior contains a 120cm³ phase change material heat storage unit filled with composite phase change material (the base is paraffin with a melting point of 58℃ + 4% CNOs carbon-based nano thermal conductive medium + 1% oleic acid dispersant), with a solid thermal conductivity of [missing information]. liquid The thermal conductivity is significantly improved compared to pure paraffin; the internal structure of the heat storage unit features a 0.5mm thick copper cross-shaped fin structure with a 5mm fin spacing, directly soldered to the heat sinks of the MOSFET and diode, with a contact thermal resistance ≤ This forms an efficient heat conduction path. The phase change material heat dissipation system also includes an active air cooling unit, a liquid level detection device, and a temperature sensor. The active air cooling unit includes a 12V, 0.18A axial flow temperature-controlled fan (airflow 15CFM) and an air duct (outlet air velocity ≥2m / s). The liquid level detection device uses a capacitive sensor, installed on the side wall of the heat storage unit, with a detection accuracy of ±1mm, and is used to monitor the melting degree of the phase change material. The temperature sensor is an NTC type, with a temperature measurement range of -40℃ to 150℃ and an accuracy of ±0.5℃, and is used to monitor the device temperature in real time. When the device temperature exceeds 55℃, the phase change material begins to melt and absorb a large amount of latent heat, keeping the temperature stable for a short time. When the liquid level detection device detects that the phase change material has completely melted, the control processing unit 4 starts the active air cooling unit, blowing cold air directly onto the fin surface through the air duct to achieve rapid heat dissipation. This composite heat dissipation method ensures that the maximum junction temperature of the power device does not exceed 95℃ when the device is running at full power in an ambient temperature of 50℃, significantly improving the reliability and lifespan of the device in high-temperature environments.

[0037] It also features a multi-level overvoltage protection circuit, using a parallel clamping + series disconnect topology for circuit access. The specific access nodes are: photovoltaic input terminal 1 (the bus loop between the two MC4 waterproof connectors and the dual MPPT control module), the DC bus after dual Boost boost (the high-voltage DC core bus inside the power conversion circuit 5), and the DC input side of the full-bridge inverter module. All of these nodes are high-risk areas for overvoltage, enabling full-link overvoltage protection.

[0038] This multi-stage overvoltage protection circuit includes a fast response protection stage and a slow recovery protection stage. The specific configuration is as follows: The fast response protection stage consists of three parallel arrays of SMDJ series bidirectional TVS diodes with different clamping voltages. The three TVS diodes (models SMDJ800A, SMDJ1000A, and SMDJ1200A) are first connected in parallel to form a protection array, and then connected in parallel between the positive and negative terminals of the three key nodes mentioned above. The SMDJ800A is mainly adapted to transient overvoltage scenarios at photovoltaic input terminal 1, while the SMDJ1... The 000A and SMDJ1200A are respectively for higher voltage level overvoltage protection on the DC bus and the input side of the full-bridge inverter module. The three sets work together to achieve transient clamping across the entire voltage range. The Vrwm (reverse working voltage) of this TVS array is 800V, 1000V, and 1200V, respectively. The VC (clamping voltage) is ≤1.2 times Vrwm, the response time is ≤10ns, and the IPP (peak pulse current) is ≥100A. It can absorb transient overvoltage energy such as lightning strikes and power grid surges at the nanosecond level, avoiding overvoltage breakdown of downstream power devices.

[0039] The slow recovery protection stage uses a double-pole double-throw normally open relay (model G6K-2P-Y, contact capacity 250V / 10A), with two sets in total. They are connected in series in the total power circuit of photovoltaic input terminal 1 (between the two MC4 connectors and the dual MPPT control module) and the main power circuit of the DC bus, respectively. The relay coil voltage is 12V, and the action time is ≤10ms. Its double-pole double-throw structure can simultaneously cut off the positive and negative poles of the corresponding circuit to achieve complete electrical isolation. When a continuous overvoltage is detected, it can completely cut off the energy transmission between the input and the subsequent circuit.

[0040] To achieve the linkage logic of the two-level protection, a high-precision voltage sampling circuit (composed of a voltage divider network of 1MΩ / 0.1% high-precision resistors and 1kΩ / 0.1% resistors, with a voltage division ratio of 1:1000) is connected in parallel at each key node. The sampling signal is transmitted to the Cortex-M4 core via the 12-bit ADC built into the STM32H745 to monitor the clamping voltage of the TVS diode in real time. When the clamping voltage of the TVS diode is detected to continuously exceed 1.1 times the rated value (i.e., 880V for SMDJ800A and SMDJ...), the circuit will detect the voltage. When the voltage of a device (1000A corresponds to 1100V, SMDJ1200A corresponds to 1320V) lasts for ≥500μs, it is determined to be a continuous overvoltage fault. The Cortex-M4 core immediately outputs a high-level drive signal to the relay coil, triggering the two sets of relays to disconnect synchronously. After the voltage sampling circuit detects that the voltage of each key node has recovered to below the rated value and stabilized for 3s, the core outputs a low level, and the relays automatically reset, ensuring that the device operates safely and reliably under overvoltage scenarios such as lightning strikes, power grid anomalies, or component failures.

[0041] The workflow is as follows: After the photovoltaic module is connected to the device through the MC4 waterproof connector at photovoltaic input terminal 1, the dual MPPT control module is immediately started. The Cortex-M7 core dynamically switches the MPPT algorithm according to the rate of change of light intensity to initially optimize the DC input power. The impedance detection unit starts simultaneously, and collects the internal resistance of the photovoltaic module and the equivalent impedance of the load terminal in real time through the high-frequency small signal injection method, and transmits the data to the matching algorithm processor. The machine learning algorithm module of the matching algorithm processor combines parameters such as light intensity and ambient temperature to quickly calculate the optimal impedance matching parameters and send instructions to the control processing unit 4. The control processing unit 4 drives the stepper according to the instructions. The motor and digital-to-analog converter circuits adjust the air gap size of the magnetic core air gap adjustment mechanism and the bias voltage of the varactor diode array, respectively, thereby dynamically adjusting the inductance value of the adjustable inductor component and the capacitance value of the adjustable capacitor component to achieve real-time optimal impedance matching between the photovoltaic input and the load. The DC power after impedance matching enters the power conversion circuit 5, where it is first boosted by a dual-boost voltage boost circuit and then converted to AC power by a full-bridge inverter circuit. During this process, the control timing circuit of the soft-switching auxiliary circuit continuously monitors the operating state of the main switch transistor. By precisely controlling the turn-on and turn-off times of the auxiliary switch transistor, the main switch transistor achieves zero-voltage turn-on and zero-current turn-off, significantly reducing [voltage degradation]. Low switching losses; the inverted AC power enters the intelligent filter network. The LC filter stage first filters out fundamental harmonics, and the active filter stage then precisely compensates for specific harmonics. The filter parameter adjustment unit monitors the load impedance characteristics in real time and automatically adjusts the cutoff frequency and quality factor of the filter circuit to ensure the purity of the output AC power. During operation, the temperature sensor monitors the temperature of the power devices in real time. When the temperature exceeds 55°C, the paraffin-based phase change material in the phase change material heat storage unit begins to melt and absorb a large amount of latent heat. The liquid level detection device monitors the melting degree of the phase change material in real time. When the phase change material is detected to be completely melted, the control processing unit 4 starts the temperature control fan of the active air cooling unit. The fan enables forced convection heat transfer. In the event of transient overvoltage such as lightning strikes, the TVS diode array of the multi-stage overvoltage protection circuit responds rapidly within nanoseconds to absorb overvoltage energy. If a sustained overvoltage occurs, the relay protection circuit immediately activates, completely cutting off the input and output circuits to achieve electrical isolation. The fast tracking processing core and the system monitoring processing core of the control processing unit 4 share data in real time through the on-chip high-speed bus. The fast tracking processing core continuously optimizes the MPPT algorithm parameters, while the system monitoring processing core monitors the working status of each module in real time, diagnoses faults in a timely manner, and provides feedback. Finally, the inverter output terminal 6 outputs a high-quality sinusoidal AC power with a total harmonic distortion rate of less than 2%.

[0042] Through the above structural design and coordinated operation, significant technical effects are achieved: the dual MPPT topology and adaptive impedance matching network work together to dynamically adapt to operating conditions, improving the overall energy conversion efficiency and achieving a power density greater than 2.5kW / L, thus solving the problem of low energy conversion efficiency caused by fixed impedance matching in existing technologies; the LLC soft-switching auxiliary circuit enables zero-voltage turn-on and zero-current turn-off of the main switch, reducing switching losses and maintaining efficiency across the entire load range, overcoming the high switching losses caused by hard switching or single topology; the composite phase change material and active air cooling work together to dissipate heat, ensuring that the maximum junction temperature of the power devices is ≤95℃ when operating at full power in a 50℃ environment, thus solving the heat dissipation problem. The single-stage protection circuit addresses the issue of device overheating, extending the device's lifespan. The multi-stage overvoltage protection circuit balances nanosecond-level transient response with complete isolation, enhancing resistance to lightning strikes and grid anomalies, thus overcoming the limitations of single-stage protection in balancing response and isolation. The intelligent adaptive filter network ensures output current THD ≤ 2%, significantly improving grid compatibility and load adaptability, and resolving the problem of severe harmonics caused by fixed filter parameters. The dual-core parallel architecture achieves a response time ≤ 10ms, improving adaptability to complex operating conditions and overcoming the response delay of single-core processors. Overall, it is suitable for efficient and stable operation under a wide range of lighting, temperature, and load variations, possessing extremely high practical value.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the accompanying drawings and the above description. However, any modifications, alterations, or variations made by those skilled in the art without departing from the scope of the present invention, utilizing the disclosed technical content, are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A high-efficiency energy conversion device based on a dual MPPT topology micro photovoltaic inverter, characterized in that: include: The photovoltaic input terminal connects to two independent photovoltaic modules. A dual MPPT control module, including a first MPPT controller and a second MPPT controller, is connected to the two photovoltaic modules respectively through the photovoltaic input terminal. An adaptive impedance matching network, using adjustable inductors and capacitors, combined with a machine learning algorithm module and high-frequency small-signal injection impedance detection, performs wide-range impedance adjustment on the DC input to the photovoltaic input terminal, forming a closed-loop impedance matching link. The power conversion circuit employs a dual Boost converter module and a full-bridge inverter module, integrating an LLC resonant soft-switching auxiliary circuit. The dual Boost converter module contains two independent boost circuits, respectively connected to the first and second MPPT controllers. The power conversion circuit integrates the LLC resonant soft-switching auxiliary circuit. The power conversion circuit achieves zero-voltage turn-on of the main switch through precise timing control. With zero-current shutdown, the DC power after impedance matching and boosting undergoes secondary boosting and inversion processing. An intelligent filtering network adaptively adjusts the filtering parameters of the inverted AC power to suppress harmonics. The control processing unit employs a dual-core processor architecture; its fast tracking processing core acquires the power generation signals from two photovoltaic modules to calculate the rate of change of light intensity. Based on this, it controls the first and second MPPT controllers to dynamically switch MPPT algorithms and outputs PWM control signals to drive the corresponding boost circuit. The system monitoring processing core of the control processing unit performs system monitoring, and the two cores share data via a high-speed bus. The DC power output from the photovoltaic modules is input through the photovoltaic input terminal, sequentially passing through the first and second MPPT controllers' maximum power point tracking, the adaptive impedance matching network's impedance matching, the boost circuit's boosting, the power conversion circuit's secondary boosting and inversion processing, and the intelligent filtering network's harmonic suppression, before outputting high-quality AC power through the inverter output terminal.

2. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: The first MPPT controller and the second MPPT controller are each configured with an independent MCU and an independent Boost converter module. Each MPPT is equipped with a high-precision ADC. The MPPT algorithm includes the perturbation-observation method and the incremental conductance method. The irradiance variation rate of the photovoltaic module output power is calculated by multiple consecutive samplings. The perturbation-observation method and the incremental conductance method are automatically switched according to the magnitude of the irradiance variation rate. When switching, the latest iteration result of the previous algorithm is retained as the initial value. After the algorithm converges, it outputs a PWM control signal with an adjustable duty cycle to drive the switching transistor of the corresponding Boost converter module.

3. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: The adaptive impedance matching network includes: an adjustable inductor component, employing a ferrite core structure, with the core air gap adjusted via a core air gap adjustment mechanism to achieve wide-range continuous adjustment; an adjustable capacitor component, composed of a varactor diode array, with wide-range continuous adjustment achieved through reverse bias voltage adjustment; an impedance detection unit, employing a high-frequency small-signal injection method to simultaneously acquire voltage and current signals before and after the injection point, separating the high-frequency detection signal from the main power signal to achieve interference-free impedance detection; and a matching algorithm processor, integrating a machine learning algorithm module, which, based on light intensity, ambient temperature, photovoltaic module internal resistance, and load impedance parameters, outputs optimal matching parameters using an offline training + online fine-tuning mode to achieve rapid convergence.

4. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: The LLC resonant soft-switching auxiliary circuit includes a resonant inductor, a resonant capacitor, an auxiliary switch, and a control timing circuit. The resonant inductor is connected in series to the main power path, and the resonant capacitor and the auxiliary switch are connected in parallel and then connected across the drain and source of the main switch. The control timing circuit captures the voltage zero-crossing point and current zero-crossing point of the main switch through a detection element, generates timing commands, and controls the complementary conduction of the auxiliary switch and the main switch to achieve zero-voltage turn-on and zero-current turn-off of the main switch.

5. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: The intelligent filtering network includes a multi-stage filtering circuit and a filtering parameter adjustment unit. The multi-stage filtering circuit includes an LC filtering stage and an active filtering stage. The LC filtering stage is composed of inductors and capacitors, and the active filtering stage is composed of operational amplifiers forming a band-stop filtering structure to compensate for specific harmonics. The filtering parameter adjustment unit uses programmable adjustment elements to automatically adjust the damping parameters and gain of the filtering circuit by collecting load impedance characteristics, thereby achieving adaptive matching of filtering parameters.

6. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: The dual-core processor architecture of the control processing unit includes a fast tracking processing core and a system monitoring processing core, which are dedicated to performing MPPT algorithm calculations and system monitoring functions, respectively. Dual-core data sharing is achieved through a high-speed bus. The system monitoring functions include temperature monitoring, fault diagnosis, power grid synchronization, and communication interaction.

7. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: It also includes a phase change material heat dissipation system, which comprises: a phase change material heat storage unit filled with paraffin-based composite phase change material and an internal fin structure to enhance heat transfer efficiency; a liquid level detection device for monitoring the melting degree of the phase change material; and an active air cooling unit, including a temperature-controlled fan and air ducts. When the temperature of the power device reaches a set threshold, the phase change material begins to melt. After the liquid level detection device detects that the phase change material has completely melted, the control processing unit starts the active air cooling unit.

8. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to claim 1, characterized in that: It also includes a multi-stage overvoltage protection circuit, comprising: a fast-response protection stage, consisting of multiple TVS diode arrays with different clamping voltages connected in parallel, adapted to the photovoltaic input terminal, DC bus, and full-bridge inverter module input side respectively, to absorb transient overvoltage energy; a slow-recovery protection stage, including a relay protection circuit, which consists of multiple sets of double-pole double-throw normally open relays connected in series in the critical power circuits to achieve complete electrical isolation; and a voltage sampling circuit, consisting of a voltage divider network composed of high-precision resistors, to monitor the voltage of each critical node in real time. When a continuous overvoltage is detected, the relay protection circuit is triggered to disconnect, and it automatically resets after the voltage stabilizes.

9. The high-efficiency energy conversion device of a micro photovoltaic inverter based on a dual MPPT topology according to any one of claims 1-8, characterized in that: Includes the following steps: The photovoltaic modules are connected to the device via waterproof connectors. The first and second MPPT controllers are activated, and a dynamic switching algorithm based on the rate of change of sunlight is used to output PWM signals to drive the Boost circuit. The impedance detection unit collects the internal resistance of the photovoltaic modules and the load impedance, and combines it with data on light intensity and ambient temperature. The machine learning algorithm module outputs the optimal matching parameter instructions to adjust the adjustable inductor and adjustable capacitor components. The DC power is input to the full-bridge inverter module after passing through the dual Boost modules. The soft-switching auxiliary circuit achieves zero-voltage turn-on and zero-current turn-off of the main switch tube through precise timing control. The AC power after inversion is subjected to harmonic suppression by an intelligent filter network, and the filter parameters are adaptively adjusted according to the load impedance. During operation, if a phase change material heat dissipation system is configured, heat dissipation is activated based on the device temperature and the melting state of the phase change material. If a multi-level overvoltage protection circuit is configured, the voltage is monitored in real time to achieve transient clamping and continuous overvoltage cutoff. The dual-core processor architecture of the control processing unit works together to quickly track and process the real-time optimization algorithm parameters of the processing core, and the system monitoring processing core monitors the system status, ultimately outputting a low-harmonic-distortion sinusoidal AC power.

Images