Photovoltaic module intelligent control method based on TinyML

Abstract

The invention discloses a TinyML-based photovoltaic module intelligent control method, and particularly relates to the technical field of photovoltaic system control, and the method comprises the steps: carrying out the pre-training of a TinyML frame under the model quantification control based on original collection data, carrying out the generalization capability evaluation of a model through k-fold cross validation according to the spatial distribution condition of the surface temperature of a photovoltaic module array, and carrying out the prediction of the surface temperature of the photovoltaic module array; the method comprises the following steps of: acquiring an illumination intensity threshold value and a temperature threshold value by using an MCU (Microprogrammed Control Unit), optimizing the model through hyper-parameter search to obtain an illumination intensity threshold value and a temperature threshold value, reading environmental data when the photovoltaic module is used by the MCU, adjusting the voltage of the photovoltaic module through a dynamic control strategy according to different environmental data, and carrying out fault diagnosis on the photovoltaic module through time sequence data of a sensor and a TinyML (Markup Language) algorithm. Through the adaptive MPPT algorithm, the energy conversion efficiency of the photovoltaic system is significantly improved, and the stability and reliability of the system are improved by dynamically adjusting the control strategy according to the environmental change.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic system control, and in particular to a photovoltaic component intelligent control method based on TinyML. Background Art

[0002] As the global demand for renewable energy increases, photovoltaic systems have been widely used due to their clean and renewable characteristics. One of the core components of a photovoltaic system is the photovoltaic module controller, whose main functions include maximum power point tracking (MPPT), inverter control, data monitoring and remote management. Traditional photovoltaic module controllers usually use fixed algorithms and hardware designs. Although they can meet basic control requirements, they are insufficient in terms of intelligence and adaptability.

[0003] There are many problems with traditional photovoltaic module controllers. Most traditional controllers use fixed MPPT algorithms, such as the perturbation and observation method (P&O) and the incremental conductance method (INC). These algorithms are prone to misjudgment when there are large changes in light and temperature, resulting in reduced system efficiency; existing controllers are unable to dynamically adjust control strategies according to environmental changes, resulting in unstable performance in complex environments; traditional controllers have weak data processing capabilities, making it difficult to achieve real-time data analysis and intelligent decision-making; traditional controllers consume high power when processing complex tasks, affecting the overall energy efficiency of the system. These problems have affected the efficiency and safety of photovoltaic modules. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to propose a photovoltaic component intelligent control method based on TinyML, which is used to solve the problem in the prior art that photovoltaic components cannot perform real-time data analysis and intelligent control.

[0005] The technical solution adopted by the present invention to solve the technical problem is a photovoltaic component intelligent control method based on TinyML, which is characterized by comprising the steps of:

[0006] S1. Pre-train the TinyML framework under model quantization control based on the original collected data. According to the spatial distribution of the surface temperature of the photovoltaic module array, the generalization ability of the model is evaluated through k-fold cross validation, and the model is optimized through hyperparameter search to obtain the light intensity threshold and temperature threshold;

[0007] S2. Deploy the trained TinyML to the MCU, and use the MCU to read the environmental data of the photovoltaic module when it is in use. When the environmental data is higher than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method; when the environmental data is lower than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the conductance increment method; when the environmental data is higher than the temperature threshold, the MCU adjusts the voltage of the photovoltaic module through the TinyML algorithm; when the environmental data is lower than the temperature threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method and the conductance increment method at the same time;

[0008] S3. When the MCU adjusts the voltage of the PV module, the MCU uses the sensor's timing data and the TinyML algorithm to diagnose the fault of the PV module. When a PV module fails, the MCU activates the alarm system and starts the edge bypass.

[0009] Furthermore, before the S1 step, TensorFlow is used to perform pruning operations on TinyML to remove redundant weights and connections.

[0010] Furthermore, when the environmental data exceeds a temperature threshold, the voltage and current of the photovoltaic component are adjusted through the TinyML to perform temperature compensation for the photovoltaic component.

[0011] Furthermore, when the environmental data is lower than the light intensity, the photovoltaic assembly is controlled by the MCU to switch to a low power consumption mode at the microampere level.

[0012] Furthermore, the low power consumption mode switches the capacitor to a micro solid-state capacitor through an LDSW communication unit.

[0013] Furthermore, the photovoltaic modules are arranged in a grid-like form, and when environmental data of one of the photovoltaic modules changes, the MPPT algorithm is selected through the TinyML.

[0014] Furthermore, the color change of each voxel in different directions is captured by the spherical harmonic function, and the time series data includes statistical features, frequency domain features and time domain features.

[0015] Furthermore, the perturbation observation method continuously applies perturbations to the working voltage of the photovoltaic module, and adjusts the voltage of the photovoltaic module accordingly according to the output power of the photovoltaic module at different voltages.

[0016] Furthermore, the conductance increment method monitors the conductance change of the photovoltaic module and obtains the maximum power point according to the voltage-current characteristic curve of the photovoltaic module, which is used to adjust the voltage of the photovoltaic module.

[0017] Furthermore, the TinyML algorithm adjusts the voltage of the photovoltaic module according to the nonlinear relationship between the environmental data and the output power.

[0018] Compared with the prior art, the present invention has at least the following beneficial effects:

[0019] (1) The present invention discloses a method and system for intelligent control of photovoltaic modules based on TinyML. The controller can dynamically adjust the control strategy according to environmental changes and significantly improve the energy conversion efficiency of the photovoltaic system through an adaptive MPPT algorithm. This improves the stability and reliability of the system.

[0020] (2) The low power consumption mode design prolongs the service life of the controller and reduces maintenance costs;

[0021] (3) Adopt intelligent fault detection and diagnosis functions to effectively prevent and reduce system failures and improve system safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of a photovoltaic module intelligent control method based on TinyML;

[0023] Figure 2 This is a schematic diagram of a photovoltaic component intelligent control system module based on TinyML. DETAILED DESCRIPTION

[0024] The following are specific embodiments of the present invention and the accompanying drawings to further describe the technical solution of the present invention, but the present invention is not limited to these embodiments.

[0025] Embodiment 1

[0026] In order to enable real-time data analysis and intelligent control of photovoltaic modules, such as Figure 1 As shown, the present invention provides a photovoltaic assembly intelligent control method based on TinyML, characterized in that it includes the steps of:

[0027] S1. Pre-train the TinyML framework under model quantization control based on the original collected data. According to the spatial distribution of the surface temperature of the photovoltaic module array, the generalization ability of the model is evaluated through k-fold cross validation, and the model is optimized through hyperparameter search to obtain the light intensity threshold and temperature threshold;

[0028] S2. Deploy the trained TinyML to the MCU, and use the MCU to read the environmental data of the photovoltaic module when it is in use. When the environmental data is higher than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method; when the environmental data is lower than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the conductance increment method; when the environmental data is higher than the temperature threshold, the MCU adjusts the voltage of the photovoltaic module through the TinyML algorithm; when the environmental data is lower than the temperature threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method and the conductance increment method at the same time;

[0029] S3. When the MCU adjusts the voltage of the PV module, the MCU uses the sensor's timing data and the TinyML algorithm to diagnose the fault of the PV module. When a PV module fails, the MCU activates the alarm system and starts the edge bypass.

[0030] The microprocessor reads raw data such as current, voltage, temperature, and gravity acceleration through the ADC (analog-to-digital converter), and collects environmental parameter data such as wind speed and humidity. Use low-pass filtering and mean filtering to remove noise from these data, and use interpolation or average method to fill in missing values ​​in the environmental and operating data collected by the sensor to ensure that the sensor data used for analysis and control is complete and continuous, and identify and eliminate obviously erroneous measurement values, so as to clean the collected raw data. The cleaned data is scaled to the range of [0,1], and the data is converted to a distribution with a mean of 0 and a standard deviation of 1. After verification, cleaning and normalization, the data is used as the input vector of the MPPT algorithm model to improve the accuracy of the algorithm.

[0031] Based on the original collected data, the TinyML framework is pre-trained under the control of model quantization, and the time series data is processed. The time series data includes statistical features, frequency domain features and time domain features. Among them, the statistical quantities such as mean, variance, maximum value and minimum value are calculated, the frequency components of the statistics are extracted through fast Fourier transform, and the power, energy, zero-crossing rate and other features are calculated. Statistics can summarize the overall characteristics of time series data and help the model better understand the data distribution. Frequency domain features help capture the periodicity and oscillation patterns in the data. These features can capture the time dependence and dynamic change patterns of the data.

[0032] Furthermore, the photovoltaic modules are arranged in a grid-like form. According to the spatial distribution of the surface temperature of the photovoltaic module array, combined with the data of multiple sensors, the Kalman filter, fuzzy logic and other technologies are used to improve the accuracy and reliability of the data. A lightweight machine learning model is trained based on the extracted features so that it can run on a resource-constrained microcontroller. The model is simplified first, TensorFlow prunes TinyML to remove redundant weights and connections, and the TensorFlow Model Optimization Toolkit is used to prune the model to remove unimportant weights and connections to reduce the size of the model. The floating-point weights are converted to low-precision fixed-point numbers (such as 8-bit integers) to reduce the model size and computational complexity, and unimportant weights or connections are removed to further compress the model. In the simplified model, the data is divided into a training set and a validation set. The generalization ability of the model is evaluated through k-fold cross-validation. The data is divided multiple times for training and validation to obtain a more stable and reliable performance estimate. The model is optimized through hyperparameter search to find the best hyperparameter combination to improve the performance of the model, so as to obtain the light intensity threshold and temperature threshold based on the trained TinyML framework model.

[0033] Furthermore, the trained TinyML is deployed in the MCU. Through component integration and platform environmental sensors, the controller edge collects the input voltage, current, temperature and the environmental data of the light intensity parameters collected by the system in real time. When the environmental data of one of the photovoltaic modules changes, the MPPT algorithm is selected through TinyML. When the environmental data is higher than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method, continuously applies perturbations to the working voltage of the photovoltaic module, and adjusts the voltage of the photovoltaic module according to the output power of the photovoltaic module at different voltages, ensuring that the system can respond quickly and maintain efficient energy conversion under high light conditions. When the environmental data is lower than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the conductance increment method, monitors the conductivity changes of the photovoltaic module, obtains the maximum power point according to the voltage-current characteristic curve of the photovoltaic module, compares the obtained conductance with the current-voltage ratio, and adjusts the working voltage of the photovoltaic array accordingly according to different ratios until it finds and stabilizes at the maximum power point, which is used to adjust the voltage of the photovoltaic module. Improve the stability and accuracy of the system to ensure that the maximum power point can be accurately tracked even under low light conditions; when the environmental data is higher than the temperature threshold, the MCU adjusts the PV module voltage through the TinyML algorithm, and adjusts the PV module voltage according to the nonlinear relationship between the environmental data and the output power, effectively handling complex nonlinear relationships, and ensuring that efficient energy conversion can be maintained even under high temperature conditions; when the environmental data is lower than the temperature threshold, the MCU adjusts the PV module voltage through the perturbation observation method and the conductance increment method at the same time, that is, using a hybrid MPPT strategy for adjustment. This strategy combines the advantages of the two algorithms, alternately executing the perturbation observation method and the conductance increment method steps, and each time determining the next adjustment direction based on the result of the previous step. Once the ambient temperature returns to the normal range, the MCU can switch back to a single MPPT method, namely the perturbation observation method or the conductance increment method. The simplified control logic is intended to improve the stability and response speed of the system under low temperature conditions. With the help of the trained TinyML algorithm model, the maximum power point tracking strategy is dynamically optimized according to the real-time changes in light intensity and ambient temperature.

[0034] Furthermore, the system monitors the operating status of photovoltaic modules in real time, including parameters such as voltage, current, and temperature. Through wireless communication technologies such as LDSW or LoRa, the data is transmitted to the cloud for analysis and management. By using the sensor's time series data, TinyML algorithm, and machine learning data analysis functions on the cloud platform, the controller can perform fault detection and diagnosis tasks. When the MCU adjusts the voltage of the photovoltaic module, the MCU uses the sensor's time series data and TinyML algorithm to diagnose the photovoltaic module fault. These fault detection technologies are based on a comprehensive analysis of current, voltage, temperature, power, and environmental time series data, and combined with machine learning functions, specific thresholds and operating trend analysis are set through different types of components. When a photovoltaic module fails, the system will activate the alarm through the MCU and automatically start the edge bypass or disconnection control mechanism according to the fault type. The microprocessor will cut off the power supply and prevent damage through relays or protection circuits. At the same time, the cloud platform will provide diagnostic suggestions and guidance services such as cleaning, maintenance, and replacement to help users quickly solve problems, effectively prevent and reduce system failures, and improve system safety.

[0035] Furthermore, in order to ensure that the photovoltaic modules always operate in the optimal state, the voltage value of the maximum power point is accurately predicted by fine-tuning the step size, integrating the temperature compensation mechanism and real-time data fusion and other cutting-edge technologies. TinyML adjusts the voltage and current of the photovoltaic modules to compensate for the temperature of the photovoltaic modules. By adopting low-power microprocessors, sensors and transmission communication units, and optimizing the hardware design, the overall energy consumption is effectively reduced. In low-light or nighttime environments, that is, when the environmental data is lower than the light intensity, the output voltage drops below the working threshold of the controller, and the MCU controls the photovoltaic modules to automatically switch to the low-power mode at the micro-ampere level. Through the ultra-low power consumption technology of the LDSW communication unit, only micro solid-state capacitors are required to ensure that the monitoring function can run continuously for more than 48 hours, thereby extending the service life of the controller and reducing maintenance costs.

[0036] Embodiment 2

[0037] In a second aspect, the present invention provides a photovoltaic module intelligent control system based on TinyML, such as Figure 2 As shown, the system performs intelligent control operations on corresponding photovoltaic components according to the modules.

[0038] The system includes:

[0039] MCU, used for data collection. Data collection involves reading data from various sensors and performing data processing. Data processing includes executing MPPT algorithms and fault detection, and controlling the output to adjust the voltage and current of the MPPT controller;

[0040] Sensors are used to collect various data. The sensors include: current sensors, which are used to measure the output current of photovoltaic modules; voltage sensors, which are used to measure the module input and bus output voltages; temperature sensors, which are used to measure the module surface and ambient temperature; acceleration sensors, which are used to measure the module installation inclination angle and vibration conditions; system environment sensors, which are used to measure environmental parameters such as wind speed, air pressure, humidity, and ambient temperature;

[0041] Communication module, optional LDSW narrowband trunking or PLC power line carrier, used for internal communication in the system to achieve wireless data transmission;

[0042] Power management module, used to supply power, ensure stable operation of MCU and sensors, and manage the micro-battery charging and discharging process;

[0043] Storage module, internal FLASH is used to store sensor data, log files, configuration parameters, etc. The storage capacity can be increased by expanding EEPROM.

[0044] The microprocessor selected is the ARM Cortex-M4 microprocessor, which has the characteristics of low power consumption and high performance. The microprocessor, sensor, voltage regulating inverter circuit, communication module, etc. are integrated into a compact package and installed in the junction box of the photovoltaic module, thereby realizing real-time data monitoring and remote management functions, making it convenient for users to understand the system status at any time and improve operation and maintenance efficiency.

[0045] In summary, a method for intelligent control of photovoltaic components based on TinyML of the present invention pre-trains the TinyML framework under model quantization control based on the original collected data, evaluates the generalization ability of the model through k-fold cross validation according to the spatial distribution of the surface temperature of the photovoltaic component array, and optimizes the model through hyperparameter search to obtain the light intensity threshold and the temperature threshold, and deploys the trained TinyML to the MCU, and reads the environmental data of the photovoltaic component when it is in use through the MCU. When the environmental data is higher than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic component through the perturbation observation method; when the environmental data is lower than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic component through the conductance increment method; when the environmental data is higher than the temperature threshold, the MCU adjusts the voltage of the photovoltaic component through the TinyML algorithm; when the environmental data is lower than the temperature threshold, the MCU adjusts the voltage of the photovoltaic component through the perturbation observation method and the conductance increment method at the same time, when the MCU adjusts the voltage of the photovoltaic component, the MCU performs fault diagnosis on the photovoltaic component through the timing data of the sensor and the TinyML algorithm, and when the photovoltaic component fails, the alarm system is activated by the MCU and the edge bypass is started. The adaptive MPPT algorithm significantly improves the energy conversion efficiency of the photovoltaic system. The controller can dynamically adjust the control strategy according to environmental changes to improve the stability and reliability of the system. In low-light or nighttime environments, it automatically uses low-power mode to extend the life of the controller and reduce maintenance costs. Operators can use real-time data monitoring and remote management functions to facilitate users to understand the system status at any time and improve operation and maintenance efficiency. It can also perform intelligent fault detection and diagnosis to effectively prevent and reduce system failures and improve system safety.

[0046] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0047] In addition, in the present invention, descriptions such as "first", "second", "one", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0048] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0049] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

Claims

1. A photovoltaic module intelligent control method based on TinyML, characterized in that: Includes steps: S1. Pre-train the TinyML framework under model quantization control based on the original collected data. According to the spatial distribution of the surface temperature of the photovoltaic module array, the generalization ability of the model is evaluated through k-fold cross validation, and the model is optimized through hyperparameter search to obtain the light intensity threshold and temperature threshold; S2. Deploy the trained TinyML to the MCU, and use the MCU to read the environmental data of the photovoltaic module when it is in use. When the environmental data is higher than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method; when the environmental data is lower than the light intensity threshold, the MCU adjusts the voltage of the photovoltaic module through the conductance increment method; when the environmental data is higher than the temperature threshold, the MCU adjusts the voltage of the photovoltaic module through the TinyML algorithm; when the environmental data is lower than the temperature threshold, the MCU adjusts the voltage of the photovoltaic module through the perturbation observation method and the conductance increment method at the same time; S3. When the MCU adjusts the voltage of the PV module, the MCU uses the sensor's timing data and the TinyML algorithm to diagnose the fault of the PV module. When a PV module fails, the MCU activates the alarm system and starts the edge bypass.

2. According to claim 1, a photovoltaic module intelligent control method based on TinyML is characterized in that: Before the S1 step, TensorFlow is used to perform pruning operations on TinyML to remove redundant weights and connections.

3. According to claim 1, a photovoltaic assembly intelligent control method based on TinyML is characterized in that: When the environmental data exceeds a temperature threshold, the voltage and current of the photovoltaic component are adjusted through the TinyML to perform temperature compensation for the photovoltaic component.

4. The photovoltaic module intelligent control method based on TinyML according to claim 1, characterized in that: When the environmental data is lower than the light intensity, the photovoltaic assembly is controlled by the MCU to switch to a low power consumption mode at the microampere level.

5. The photovoltaic module intelligent control method based on TinyML according to claim 4, characterized in that: The low power consumption mode switches the capacitor to a micro solid-state capacitor through the LDSW communication unit.

6. The photovoltaic module intelligent control method based on TinyML according to claim 1, characterized in that: The photovoltaic modules are arranged in a grid form, and when the environmental data of one of the photovoltaic modules changes, the MPPT algorithm is selected through the TinyML.

7. The photovoltaic module intelligent control method based on TinyML according to claim 1, characterized in that: The time series data includes statistical features, frequency domain features and time domain features.

8. The photovoltaic module intelligent control method based on TinyML according to claim 1, characterized in that: The perturbation observation method continuously applies perturbations to the working voltage of the photovoltaic module, and adjusts the voltage of the photovoltaic module accordingly according to the output power of the photovoltaic module at different voltages.

9. The photovoltaic module intelligent control method based on TinyML according to claim 1, characterized in that: The conductance increment method monitors the conductance change of the photovoltaic module and obtains the maximum power point according to the voltage-current characteristic curve of the photovoltaic module, which is used to adjust the voltage of the photovoltaic module.

10. The photovoltaic module intelligent control method based on TinyML according to claim 1, characterized in that: The TinyML algorithm adjusts the voltage of the photovoltaic module according to the nonlinear relationship between the environmental data and the output power.

Images