Novel SVG device capable of automatically following new energy power generation state

Abstract

The invention discloses a novel SVG device capable of automatically following a new energy power generation state. The novel SVG device comprises a new energy power generation state monitoring module; the multi-source power prediction module is used for predicting new energy power generation power from 10 seconds to 1 hour in the future according to the monitoring data; the dynamic compensation control module is used for dynamically adjusting the reactive current output by the SVG according to the prediction result; the composite heat dissipation module is used for automatically adjusting heat dissipation power according to the temperature threshold value; the fault self-diagnosis module is used for monitoring the equipment state in real time and early warning and positioning faults; through second-level response to new energy power generation fluctuation, the power grid voltage fluctuation rate is reduced from + / -5% to + / -1.2%, the power factor is increased to 0.99 or above, and the power grid accident rate is reduced; and the self-adaptive heat dissipation system reduces the equipment failure rate, reduces the operation and maintenance cost, supports high-altitude, extremely cold and high-temperature deserts and other extreme environments, covers the requirements of new energy power stations, and has huge market potential.

Description

Technical Field

[0001] This invention relates to the field of new energy power generation and reactive power compensation technology, specifically a novel SVG device that automatically follows the state of new energy power generation. Background Technology

[0002] By the end of 2024, the global installed capacity of new energy power generation exceeded 4.5 billion kilowatts, with China accounting for over 32%, of which wind and solar power accounted for 28%. The International Energy Agency (IEA) predicts that by 2030, new energy power generation will account for 40% of global electricity supply, but its intermittent and random characteristics pose a severe challenge to grid stability: the power fluctuation rate of onshore wind power can reach 30% within 10 minutes, and the power fluctuation rate of offshore wind power can reach 50% due to more drastic changes in wind speed; while solar power generation is affected by day and night, seasons, and cloud cover, with daytime power fluctuation rate reaching 50%, requiring energy storage systems to maintain output at night; at the same time, the charging and discharging power fluctuation of lithium-ion batteries in energy storage systems causes grid frequency deviations, requiring dynamic reactive power compensation to maintain voltage stability.

[0003] Traditional SVG devices employ a fixed control strategy, which has the following core drawbacks: Response lag: The response time of a conventional SVG is about 20-50ms, which cannot match the second-level fluctuations of new energy power generation (such as the 10ms-level power drop in photovoltaic power generation due to cloud cover), resulting in excessive grid voltage fluctuation rate. The compensation strategy is too simple: it operates with preset parameters and cannot dynamically adjust the compensation amount according to the real-time status of new energy power generation (such as wind speed, light intensity, and energy storage charging and discharging power), resulting in low compensation accuracy (deviation > 10%). Poor environmental adaptability: In high-altitude (5000m), extremely cold (-40℃), and high-temperature desert (60℃) environments, the equipment's heat dissipation and insulation performance decrease, resulting in an annual failure rate as high as 20%. Weak fault diagnosis capability: Lacking a self-diagnostic module, equipment failures require manual troubleshooting, with a mean time to repair (MTTR) of up to 8 hours, affecting the continuous operation of new energy power plants.

[0004] In recent years, some companies have improved SVG performance by optimizing control algorithms or adding sensors, and adopted fast-response control algorithms, reducing the response time to 5ms. However, they have not established a linkage mechanism between new energy power generation prediction and compensation, and the deviation between compensation and actual demand still reaches 8%. They have integrated wind speed prediction modules, but these are only applicable to wind power scenarios and cannot be compatible with multiple energy forms such as photovoltaics and energy storage. They have adopted liquid cooling systems, but have not combined them with air cooling to form a composite heat dissipation system, resulting in a 30% decrease in heat dissipation efficiency in extreme environments. To address this, a novel SVG device that automatically follows the state of new energy power generation is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a novel SVG device that automatically follows the state of new energy power generation, so as to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a novel SVG device that automatically follows the status of new energy power generation, including a new energy power generation status monitoring module for real-time acquisition of wind speed, light intensity, power generation, and energy storage charging and discharging power data; The multi-source power prediction module includes an LSTM neural network prediction unit, a CNN feature extraction unit, and an attention mechanism fusion unit, which is used to predict the power generation of new energy sources in the next 10 seconds to 1 hour based on monitoring data. The dynamic compensation control module includes a fuzzy PID control unit, an MPC optimization unit, and a DRL learning unit, which are used to dynamically adjust the reactive current output of the SVG based on the prediction results. The composite heat dissipation module includes a liquid cooling subsystem, an air cooling subsystem, and a phase change material heat dissipation subsystem, which is used to automatically adjust the heat dissipation power according to the temperature threshold. The fault self-diagnosis module includes an expert system reasoning unit and a machine learning recognition unit, which are used to monitor the equipment status in real time and provide early warnings to locate faults. The communication and protection module supports multiple communication protocols and protection functions to ensure the safe and stable operation of the equipment.

[0007] According to the above technical solution, the new energy power generation status monitoring module includes an anemometer, a light sensor, a current / voltage transformer, an energy storage charging and discharging monitoring module, a vibration sensor, and an acoustic sensor.

[0008] According to the above technical solution, the LSTM neural network prediction unit of the multi-source power prediction module adopts the Adam optimizer.

[0009] According to the above technical solution, the fuzzy PID control unit of the dynamic compensation control module dynamically adjusts K based on the deviation ΔP between the predicted power and the actual power and the rate of change of the deviation ΔP′. p K i K d The parameters and formula are as follows: IQ(t) = K p •ΔP(t)+K i •∫0 t ΔP(τ)dτ+K d •ΔP′(t) Among them, K p K i K dThe parameters are adjusted in real time by the fuzzy inference engine. The input variable range is: ΔP∈[-50MW,50MW], ΔP′∈[-10MW / s, 10MW / s].

[0010] According to the above technical solution, the liquid cooling subsystem of the composite heat dissipation module includes a water cooling plate, an internal circulation pipe, a gas-water separator, and a buffer water tank.

[0011] According to the above technical solution, the expert system reasoning unit of the fault self-diagnosis module includes a rule base and a reasoning engine. The rule base covers the diagnostic rules for common equipment faults. The machine learning recognition unit adopts the random forest algorithm. The training data includes historical fault data and normal operation data. The feature variables include temperature, current, voltage, vibration, and acoustics.

[0012] According to the above technical solution, the communication and protection module supports IEC61850-90-5, IEC61850-90-7 and 5G communication protocols, and has overcurrent, overvoltage, undervoltage, overtemperature and insulation fault protection functions.

[0013] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: This invention reduces grid voltage fluctuation rate from ±5% to ±1.2% and improves the power factor to over 0.99 by responding to new energy power generation fluctuations within seconds, thereby reducing grid accident rate; it enhances grid stability; the adaptive cooling system reduces equipment failure rate and lowers operation and maintenance costs; it supports extreme environments such as high altitudes, extreme cold, and high-temperature deserts, covering the needs of new energy power plants, and has huge market potential; dynamic compensation strategies reduce reactive power loss and improve power generation efficiency; the fault self-diagnosis module shortens the average repair time and improves equipment availability; it is compatible with multiple energy forms, supporting various new energy forms such as wind power, photovoltaics, and energy storage, and allows for rapid switching of control strategies through parameter configuration modules; it supports multiple communication protocols and protection functions to ensure safe and stable equipment operation. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a system architecture diagram of the SVG device of the present invention; Figure 2 This is a flowchart of the dynamic reactive power compensation algorithm of the present invention; Detailed Implementation

[0015] 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 some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1-2 The present invention provides a technical solution: a novel SVG device that automatically follows the status of new energy power generation, including a new energy power generation status monitoring module for real-time collection of wind speed, light intensity, power generation, and energy storage charging and discharging power data; The multi-source power prediction module includes an LSTM neural network prediction unit, a CNN feature extraction unit, and an attention mechanism fusion unit, which is used to predict the power generation of new energy sources in the next 10 seconds to 1 hour based on monitoring data. The dynamic compensation control module includes a fuzzy PID control unit, an MPC optimization unit, and a DRL learning unit, which are used to dynamically adjust the reactive current output of the SVG based on the prediction results. The composite heat dissipation module includes a liquid cooling subsystem, an air cooling subsystem, and a phase change material heat dissipation subsystem, which is used to automatically adjust the heat dissipation power according to the temperature threshold. The fault self-diagnosis module includes an expert system reasoning unit and a machine learning recognition unit, which are used to monitor the equipment status in real time and provide early warnings to locate faults. The communication and protection module supports multiple communication protocols and protection functions to ensure the safe and stable operation of the equipment.

[0017] Fault self-diagnosis algorithm: Composite heat dissipation system design The composite heat dissipation system combines liquid cooling, air cooling, and phase change materials to achieve efficient heat dissipation. The system automatically adjusts the heat dissipation mode based on temperature thresholds. When the temperature is ≤40℃, only the liquid cooling system operates, and the water-cooled plate temperature is maintained at 35-40℃; When the temperature exceeds 40℃, the air-cooling system will activate, and the variable frequency fan speed will automatically adjust according to the temperature gradient (500-3000 rpm). The airflow adjustment range is 0-500 m³ / h. 3 / h; When the temperature is ≥55℃, the phase change material is triggered to dissipate heat. The paraffin-based PCM absorbs the instantaneous heat to prevent the temperature from rising suddenly. When the temperature is ≥60℃, an alarm will sound and the machine will shut down, and the fault self-diagnosis module will be activated to investigate the cause of the fault.

[0018] Specifically, the new energy power generation status monitoring module includes an anemometer, a light sensor, a current / voltage transformer, an energy storage charging and discharging monitoring module, a vibration sensor, and an acoustic sensor; this device adopts a modular design, and the hardware architecture includes: a monitoring subsystem: Anemometer: Range 0-60 m / s, accuracy ±0.1 m / s, installed on top of the wind farm tower; employs an ultrasonic anemometer (model: WindMaster 3D), calculating three-dimensional wind speed by measuring the time difference of ultrasonic wave propagation in air. Installed on top of the wind farm tower (80 meters above ground), equipped with an automatic heating device to prevent icing. Data acquisition frequency 1 kHz, transmitted to the main controller via RS485 bus. The ultrasonic transmitter periodically emits pulses, and the receiver captures the signal to calculate the wind speed component. A Kalman filter algorithm is used to eliminate turbulence interference, outputting stable wind speed data for power prediction.

[0019] Light sensor: Measurement range 0-2000W / m 2 Accuracy ±1W / m 2 The sensor is mounted on the surface of the photovoltaic array; it employs a silicon-based photovoltaic sensor (model: LI-250A) with an anti-reflective coating to enhance response sensitivity. One sensor is installed on each 50kW photovoltaic array surface (parallel to the panel). Data is synchronized to the control module via a CAN bus. The PN junction inside the sensor generates a photocurrent when illuminated, and the magnitude of the current is proportional to the irradiance. Calibration using an IV characteristic curve eliminates the influence of temperature on the output signal.

[0020] Current / voltage transformer: range 0-1000A / 0-1000V, accuracy ±0.5%, installed at the grid connection point of new energy power generation; Energy storage charge and discharge monitoring module: Real-time acquisition of battery SOC, charge and discharge power, temperature and other data, with an accuracy of ±1%; Vibration sensor: range ±10g, accuracy ±0.1g, mounted on the surface of the SVG power module, used to monitor mechanical vibration; Acoustic sensor: range 30-20000Hz, accuracy ±1dB, mounted on the surface of the heat sink, used to monitor abnormal fan noise.

[0021] Control subsystem: Main controller: adopts a DSP+FPGA+ARM tri-core architecture, with a main frequency of 1GHz, and runs the Linux real-time operating system; Power module: Adopts cascaded H-bridge topology, single module capacity 500kvar, withstand voltage 10kV, efficiency ≥98%; Energy storage interface: Supports various energy storage methods such as lithium-ion batteries, supercapacitors, and flywheel energy storage; charging and discharging power is adjustable. Communication module: Supports multiple communication methods such as 5G, fiber optic Ethernet, RS485, and CAN bus, and is compatible with the IEC 61850 protocol.

[0022] Specifically, the LSTM neural network prediction unit of the multi-source power prediction module uses the Adam optimizer; the multi-source power prediction algorithm is based on LSTM neural network, CNN and attention mechanism to construct a multi-source power prediction model. A four-layer LSTM architecture (input layer, two hidden layers, and output layer) is used, with 64 neurons in each layer. Input data includes historical wind speed, light intensity, power generation, energy storage SOC, etc., with a time window set to 1 hour. The model is deployed on a DSP+FPGA controller, updating the prediction results every 100ms. Information flow is controlled through forget gates, input gates, and output gates to capture long-term dependencies in the time series. The model is trained using the Adam optimizer, with the mean squared error (MSE) loss function. Training data covers all four seasons and is automatically updated every 15 minutes. Model parameter input data includes historical wind speed, light intensity, power generation, energy storage charging and discharging power, temperature, humidity, etc., and the output is the predicted value of new energy power generation for the next 10 seconds to 1 hour. The model was trained using the Adam optimizer with a learning rate of 0.001, a batch size of 32, and 1000 training cycles, achieving an accuracy of ≥95%.

[0023] A multi-head attention layer is added after the LSTM output layer, and weights are assigned by calculating the dot product of the query vector, key vector, and value vector. Multi-source data such as wind speed, illumination, and energy storage are fused to improve prediction accuracy. The attention weights are dynamically allocated based on the importance of the input features. For example, in cloud-occupied scenarios, the weight of illumination data is automatically increased, while the weight of wind speed data is decreased, achieving adaptive feature fusion.

[0024] Specifically, the dynamic reactive power compensation algorithm employs fuzzy PID control, MPC, and DRL algorithms to achieve dynamic adjustment of reactive current. Fuzzy PID control dynamically adjusts K based on the deviation ΔP between predicted and actual power and the rate of change of that deviation (ΔP'). p K i K d The parameters and formula are as follows: IQ(t) = K p •ΔP(t)+K i •∫0 t ΔP(τ)dτ+Kd•ΔP′(t) Among them, K p K i K dThe parameters are adjusted in real time by the fuzzy inference engine. The input variable range is: ΔP∈[-50MW, 50MW], ΔP′∈[-10MW / s, 10MW / s]. The MPC algorithm solves for the optimal reactive current in each control cycle through a rolling optimization strategy, taking into account constraints (such as power module capacity and voltage limits). The DRL algorithm uses a Q-learning algorithm to learn the optimal compensation strategy through interaction with the environment, thereby improving long-term stability. Specifically, the composite heat dissipation module comprises two systems: a liquid cooling system and an external circulation system. The liquid cooling plate is made of copper with a thermal conductivity ≥400W / m·K and is directly bonded to the IGBT power module. The inner circulation pipe is made of stainless steel with a diameter of 10mm and an adjustable flow rate of 0-10L / min. The external circulation loop uses an air-water separator and a buffer tank with a capacity of 20L to achieve cooling water recycling. The liquid cooling plate adopts a microchannel structure (channel width 0.5mm, height 2mm) and is bonded to the IGBT module using thermally conductive silicone grease (thermal conductivity 5W / m·K). The inner circulation pipe uses a stainless steel corrugated pipe, and the flow rate is adjusted by an electronic throttle valve. The external circulation loop is equipped with a plate heat exchanger, where the cooling water exchanges heat with the fan coil unit.

[0025] Air-cooled system: The variable frequency fan uses a DC brushless motor with an adjustable speed of 500-3000 rpm and an adjustable air volume range of 0-500 m³ / h. 3 / h; The external circulation loop pipe is made of aluminum, with a pipe diameter of 20mm and a heat sink spacing of 5mm; Phase change material: Paraffin-based PCM with a phase change temperature of 40℃ and a thermal conductivity ≥2W / m·K is installed on the surface of the power module to absorb instantaneous heat. Temperature sensor: PT100 platinum resistance thermometer, range -50℃ to +200℃, accuracy ±0.1℃, installed on the surface of power module, water cooling plate, and fan for real-time temperature monitoring.

[0026] Specifically, the expert system reasoning unit of the fault self-diagnosis module includes a rule base and an inference engine. The rule base covers diagnostic rules for common equipment faults; the machine learning recognition unit uses a random forest algorithm, with training data including historical fault data and normal operation data, and feature variables including temperature, current, voltage, vibration, and acoustics. Based on the expert system and machine learning algorithm, fault early warning and location are achieved. The expert system includes a rule base and an inference engine. The rule base covers diagnostic rules for common equipment faults (such as over-temperature, over-voltage, and insulation faults); the inference engine matches real-time monitoring data with the rule base to generate fault early warning information. The expert system inference unit rule base contains 200 fault diagnosis rules, covering common faults such as over-temperature, over-voltage, and insulation faults. The inference engine uses a forward inference chain, matching rules with real-time monitoring data to generate fault codes and suggested measures. For example, when the IGBT module temperature is detected to be >55℃ and the cooling system is normal, the "heat dissipation system fault" rule is triggered, generating an alarm code and prompting to check the fan or water cooling plate.

[0027] A random forest classifier was employed, with training data including historical fault samples (such as abnormal fan noise and excessive current harmonics) and normal samples. Feature variables included temperature, vibration, acoustics, and current harmonics, resulting in a total of 12 features. Feature engineering was used to optimize feature weights; for example, the frequency domain features of vibration signals were given increased weight in bearing fault identification. The model accuracy was ensured to be ≥90% through cross-validation, with a false positive rate ≤5%.

[0028] Specifically, the communication and protection module supports IEC61850-90-5, IEC61850-90-7 and 5G communication protocols, and has overcurrent, overvoltage, undervoltage, overtemperature and insulation fault protection functions; When this invention is used, the new energy power generation status monitoring module is used to collect data on wind speed, light intensity, power generation, and energy storage charging and discharging power in real time; the multi-source power prediction module predicts the new energy power generation for the next 10 seconds to 1 hour based on the monitoring data; the dynamic compensation control module dynamically adjusts the reactive current output of the SVG based on the prediction results; the composite heat dissipation module automatically adjusts the heat dissipation power based on the temperature threshold; the fault self-diagnosis module monitors the equipment status in real time and provides early warnings to locate faults; and the communication and protection module ensures the safe and stable operation of the equipment.

[0029] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A novel SVG device that automatically follows the state of new energy power generation, characterized in that: It includes a new energy power generation status monitoring module, used to collect real-time data on wind speed, light intensity, power generation, and energy storage charging and discharging power; The multi-source power prediction module includes an LSTM neural network prediction unit, a CNN feature extraction unit, and an attention mechanism fusion unit, which is used to predict the power generation of new energy sources in the next 10 seconds to 1 hour based on monitoring data. The dynamic compensation control module includes a fuzzy PID control unit, an MPC optimization unit, and a DRL learning unit, which are used to dynamically adjust the reactive current output of the SVG based on the prediction results. The composite heat dissipation module includes a liquid cooling subsystem, an air cooling subsystem, and a phase change material heat dissipation subsystem, which is used to automatically adjust the heat dissipation power according to the temperature threshold. The fault self-diagnosis module includes an expert system reasoning unit and a machine learning recognition unit, which are used to monitor the equipment status in real time and provide early warnings to locate faults. The communication and protection module supports multiple communication protocols and protection functions to ensure the safe and stable operation of the equipment.

2. The novel SVG device for automatically following the state of new energy power generation according to claim 1, characterized in that: The new energy power generation status monitoring module includes an anemometer, a light sensor, a current / voltage transformer, an energy storage charging and discharging monitoring module, a vibration sensor, and an acoustic sensor.

3. The novel SVG device for automatically following the state of new energy power generation according to claim 2, characterized in that: The LSTM neural network prediction unit of the multi-source power prediction module uses the Adam optimizer.

4. The novel SVG device for automatically following the state of new energy power generation according to claim 3, characterized in that: The fuzzy PID control unit of the dynamic compensation control module dynamically adjusts K based on the deviation ΔP between the predicted power and the actual power and the rate of change of the deviation ΔP′. p K i K d The parameters and formula are as follows: IQ(t)=K p •ΔP(t)+K i •∫0 t ΔP(τ)dτ+K d •ΔP′(t) Among them, K p K i K d The parameters are adjusted in real time by the fuzzy inference engine. The input variable range is: ΔP∈[-50MW, 50MW], ΔP′∈[-10MW / s, 10MW / s].

5. The novel SVG device for automatically following the state of new energy power generation according to claim 4, characterized in that: The liquid cooling subsystem of the composite heat dissipation module includes a water cooling plate, an internal circulation pipe, an air-water separator, and a buffer water tank.

6. The novel SVG device for automatically following the state of new energy power generation according to claim 5, characterized in that: The expert system reasoning unit of the fault self-diagnosis module includes a rule base and a reasoning engine. The rule base covers diagnostic rules for common equipment faults. The machine learning recognition unit uses the random forest algorithm. The training data includes historical fault data and normal operation data. The feature variables include temperature, current, voltage, vibration, and acoustics.

7. The novel SVG device for automatically following the state of new energy power generation according to claim 6, characterized in that: The communication and protection module supports IEC61850-90-5, IEC61850-90-7 and 5G communication protocols, and has overcurrent, overvoltage, undervoltage, overtemperature and insulation fault protection functions.

Images