Multi-power intelligent photovoltaic combiner box based on AI drive and working method

Abstract

The invention discloses a multi-power intelligent photovoltaic combiner box based on AI driving and a working method, and relates to the technical field of solar photovoltaic accessories. The photovoltaic combiner box supports the access of switching power supplies with different voltage levels, and comprises a plurality of independent power supply access loops, a photovoltaic access unit, a convergence unit, a current diverter, a load cut-in unit and an AI monitoring module. The AI monitoring module is combined with reinforcement learning and a fuzzy control strategy, so that dynamic adjustment of current distribution, load switching optimization and fault prediction and response are realized, the problem of countercurrent of a multi-power system caused by inconsistent busbar voltage is solved, the stability and the intelligent level of the system are improved, the construction and operation and maintenance cost is reduced, and the system is suitable for large-scale popularization and application. The photovoltaic module is widely applicable to photovoltaic application in communication machine rooms and other complex power supply environments.

Description

Technical Field

[0001] The present invention relates to the technical field of solar photovoltaic accessories, and particularly relates to an AI-driven multi-power intelligent photovoltaic busbar box and a working method thereof. Background Technique

[0002] In the technology of solar power supply systems for communication, the busbar box mainly undertakes the functions of current collection and circuit protection. Generally, it consists of 1 to N photovoltaic inputs, circuit breaker or fuse protection for the corresponding number of circuits, busbar copper bars, lightning protection units, power measurement units, communication units, and load switching units. Generally, 1 busbar box is connected to 1 switching power supply and superimposed into the current switching power supply to provide green energy with priority for photovoltaic use.

[0003] However, in existing communication machine rooms, due to reasons such as multi-operator sharing and historical construction, many machine rooms are powered by multiple switching power supply systems; with the construction of the minimalist stations of operators, there are also scenarios where a site has both a switching power supply and a new type of blade power supply, etc., with dual-power supply.

[0004] For the above multi-power scenarios, if the existing scheme is used with a single busbar box and the output unit is connected to different switching power supplies through multiple wires, since it is impossible to ensure that the busbar voltages of the two switching power supplies are the same, unacceptable scenarios such as reverse current or only single-power operation will occur and it cannot be used. If multiple sets of photovoltaic devices are used in a multi-power access site, it will undoubtedly increase the number of platform entries and maintenance, increase the number of fault points in the future, and also increase the cost.

[0005] Therefore, how to design an intelligent photovoltaic busbar box that can support independent access of multiple powers, avoid reverse current problems caused by inconsistent busbar voltages, and achieve dynamic regulation of current distribution and load switching management through intelligent optimization has become an urgent technical problem to be solved. Summary of the Invention

[0006] In view of the above problems, the present invention provides an AI-driven multi-power intelligent photovoltaic busbar box and a working method thereof. Combining AI-driven fault prediction and control optimization algorithms, and through advanced technologies such as reinforcement learning and fuzzy control to perform fault prediction and response adjustment on the system, it will provide important technical support for the efficient operation of the multi-power photovoltaic system in the communication machine room.

[0007] The present invention achieves the above object through the following technical solutions:

[0008] An AI-driven multi-power intelligent photovoltaic busbar box, the busbar box supports the access of switching power supplies with different voltage levels, and includes: a plurality of power access circuits, and each power access circuit works independently and does not interfere with each other. Each independent power access circuit includes:

[0009] Photovoltaic access unit, which is used to access an independent photovoltaic array and transmit DC electrical energy to the busbar unit;

[0010] Busbar unit, which is used to collect the current of the photovoltaic array and uniformly transmit it to the current shunt;

[0011] Current shunt, which is used to monitor and distribute the current from the busbar unit;

[0012] Load switching-in unit, which is used to determine the access and switching of the switching power supply according to the load demand and the health status of the circuit, support the situation where any one path is powered or all paths are powered, and automatically switch to other power supply paths when a fault occurs in any one power supply path;

[0013] Lightning protection unit, which is used to integrate lightning protection to prevent the circuit from being affected by lightning strikes or voltage surges;

[0014] The busbar box further includes an AI monitoring module, which is used to detect the health status of the circuit by combining reinforcement learning and fuzzy control strategies, adjust the current distribution and power supply switching in real time, and adjust the fuzzy control rules according to real-time data, predict potential faults, and provide fault diagnosis and response.

[0015] As a preferred solution of the present invention, the busbar box further includes:

[0016] Grounding unit, which is used to provide electrical grounding protection for the busbar box and all power supply access circuits;

[0017] Local load power supply unit, which is used to provide power support for local loads according to the electrical energy output by the circuit;

[0018] Multi-loop metering unit, which is used to measure the electricity quantity of each power supply access circuit, monitor the voltage, current and power of each circuit in real time, and upload them to the communication unit;

[0019] Communication unit, which is used to transmit the data uploaded by the multi-loop metering unit and the operation status information of the busbar box to the AI monitoring module for data analysis, fault diagnosis and optimization.

[0020] As a preferred solution of the present invention, the AI monitoring module includes:

[0021] Fault detection unit, which is used to monitor the current, voltage and temperature of each power supply access circuit in real time, judge whether there is a fault, and feedback the fault information to the load switching decision unit;

[0022] Load switching decision unit, which is used to intelligently judge whether it is necessary to switch the power supply access circuit according to the fault information provided by the fault detection unit and the real-time load demand to ensure stable power supply for the load. If necessary, it issues an instruction to automatically switch to another power supply path through the load switching-in unit;

[0023] The current optimization control unit is used to dynamically adjust the current distribution according to the real-time current and load demand, and optimize the current output of each loop through a fuzzy control strategy;

[0024] The learning and prediction unit is used to analyze the historical operation data and real-time monitoring data through a reinforcement learning algorithm, predict potential loop faults, and generate an optimization control strategy to adjust the current distribution and load switching decision in advance.

[0025] As a preferred solution of the present invention, the fault detection unit performs fault mode recognition based on the Q-learning algorithm, calculates the Q value of each loop, selects the optimal strategy and updates it dynamically. The specific implementation process is as follows:

[0026] Define the state space S and the action space A. The state space S represents the current, voltage and temperature of each loop, and the action space A represents the fault detection and diagnosis actions;

[0027] When the fault detection unit is in the current state s t it selects the action a t and judges whether the health state of the loop is normal;

[0028] After executing the action a t a state transition is performed, from the current state s t to the new state s t+1 and the immediate reward value r t+1 is calculated according to the executed action;

[0029] If the current, voltage or temperature of the loop is within the normal range, the immediate reward value r t+1 is +1, indicating normal operation; if the state exceeds the normal range and a fault such as current overload, voltage abnormality or overheating occurs, the immediate reward value r t+1 is -1, indicating a fault state;

[0030] Once the immediate reward value r t+1 is obtained, the update formula of the Q-learning algorithm is used to update the Q value corresponding to the current state s t and the selected action a t . The update formula is:

[0031]

[0032] In the formula, Q(s t ,a t ) represents the Q value of executing the action a t in the state s t ; α is the learning rate; r t+1 is the immediate reward value given according to the health state of the loop; γ is the discount factor; a′ is the next state s t+1The best action among all actions below;

[0033] Whenever the state changes or different rewards are generated when performing an action, the Q-learning algorithm dynamically adjusts the Q value, making the fault detection strategy gradually tend to be optimal.

[0034] As a preferred embodiment of the present invention, the load switching decision unit performs decision optimization based on the multi-armed bandit problem and selects a power supply switching strategy, specifically including:

[0035] Each power supply access loop is regarded as a lever, and each time any power supply access loop is selected for switching, a reward will be obtained. The reward r t represents the stability of the load after power supply switching. r t being +1 indicates that the load is stable and the power supply meets the demand. r t being -1 indicates that the load is unstable and the power supply is insufficient;

[0036] Based on the current reward and historical rewards, use the greedy strategy or ε-greedy strategy to select a power supply access loop for switching, and calculate the reward value r according to the power supply stability of the load after switching t ;

[0037] By recording the power supply access loop and reward selected each time, gradually accumulate data and update the reward value, and calculate the reward for each power supply access loop. The formula is:

[0038]

[0039] In the formula, Q(A i ) represents the reward value of the power supply access loop A i ; represents the immediate reward after the t-th selection when selecting the power supply access loop A i ; N i is the number of times the power supply access loop A i is selected; T is the total number of decisions;

[0040] Select the optimal power supply access loop according to the updated reward value. The power supply access loop with the highest reward is selected as the first choice for the next power supply switching. If the rewards of multiple power supply access loops are the same, select through the ε-greedy strategy;

[0041] Continuously record the rewards of each power supply switching, update the reward values of each power supply access loop, and adjust the selection strategy.

[0042] As a preferred embodiment of the present invention, the current optimization control unit monitors the current, load demand, and operating status of the power supply of each loop in real time, and evaluates whether the current distribution meets the expectation;

[0043] According to the non-linear relationship between current and load demand, a fuzzy control strategy is adopted to dynamically adjust the current distribution to optimize the current output of each loop.

[0044] The fuzzy control strategy adjusts the current based on fuzzy rules. When the current of any loop is lower than the preset threshold, it is judged according to the fuzzy control rules whether the current output of this loop needs to be increased, and corresponding adjustments are made according to the load demands of other loops.

[0045] As a preferred solution of the present invention, the learning and prediction unit adopts a deep reinforcement learning algorithm to predict potential loop faults and generate an optimized control strategy, including the following steps:

[0046] By analyzing historical fault patterns, current, voltage, temperature, load demand and real-time operation data, a deep reinforcement learning model is established to predict possible future system faults.

[0047] Through a deep neural network DNN, learn the potential patterns of load switching, fault occurrence and current distribution. During the training process, use Q-learning or Actor-Critic algorithm to optimize the current distribution and load switching decisions according to historical data, and predict the probability of fault occurrence.

[0048] Based on the learned strategies and models, adjust the current distribution and load switching decisions in advance, use the deep reinforcement learning model to predict the potential risks of fault occurrence, adjust the current distribution in advance, and optimize the load switching decisions.

[0049] A working method of an AI-driven multi-power intelligent photovoltaic busbar box, which is applied to the AI-driven multi-power intelligent photovoltaic busbar box as described above. The method includes:

[0050] Connect each independent photovoltaic array to the busbar box through the photovoltaic access unit. The busbar unit collects the DC electric energy from each photovoltaic array and transmits it to the current shunt for current monitoring and distribution.

[0051] The current shunt distributes the current according to the current from the busbar unit and monitors the current status of each loop in real time. During the current distribution process, the AI monitoring module adjusts the current distribution in combination with real-time data to ensure that the current output of each loop meets the load demand and avoid overload or uneven current.

[0052] According to the load demand and the health status of the loop, the load switching unit intelligently judges the access status of the current power supply, and automatically switches the power supply according to the prediction or fault diagnosis result. If any power supply fails, it will automatically switch to other available power supplies according to the judgment of the AI monitoring module to ensure continuous power supply to the load.

[0053] The AI monitoring module continuously detects the health status of the system loop by combining reinforcement learning and fuzzy control strategies, and adjusts the current distribution and power switching strategies in real time. By learning historical data and real-time monitoring information, it predicts the occurrence of potential faults and takes adjustment measures in advance according to the prediction results, reducing the probability of fault occurrence and ensuring the stable operation of the system;

[0054] When a potential fault or load demand change is predicted, the current distribution strategy is dynamically adjusted using fuzzy control rules to optimize the current output of each loop.

[0055] The beneficial effects of the present invention are as follows: By connecting multiple independent power sources to the loop, the independent operation and non-interference of power sources with different voltage levels are achieved, solving the problems of reverse current and single-power operation caused by inconsistent busbar voltages in a multi-power system; The design of the photovoltaic access unit, the busbar unit, and the current shunt, combined with the intelligent switching function of the load switching unit, can automatically switch the power source according to the load demand and the loop health status, ensuring continuous power supply in case of a fault in any power source, improving the reliability of the system; The AI monitoring module combines reinforcement learning and fuzzy control strategies to achieve dynamic adjustment of current distribution and load switching, and has the ability of fault prediction and diagnosis, effectively reducing the fault risk and ensuring the stable operation of the system. It also integrates a lightning protection module and an electricity metering unit, providing circuit protection and electricity monitoring capabilities, and further reducing the operation and maintenance costs through the remote monitoring function, comprehensively improving the intelligence and practicality of the photovoltaic busbar box. Brief Description of the Drawings

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts.

[0057] Among them:

[0058] Figure 1 is the module structure diagram of the photovoltaic busbar box in the embodiment of the present invention;

[0059] Figure 2 is the method flow chart in the embodiment of the present invention. Detailed Embodiments

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present invention fall within the scope of protection of the present invention.

[0061] As Figure 1 shown, this is an embodiment of the present invention. This embodiment provides an AI-driven multi-power intelligent photovoltaic busbar box that supports the access of switching power supplies with different voltage levels, includes multiple power access circuits, and each power access circuit works independently without interference;

[0062] Taking a dual-power supply as an example in this embodiment, the input and output are divided into 2 circuits. The input branches of each circuit are required to be flexibly configurable. The output branches are 2 circuits, and the two circuits are independent. Typical configurations are such as 4+4, 8+8, 4+8, etc. Each independent power access circuit includes:

[0063] Photovoltaic access units (loop 1 photovoltaic access unit, loop 2 photovoltaic access unit), which are used to access independent photovoltaic arrays and transmit DC electrical energy to the busbar unit;

[0064] Busbar units (loop 1 busbar unit, loop 2 busbar unit), which are used to collect the current of the photovoltaic array and uniformly transmit it to the current shunt;

[0065] Current shunts (loop 1 current shunt, loop 2 current shunt), which are used to monitor and distribute the current from the busbar unit;

[0066] Load switching-in units (loop 1 load switching-in unit, loop 2 load switching-in unit), which are used to determine the access and switching of the switching power supply according to the load demand and the health status of the circuit, support the situation where any one circuit is powered or all circuits are powered, and automatically switch to other circuits when any one power supply fails;

[0067] Lightning protection units (loop 1 lightning protection unit, loop 2 lightning protection unit), which are used to integrate lightning protection to avoid the influence of the circuit being struck by lightning or voltage surges; have independent SPD module lightning protection. The nominal operating voltage of the SPD should be greater than 1.2 times the maximum operating voltage of the superposition light controller. The rated discharge current In of the SPD ≥ 10 kA, and the lightning protection module has a remote signaling function;

[0068] It also includes:

[0069] AI monitoring module, which is used to detect the health status of the circuit by combining reinforcement learning and fuzzy control strategies, adjust the current distribution and power supply switching in real time, and adjust the fuzzy control rules according to real-time data, predict potential faults, and provide fault diagnosis and response;

[0070] The AI monitoring module has the function of accessing the tower monitoring platform, and can monitor the working status of the system in real time; collect and store the system operation parameters; adopt a wireless transmission module to realize data transmission between the controller and the monitoring platform through the MQTT protocol, and support remote upgrade. The monitoring contents are as follows:

[0071] (1) Photovoltaic adapter

[0072] Telemetry: adapter manufacturer, adapter model, adapter software version, adapter power-on / off status, adapter temperature, adapter input voltage / current, adapter output voltage / current, adapter output power.

[0073] Tele-signal: adapter fault, adapter protection trigger, adapter communication alarm, adapter high-temperature alarm.

[0074] (2) Photovoltaic controller

[0075] Telemetry: system type, IMSI, MSISDN, SIM card ICCID, software and hardware version, controller manufacturer, controller model, total system power, cumulative power generation.

[0076] Tele-signal: system output voltage too high alarm, system lightning protection device status, controller fault alarm.

[0077] Tele-control: system zero point time calibration.

[0078] (3) DC ammeter

[0079] Telemetry: ammeter number, total DC current, DC voltage, total active power, total active energy, daily active energy, daily forward active energy and total forward active energy at peak, valley and flat periods.

[0080] Tele-signal: fault alarm.

[0081] Tele-adjustment: start and end times of peak, valley and flat periods.

[0082] (4) Environmental quantity

[0083] Telemetry: temperature, humidity, wind direction, wind speed, irradiance intensity.

[0084] Grounding unit, which is used to provide electrical grounding protection for the busbar box and all power supply access circuits; the protected grounding point should have obvious signs, and the resistance between the shell and all accessible non-electrified metal parts and the protected grounding point should not be greater than 0.1Ω;

[0085] A local load power supply unit for providing power support to a local load according to the electric energy output by a circuit;

[0086] A multi-circuit metering unit for measuring the electricity consumption of each power access circuit, real-time monitoring the voltage, current and power of each circuit, and uploading them to the communication unit; the DC ammeters used by the multi-circuit metering unit shall comply with "GB / T 33708-2017 Static DC Watt-hour Meter", have CPA certification, an accuracy not lower than Class 1, and have forward, reverse and time-of-use electricity metering functions. Requirements for DC ammeters:

[0087] a) The ammeter shall have CPA metering certification;

[0088] b) The ammeter shall have the function of separately metering forward and reverse electric energy;

[0089] c) The ammeter shall be able to measure real-time DC voltage, current and power parameters;

[0090] d) The ammeter shall have an RS-485 communication interface and support DL / T 645-2007 protocol or Modbus-RTU protocol;

[0091] e) The auxiliary power supply of the ammeter shall support DC 20V to 60V;

[0092] f) The accuracy of the ammeter is Class 1;

[0093] g) Support dual-circuit metering ability, and can display the voltage, current and electric energy of two DC branches on the display screen;

[0094] A communication unit for transmitting the data uploaded by the multi-circuit metering unit and the operation status information of the busbar trunking unit to the AI monitoring module for data analysis, fault diagnosis and optimization;

[0095] A box body, which is firm, has a flat surface, uniform color, and no defects such as rust, wrinkles and flow marks; the signs are complete and clear; the lock should be easy to operate and flexible and reliable; a warning sign indicating that the metal parts inside the box are electrified is marked at a prominent position. All cables inside the busbar trunking unit use flame-retardant cables, and communication cables and signal acquisition cables all need to use shielded cables. Installation holes are arranged on the back of the box body, and the specific arrangement position and hole diameter shall be determined in cooperation with the bracket to ensure smooth installation and fixation.

[0096] Furthermore, the AI monitoring module includes a fault detection unit, a load switching decision-making unit, a current optimization control unit and a learning and prediction unit.

[0097] The fault detection unit is used for real-time monitoring the current, voltage and temperature of each power access circuit, judging whether there is a fault, and feeding back the fault information to the load switching decision-making unit;

[0098] Specifically, the fault detection unit performs fault mode recognition based on the Q-learning algorithm, calculates the Q value of each loop, selects the optimal strategy and updates it dynamically. The specific implementation process is as follows:

[0099] Define the state space S and the action space A. The state space S represents the current, voltage and temperature of each loop, and the action space A represents fault detection and diagnosis actions, such as normal operation, current overload alarm, voltage anomaly alarm, etc.;

[0100] When the fault detection unit is in the current state s t it selects the action a t and determines whether the health state of the loop is normal;

[0101] After executing the action a t a state transition occurs, from the current state s t to the new state s t+1 , and the immediate reward value r t+1 is calculated according to the executed action;

[0102] If the current, voltage or temperature of the loop is within the normal range, the immediate reward value r t+1 is +1, indicating normal operation; if the state exceeds the normal range and a fault such as current overload, voltage anomaly or overheating occurs, the immediate reward value r t+1 is -1, indicating a fault state;

[0103] Once the immediate reward value r t+1 is obtained, the update formula of the Q-learning algorithm is used to update the Q value corresponding to the current state s t and the selected action a t . The update formula is:

[0104] Q(s t ,a t )←Q(s t ,a t )+α[r t+1 +γm a a ′ xQ(s t+1 ,a′)-Q(s t ,a t )];

[0105] In the formula, Q(s t ,a t ) represents the Q value of executing the action a t in the state s t (i.e., the expected benefit of executing the action); α is the learning rate, which is used to control the influence degree of new experience on the Q value update; r t+1$r$ is the immediate reward value given according to the health state of the circuit. If the circuit state is normal, the reward is positive; if a fault (such as overload, overvoltage or overtemperature) is detected, the reward is negative; $\gamma$ is the discount factor, indicating the impact of future rewards on the current decision; $a'$ is the best action among all actions in the next state $s$. t+1 among all actions in the next state $s$.

[0106] Whenever the state changes or different rewards are generated by executing an action, the Q - learning algorithm dynamically adjusts the Q - value, making the fault detection strategy gradually tend to be optimal.

[0107] Through the Q - learning algorithm, the system selects an action $a$ according to the current current, voltage and temperature of the circuit (i.e., the current state $s$). t (e.g., marked as "normal" or "fault"). Then, the system observes the state transition caused by executing this action (such as current overload, over - voltage, etc.), and updates the Q - value according to the new state $s$. t so as to make better fault recognition decisions in the future. t+1

[0108] By using the Q - learning algorithm for fault mode recognition, the optimal detection strategy is selected by dynamically updating the Q - value, so as to achieve accurate fault judgment and response; through the self - learning ability of reinforcement learning, the fault detection unit can continuously optimize the fault detection accuracy according to historical data and real - time monitoring data, improve the intelligent level of the system, efficiently identify faults in the multi - power system, provide accurate feedback and ensure the reliability and stability of the system.

[0109] The load switching decision - making unit is used to intelligently judge whether it is necessary to switch the power access circuit according to the fault information provided by the fault detection unit and in combination with the real - time load demand to ensure stable power supply for the load. If necessary, it issues an instruction to automatically switch to another power supply through the load switching - in unit.

[0110] Specifically, the load switching decision - making unit optimizes the decision - making based on the multi - armed bandit problem and selects the power - switching strategy. The multi - armed bandit problem is a classic decision - making optimization problem, mainly used to handle the situation of selecting the optimal action (i.e., which "lever" to pull) in an uncertain environment. In this problem, assume there are multiple "levers" (actions), the reward distribution of each lever is unknown, and the goal of the decision - maker is to maximize the total reward by selecting these levers multiple times. In the load switching decision - making, each power supply circuit or standby power supply is equivalent to a "lever", and each decision to switch a power supply circuit will get a "reward", and the reward value is directly related to the power supply stability of the load after the power supply is switched (whether it can meet the load demand).

[0111] Specifically, it includes:

[0112] ​Each power supply access loop is regarded as a lever. Each time any power supply access loop is selected for switching, a reward will be obtained, and the reward is r t represents the stability of the load after power supply switching. r t being +1 indicates that the load is stable and the power supply meets the demand. r t being -1 indicates that the load is unstable and the power supply is insufficient;

[0113] Based on the current reward and historical rewards, use the greedy strategy or ε-greedy strategy to select a power supply access loop for switching, and calculate the reward value r according to the stability of the load power supply after switching t ;

[0114] Greedy strategy: Select the power supply loop with the maximum reward (i.e., the load is stable and the power supply is reliable). Each time of switching, select the action with the maximum Q value;

[0115] ε-greedy strategy: Most of the time, select the current optimal power supply loop (the maximum Q value), and occasionally select other loops for exploration to ensure that the system continuously tries different power configurations and avoids falling into local optima;

[0116] By recording the power supply access loop and reward selected each time, gradually accumulate data and update the reward value, and calculate the reward for each power supply access loop. The formula is:

[0117]

[0118] In the formula, Q(A i ) represents the reward value of the power supply access loop A i (i.e., the selected "lever" or power supply loop); represents the immediate reward after the t-th selection when selecting the power supply access loop A i . The reward value is usually an indication of whether the load power supply is stable; N i is the number of times the power supply access loop A i is selected, that is, the frequency of loop A i being pulled as a "lever"; T is the total number of decisions. During the learning process, T represents the time step or the number of selections;

[0119] Select the optimal power supply access loop according to the updated reward value. The power supply access loop with the highest reward is selected as the first choice for the next power supply switching. If the rewards of multiple power supply access loops are the same, select through the ε-greedy strategy;

[0120] Continuously record the rewards of each power supply switching, update the reward values of each power supply access loop, and adjust the selection strategy.

[0121] Through the above process, the load switching decision-making unit can utilize the optimization method of the multi-armed bandit problem in a multi-power system and select the optimal power switching scheme based on the reward maximization strategy. By using the ε-greedy strategy to balance exploration and exploitation, it continuously accumulates the selection experience of power circuits, optimizes the power switching decision, and thus ensures the stability of load power supply.

[0122] The current optimization control unit is used to dynamically adjust the current distribution according to the real-time current and load demand, and optimize the current output of each circuit through the fuzzy control strategy;

[0123] Specifically, the current optimization control unit monitors the current, load demand, and operating status of the power supply of each circuit in real time, and evaluates whether the current distribution meets the expectations;

[0124] According to the non-linear relationship between the current and the load demand, the fuzzy control strategy is used to dynamically adjust the current distribution to optimize the current output of each circuit;

[0125] The fuzzy control strategy adjusts the current based on fuzzy rules. When the current of any circuit is lower than the preset threshold, it judges whether to increase the current output of this circuit according to the fuzzy control rules, and makes corresponding adjustments according to the load demands of other circuits.

[0126] The current optimization control unit combines historical data and real-time monitoring data to predict the change trend of current demand, adjusts the current distribution in advance, ensures the stable operation of the system, and maximizes the load balance and system efficiency.

[0127] The learning and prediction unit is used to analyze the historical operation data and real-time monitoring data through the reinforcement learning algorithm, predict potential circuit faults, and generate an optimization control strategy to adjust the current distribution and load switching decision in advance.

[0128] Specifically, the learning and prediction unit adopts the deep reinforcement learning algorithm to predict potential circuit faults and generate an optimization control strategy, including the following steps:

[0129] By analyzing the historical fault patterns, current, voltage, temperature, load demand, and real-time operation data, a deep reinforcement learning model is established to predict possible future system faults; the input features of the model include historical fault records and real-time data, and feature engineering is used to preprocess the data; learning is carried out through a deep neural network (DNN) or other suitable neural network architectures (such as LSTM or CNN) to predict possible future system faults;

[0130] Through the deep neural network DNN, learn the potential patterns of load switching, fault occurrence, and current distribution. During the training process, the Q-learning or Actor-Critic algorithm is used to optimize the current distribution and load switching decision according to historical data, and predict the probability of fault occurrence;

[0131] By using the learned strategies and models, adjust the current distribution and load switching decisions in advance, and use the deep reinforcement learning model to predict the potential risks of fault occurrence, adjust the current distribution in advance, and optimize the load switching decisions, so as to ensure that preventive control is achieved by adjusting the load switching strategy before the fault occurs, thereby reducing the probability of fault occurrence and ensuring the stable operation of the system.

[0132] As Figure 2 shown, this is another embodiment of the present invention. This embodiment provides a working method for an AI-driven multi-power intelligent photovoltaic combiner box, which is applied to the AI-driven multi-power intelligent photovoltaic combiner box as described above, and includes the following steps:

[0133] Photovoltaic array access and current collection: Each independent photovoltaic array is connected to the combiner box through the photovoltaic access unit, and the current collection unit collects the DC electrical energy from each photovoltaic array and transmits it to the current shunt for current monitoring and distribution;

[0134] Current distribution and monitoring: The current shunt distributes the current from the current collection unit and monitors the current status of each circuit in real time. During the current distribution process, the AI monitoring module adjusts the current distribution in combination with real-time data to ensure that the current output of each circuit meets the load requirements and avoid overload or uneven current;

[0135] Load switching control: According to the load demand and the health status of the circuit, the load switching unit intelligently judges the access status of the current power supply, and automatically switches the power supply according to the prediction or fault diagnosis result. If any power supply fails, it will automatically switch to other available power supplies according to the judgment of the AI monitoring module to ensure continuous power supply to the load;

[0136] Fault prediction and response: The AI monitoring module continuously detects the health status of the system circuit by combining reinforcement learning and fuzzy control strategies, and adjusts the current distribution and power supply switching strategies in real time. By learning historical data and real-time monitoring information, it predicts the occurrence of potential faults and takes adjustment measures in advance according to the prediction results to reduce the probability of fault occurrence and ensure the stable operation of the system;

[0137] Current adjustment and optimization control: When potential faults or load demand changes are predicted, use fuzzy control rules to dynamically adjust the current distribution strategy and optimize the current output of each circuit.

[0138] In summary, through the design of multiple power access circuits, each circuit can work independently without interference, supporting the access of switching power supplies with different voltage levels, and solving the reverse current problem caused by inconsistent bus voltages in a multi-power system. At the same time, it avoids the problem of independently configuring photovoltaic equipment for each power system in the traditional solution, reducing the construction and maintenance costs of the system.

[0139] Through the photovoltaic access unit, the busbar unit and the current shunt, combined with the AI monitoring module to dynamically adjust the current distribution, ensuring the balance and efficient output of the current distribution in each circuit. The load switching unit can automatically switch the power supply according to the load demand and the circuit health status, ensuring continuous and stable power supply of the system in case of faults.

[0140] The AI monitoring module uses reinforcement learning and fuzzy control strategies to detect and predict the health status of the circuit in real time, dynamically adjusting the current distribution and load switching strategies before the occurrence of faults, achieving early prevention and rapid response to faults, significantly reducing the fault risk during the operation of the system, and improving the reliability and stability of the photovoltaic system.

[0141] The integrated design of the lightning protection unit effectively protects the system from lightning strikes and voltage surges, providing a high level of safety guarantee for the operation of the photovoltaic system in complex environments.

[0142] The overall technical solution optimizes the current distribution and power supply switching strategies by combining historical data and real-time monitoring data, maximizing the power generation efficiency of the photovoltaic system, flexibly adapting to the dynamic changes in load demand, improving the overall operation efficiency and adaptability of the system, breaking through the limitations of the existing technology, and realizing an intelligent, multi-functional and highly reliable photovoltaic busbar box system in multi-power scenarios, which is particularly suitable for the efficient management requirements of communication machine rooms, minimalist sites and other complex power environments.

[0143] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of various changes or substitutions, and these should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. An AI-driven multi-power intelligent photovoltaic combiner box, characterized in that: The combiner box supports the access of switch power supplies of different voltage levels, including: multiple power access circuits, and each power access circuit works independently without interfering with each other, and each independent power access circuit includes: Photovoltaic access unit, used to access an independent photovoltaic array and transmit DC power to the confluence unit; The confluence unit is used to collect the current of the photovoltaic array and transmit it to the current shunt; A current shunt for monitoring and distributing the current from the busbar unit; The load cut-in unit is used to determine the access and switching of the switching power supply according to the load demand and the health status of the circuit. It supports the situation where any one power supply has power or both power supplies have power. When any power supply fails, it automatically switches to other power supplies. Lightning protection unit, used for integrated lightning protection to prevent the circuit from being affected by lightning strikes or voltage surges; The combiner box also includes an AI monitoring module, which is used to detect the health status of the circuit by combining reinforcement learning with fuzzy control strategies, adjust current distribution and power switching in real time, and adjust fuzzy control rules according to real-time data to predict potential faults and provide fault diagnosis and response.

2. According to claim 1, a multi-power intelligent photovoltaic combiner box based on AI driving is characterized in that: The combiner box also includes: Grounding unit, used to provide electrical grounding protection for the combiner box and all power access circuits; A local load power supply unit is used to provide power support to the local load according to the electric energy output by the loop; The multi-circuit metering unit is used to measure the power of each power access circuit, monitor the voltage, current and power of each circuit in real time, and upload them to the communication unit; The communication unit is used to transmit the data uploaded by the multi-circuit metering unit and the operating status information of the combiner box to the AI ​​monitoring module for data analysis, fault diagnosis and optimization.

3. According to claim 1, the AI-driven multi-power intelligent photovoltaic combiner box is characterized in that: The AI ​​monitoring module includes: A fault detection unit is used to monitor the current, voltage and temperature of each power supply access circuit in real time, determine whether there is a fault, and feed back the fault information to the load switching decision unit; The load switching decision unit is used to intelligently determine whether it is necessary to switch the power supply access circuit to ensure stable power supply to the load based on the fault information provided by the fault detection unit and the real-time load demand. If necessary, it will issue a command to automatically switch to another power supply through the load cutting unit; Current optimization control unit, used to dynamically adjust current distribution according to real-time current and load demand, and optimize the current output of each circuit through fuzzy control strategy; The learning and prediction unit is used to analyze historical operation data and real-time monitoring data through reinforcement learning algorithms, predict potential circuit failures, and generate optimized control strategies to adjust current distribution and load switching decisions in advance.

4. According to claim 3, the AI-driven multi-power intelligent photovoltaic combiner box is characterized in that: The fault detection unit performs fault mode recognition based on the Q learning algorithm, calculates the Q value of each loop, selects the optimal strategy and dynamically updates it. The specific implementation process is as follows: Define the state space S and action space A. The state space S represents the current, voltage and temperature of each circuit, and the action space A represents the fault detection and diagnosis actions. When the fault detection unit is in the current state s t When selecting action a t , determine whether the health status of the circuit is normal; Execute action a t Then the state is transferred from the current state s t Transition to new state s t+1 , and calculate the immediate reward value r based on the action performed t+1 ; If the current, voltage or temperature of the circuit is within the normal range, the instant reward value r t+1 +1 indicates normal operation; if the status is beyond the normal range, such as current overload, voltage abnormality or overtemperature, the instant reward value r t+1 -1 indicates a fault state; Once the instant reward value r is obtained t+1 , then use the update formula of the Q learning algorithm to update the current state s t and the selected action a t The corresponding Q value update formula is: Q(s t ,a t )←Q(s t ,a t )+α[r t+1 +γm a a ′ xQ(s t+1 ,a′)-Q(s t ,a t )]; In the formula, Q(s t ,a t ) means in state s t Next, perform action a t Q value; α is the learning rate; r t+1 is the immediate reward value given according to the health status of the loop; γ is the discount factor; a′ is the next state s t+1 The best move among all the moves; Whenever the state changes or the execution of an action generates a different reward, the Q-learning algorithm dynamically adjusts the Q value so that the fault detection strategy gradually approaches the optimal one.

5. According to claim 3, the AI-driven multi-power intelligent photovoltaic combiner box is characterized in that: The load switching decision unit performs decision optimization based on the multi-armed bandit problem and selects a power switching strategy, specifically including: Each power access circuit is regarded as a lever. Each time you select any power access circuit to switch, it will bring a reward, the reward r t Represents the stability of the load after power switching, r t +1 means the load is stable and the power supply meets the demand. t -1 means the load is unstable and the power supply is insufficient; Based on the current return and historical return, a greedy strategy or ε-greedy strategy is used to select a power access loop for switching, and the return value r is calculated based on the load power supply stability after switching. t ; By recording the power access circuit and the return each time, the data is gradually accumulated and the return value is updated. The return of each power access circuit is calculated. The formula is: In the formula, Q(A i ) indicates power supply access circuit A i The return value; r t (i) Indicates that when selecting power supply access circuit A i When N i Connect the power supply to loop A i The number of times selected; T is the total number of decisions made; The optimal power access loop is selected according to the updated reward value. The power access loop with the highest reward is selected as the first choice for the next power switch. If the rewards of multiple power access loops are the same, the ε-greedy strategy is used to select; Continuously record the feedback of each power switching, update the feedback value of each power access loop, and adjust the selection strategy.

6. The AI-driven multi-power intelligent photovoltaic combiner box according to claim 3 is characterized in that: The current optimization control unit monitors the current, load demand and operating status of each circuit in real time, and evaluates whether the current distribution meets expectations; According to the nonlinear relationship between current and load demand, the fuzzy control strategy is used to dynamically adjust the current distribution to optimize the current output of each circuit; The fuzzy control strategy adjusts the current based on fuzzy rules. When the current of any loop is lower than a preset threshold, the fuzzy control rules are used to determine whether the current output of the loop needs to be increased, and corresponding adjustments are made based on the load requirements of other loops.

7. The AI-driven multi-power intelligent photovoltaic combiner box according to claim 3 is characterized in that: The learning and prediction unit uses a deep reinforcement learning algorithm to predict potential loop failures and generate an optimized control strategy, including the following steps: By analyzing historical failure modes, current, voltage, temperature, load demand, and real-time operation data, a deep reinforcement learning model is built to predict possible future system failures; Through the deep neural network DNN, the potential patterns of load switching, fault occurrence and current distribution are learned. During the training process, Q learning or Actor-Critic algorithm is used to optimize the current distribution and load switching decisions based on historical data and predict the probability of fault occurrence. Through the learned strategies and models, the current distribution and load switching decisions are adjusted in advance. The deep reinforcement learning model is used to predict the potential risk of fault occurrence, adjust the current distribution in advance, and optimize the load switching decision.

8. A working method of a multi-power intelligent photovoltaic combiner box based on AI driving, applied to the multi-power intelligent photovoltaic combiner box based on AI driving according to any one of claims 1 to 7, characterized in that: The method comprises: Each independent photovoltaic array is connected to the combiner box through the photovoltaic access unit. The combiner unit collects the DC power from each photovoltaic array and transmits it to the current shunt for current monitoring and distribution; The current shunt distributes the current according to the current from the confluence unit and monitors the current status of each circuit in real time. During the current distribution process, the AI ​​monitoring module adjusts the current distribution based on real-time data to ensure that the current output of each circuit meets the load requirements and avoids overload or uneven current. According to the load demand and the health status of the circuit, the load cut-in unit intelligently determines the access status of the current power supply, and automatically switches the power supply according to the prediction or fault diagnosis results. If any power supply fails, it will automatically switch to other available power supplies according to the judgment of the AI ​​monitoring module to ensure continuous power supply to the load; The AI ​​monitoring module continuously detects the health status of the system circuit by combining reinforcement learning and fuzzy control strategies, and adjusts the current distribution and power switching strategies in real time. It predicts the occurrence of potential faults by learning historical data and real-time monitoring information, and takes adjustment measures in advance based on the prediction results to reduce the probability of faults and ensure the stable operation of the system. When potential faults or changes in load demand are predicted, the current distribution strategy is dynamically adjusted using fuzzy control rules to optimize the current output of each circuit.

Images