Power supply control cabinet system based on precise control and use method thereof

Abstract

The invention provides a power supply control cabinet system based on precise control and a use method thereof, and relates to the technical field of power supply control cabinets, and the system comprises a power supply module which is used for converting an input AC power supply into a stable DC power supply; the control module is used for receiving an operation instruction and processing data; the sampling module is used for sampling voltage, current and temperature parameters of the power supply control cabinet in real time; the input and output module is used for receiving a signal from external setting; the communication module is used for realizing data communication between the power supply control cabinet system and external equipment; the display and operation panel module is used for providing an interface for interaction between a user and the power supply control cabinet; according to the power supply control cabinet system, the long-short-term memory neural network and the reinforcement learning algorithm are fused, so that the probability that complex faults cannot be accurately diagnosed and predicted is reduced, and the reliability and the fault-tolerant capability of the whole power supply control cabinet system are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply control cabinets, and in particular to a power supply control cabinet system based on precision control and a use method thereof. Background Art

[0002] A power control cabinet is a device used to centrally manage and control power. It usually contains various electrical components, such as circuit breakers, contactors, relays, timers, overload protection devices, indicator lights, buttons, etc., which are used to distribute, control and protect power. Power control cabinets are widely used in industrial, commercial and residential buildings and other fields, and are an indispensable part of modern electrical systems.

[0003] When the existing precision-controlled power control cabinet system is in use, various sensors are deployed inside the power control cabinet to monitor the environmental parameters and electrical parameters inside the control cabinet in real time. The data acquisition system is responsible for receiving the data from the sensors and performing preliminary processing and conversion for subsequent storage and analysis. Then, a professional database system is used to store the operating data of the power control cabinet. These database systems usually have efficient data storage and retrieval capabilities. Secondly, data processing and analysis are performed to clean and preprocess the collected data, remove abnormal data, and fill in missing data. Then, statistical analysis methods are used to conduct in-depth analysis of the data, and a model is established to predict the failure risk of the equipment and discover potential problems. According to the data analysis results, the potential failure risk of the equipment is discovered in advance, and corresponding maintenance measures are taken to prevent it. In addition, through data analysis, the operating parameters of the equipment are optimized. The operating environment of the power control cabinet system is complex and changeable, and there may be multiple factors coupled with each other. The existing analysis methods may not be able to handle these complex relationships, resulting in the inability to accurately diagnose and predict complex faults, reducing the reliability and fault tolerance of the entire power control cabinet system.

[0004] Therefore, it is necessary to provide a new power supply control cabinet system based on precise control and its use method to solve the above technical problems. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a power control cabinet system based on precision control and a method of using the same.

[0006] A precisely controlled power supply control cabinet system provided by the present invention includes a power supply module for converting an input AC power supply into a stable DC power supply;

[0007] The control module is responsible for receiving operation instructions, processing data, and adjusting the system operation state according to the set parameters, wherein the control module includes a main control unit, an AI algorithm unit, an instruction generation unit and a communication unit. The main control unit is used to receive the real-time data of voltage, current and temperature transmitted by the sampling module for calculation and analysis. The AI ​​algorithm unit is used to fuse the long short-term memory neural network with the reinforcement learning algorithm, and learn and analyze the long-term operation data of the power control cabinet through the fused algorithm, and automatically adjust the control strategy according to the real-time monitored voltage, current and temperature real-time data. The instruction generation unit is used to generate mode switching instructions according to different work requirements and scenarios. The communication unit is used to transmit the real-time data of voltage, current and temperature obtained by the sampling module in the power control cabinet;

[0008] The sampling module is used to sample the voltage, current and temperature parameters of the power control cabinet in real time and transmit the data to the control module;

[0009] Input and output modules, the input module is used to receive signals from external devices, while the output module is used to send instructions to the control module;

[0010] Communication module, used to realize data communication between the power control cabinet system and external equipment;

[0011] The display and operation panel module is used to provide an interface for users to interact with the power control cabinet and can display the operating status of the equipment in real time;

[0012] The protection module is used to cut off the power supply or take other protective measures in time when an abnormal situation occurs in the system.

[0013] Preferably, the AI ​​algorithm unit specifically comprises the following steps when fusing the long short-term memory neural network with the reinforcement learning algorithm:

[0014] S1. Problem definition and environment modeling: determine the system goal and define the state space. At the same time, determine the executable action set based on the system control method, and then set the reward function.

[0015] S2. Construction and training of long short-term memory neural network. Collect historical operation data of the power control cabinet system and pre-process the data. Then, according to the complexity of the power control cabinet system and the data characteristics, confirm the number of layers and neuron data of the long short-term memory neural network. Finally, use the pre-processed long short-term memory neural network for training.

[0016] S3, reinforcement learning algorithm selection and design, according to the characteristics and requirements of the problem, select the reinforcement learning algorithm, and then carry out the interaction design between the intelligent agent and the environment, define the interaction process between the intelligent agent and the power control cabinet system environment, the intelligent agent selects actions according to the current state, and the environment updates the state according to the action of the intelligent agent and returns the reward;

[0017] S4. Fusion architecture design: First, the trained LSTM neural network is used as a feature extractor to extract features from the input sequence data, and the output of the LSTM neural network is used as the input of the reinforcement learning agent. During the training process, the loss of the LSTM neural network and the loss of reinforcement learning can be combined by designing a joint loss function, and the parameters of the two parts can be updated simultaneously using the gradient descent method, and then the fused overall architecture is built;

[0018] S5, training and optimization, let the fused intelligent agent and the power control cabinet system environment conduct a lot of interactive training. At each time step, the intelligent agent selects an action based on the current state, and the environment updates the state and returns the reward based on the action. At the same time, during the training process, the hyperparameters of the long short-term memory neural network and the reinforcement learning algorithm need to be adjusted, and the performance of the fusion model should be evaluated regularly;

[0019] S6. Deployment and application: deploy the trained fusion model to the control module of the power control cabinet system, and during actual use, monitor the operating status of the power control cabinet system in real time and collect new data.

[0020] Preferably, the main control unit includes a core processor and a storage circuit, and the core processor is used for complex logic control and high-speed data processing.

[0021] Preferably, the instruction generation unit is responsible for generating accurate and effective instructions according to the operating status and control requirements of the system.

[0022] Preferably, the communication unit is responsible for building a bridge for information interaction between various components within the system and between the system and external devices, and the communication unit includes a serial communication interface, Ethernet communication and CAN bus communication. The serial communication interface is used for short-distance, point-to-point communication, Ethernet communication is used for application scenarios with high requirements for data transmission rate and real-time performance, and CAN bus communication is used for power control cabinet systems with high requirements for communication reliability and real-time performance.

[0023] Preferably, the power module includes a rectifier, a filter, a voltage stabilizer and a protection circuit. The rectifier is used to convert alternating current into direct current. The filter is used to smooth the rectified direct current to reduce voltage and current fluctuations. The voltage stabilizer is used to maintain the stability of the output voltage under various conditions and provide a stable and reliable power supply for the electronic components and equipment in the power control cabinet. The protection circuit is used to take protective measures in time when an abnormal situation occurs in the power module to prevent damage to the power module itself and other equipment and circuits connected to it.

[0024] Preferably, the various monitored data include voltage, current, temperature, and power parameters.

[0025] A method for using a precisely controlled power control cabinet system provided in a second aspect of the present invention is applicable to the precisely controlled power control cabinet system, and is characterized in that it comprises the following steps:

[0026] S10, system initialization, comprehensive inspection of various hardware devices in the power control cabinet, start the control system software, initialize the system parameters, and calibrate various sensors installed in the power control cabinet;

[0027] S20, data acquisition, the data acquisition system reads the environmental parameters and electrical parameters inside the power control cabinet from each sensor according to the set sampling period, and transmits the collected data to the data storage module through the communication interface for storage;

[0028] S30, data analysis and processing, cleaning the collected data, removing outliers, noise and erroneous data, filling in missing data, and extracting useful features from the cleaned and preprocessed data;

[0029] S40, formulate corresponding control strategies according to the results of data analysis and fault diagnosis, combined with the control objectives and constraints of the system, and then convert the formulated control strategies into specific control instructions, monitor the execution effect of the control instructions in real time, collect the feedback data of the system through sensors, and compare them with the control objectives;

[0030] S50. Develop a regular inspection plan to conduct comprehensive inspections and maintenance on the power control cabinet system. When a system failure occurs, respond to fault alarms in a timely manner and organize maintenance personnel to troubleshoot and repair the fault.

[0031] Compared with the related art, the power control cabinet system based on precise control and the use method thereof provided by the present invention have the following beneficial effects:

[0032] The present invention determines the system goal by fusing the long short-term memory neural network with the reinforcement learning algorithm, clarifies the precise control goal to be achieved by the power control cabinet system, and defines the state space. At the same time, according to the control mode of the system, the historical operation data of the power control cabinet system is collected, and the number of layers of the long short-term memory neural network, neuron data, and reinforcement learning algorithm selection and design are confirmed according to the complexity of the system and the data characteristics. According to the characteristics and requirements of the problem, the interaction design between the intelligent agent and the environment is carried out, and the interaction process between the intelligent agent and the power control cabinet system environment is defined. The fusion architecture design uses the trained long short-term memory neural network as a feature extractor to extract features from the input sequence data, combines the loss of the long short-term memory neural network with the loss of reinforcement learning, and allows the fused intelligent agent to perform a large amount of interactive training with the power control cabinet system environment. The environment updates the state according to the action and returns the reward. It is necessary to adjust the hyperparameters of the long short-term memory neural network and the reinforcement learning algorithm, and regularly evaluate the performance of the fusion model. The trained fusion model is deployed in the control module of the power control cabinet system. This device reduces the probability of being unable to accurately diagnose and predict complex faults, and reduces the reliability and fault tolerance of the entire power control cabinet system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A structural block diagram of a precisely controlled power supply control cabinet system provided by the present invention;

[0034] Figure 2 A flowchart of the fusion of the long short-term memory neural network and the reinforcement learning algorithm provided by the present invention;

[0035] Figure 3 A flowchart of the long short-term memory neural network provided by the present invention;

[0036] Figure 4 A flowchart of the training process provided by the present invention;

[0037] Figure 5 A flowchart of a method for using a precisely controlled power control cabinet system provided by the present invention. DETAILED DESCRIPTION

[0038] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.

[0039] Please refer to Figure 1 , Figure 2 , Figure 3 , Figure 4 as well as Figure 5 ,in, Figure 1 A structural block diagram of a precisely controlled power supply control cabinet system provided by the present invention; Figure 2A flowchart of the fusion of the long short-term memory neural network and the reinforcement learning algorithm provided by the present invention; Figure 3 A flowchart of the long short-term memory neural network provided by the present invention; Figure 4 A flowchart of the training process provided by the present invention; Figure 5 A flowchart of a method for using a precisely controlled power control cabinet system provided by the present invention.

[0040] Embodiment 1

[0041] In the specific implementation process, Figures 1 to 4 As shown, a precisely controlled power control cabinet system includes a power module for converting input AC power into a stable DC power;

[0042] It should be noted that the power module includes a rectifier, a filter, a voltage stabilizer and a protection circuit. The rectifier is used to convert AC power into DC power. The filter is used to smooth the rectified DC power to reduce voltage and current fluctuations. The voltage stabilizer is used to maintain the stability of the output voltage under various conditions and provide a stable and reliable power supply for the electronic components and equipment in the power control cabinet. The protection circuit is used to take protective measures in time when an abnormal situation occurs in the power module to prevent damage to the power module itself and other equipment and circuits connected to it.

[0043] In addition, the design points of the power module can be divided into: stability design, reliability design, heat dissipation design, electromagnetic compatibility design and maintainability design;

[0044] Stability design, using high-quality electronic components, optimizing circuit layout and wiring, reducing the impact of electromagnetic interference and parasitic parameters, ensuring that the power module can operate stably under different working conditions;

[0045] Reliability design, adding redundant design and protection circuit to improve the reliability and fault resistance of the power module;

[0046] Heat dissipation design: The power module will generate heat during operation, so effective heat dissipation design is required. Heat sinks, fans and other heat dissipation devices can be used to ensure that the temperature of the power module is within a reasonable range and extend the service life of the components.

[0047] Electromagnetic compatibility design: In the design of the power module, the electromagnetic compatibility problem should be considered, and shielding, filtering and other measures should be taken to reduce the electromagnetic interference of the power module to the surrounding environment, while improving its own anti-interference ability;

[0048] Maintainability design: When designing the power module, it is necessary to consider the convenience of maintenance and repair. The modular design makes it easy to disassemble and replace each part, and at the same time provides necessary test points and indicator lights to facilitate fault diagnosis and troubleshooting;

[0049] The control module is responsible for receiving operation instructions, processing data, and adjusting the system operation status according to the set parameters. The control module includes a main control unit, an AI algorithm unit, an instruction generation unit and a communication unit. The main control unit is used to receive the real-time data of voltage, current and temperature from the sampling module for calculation and analysis. The AI ​​algorithm unit is used to fuse the long short-term memory neural network with the reinforcement learning algorithm, and learn and analyze the long-term operation data of the power control cabinet through the fused algorithm. At the same time, according to various data monitored in real time, the control strategy is automatically adjusted to achieve adaptive control. The instruction generation unit is used to generate mode switching instructions according to different work requirements and scenarios. The communication unit is used to transmit various data obtained by the sampling module in the power control cabinet;

[0050] It should be noted that the Long Short-Term Memory (LSTM) neural network is a special recurrent neural network (RNN) that is specifically designed to solve the gradient vanishing and gradient exploding problems encountered by traditional recurrent neural networks when processing long sequence data.

[0051] The various data monitored include voltage, current, temperature, and power parameters;

[0052] The AI ​​algorithm unit integrates the long short-term memory neural network with the reinforcement learning algorithm, including the following steps:

[0053] S1. Problem definition and environment modeling, determine the system goals, clarify the precise control goals to be achieved by the power control cabinet system, and define the state space. At the same time, according to the control method of the system, determine the executable action set, and then set the reward function;

[0054] It should be noted that the core goal of the power control cabinet is to achieve precise control of the power supply, ensure stable and efficient output and meet the needs of various loads;

[0055] Determine the key variables. There are many variables that need to be paid attention to, including the voltage fluctuation and frequency change of the mains on the input side, the voltage, current, power on the output side, and the temperature and heat dissipation of the components inside the power supply;

[0056] Environmental modeling, state space definition: The state space should contain all key information reflecting the real-time status of the power control cabinet. In addition to the voltage, current, power and other data at the current moment, the long short-term memory neural network is used to process sequence data, and the sequence of these parameter changes over a period of time (such as the past 10 sampling cycles) is included;

[0057] Action space definition: The action space is determined according to the control method of the power control cabinet. Common actions include adjusting the transformer ratio, which can be set to adjust with a fixed step size within a certain range to control the on and off of the switch elements.

[0058] Reward function design: The reward function is the key to guiding the agent to learn the optimal strategy. If the goal is to stabilize the output voltage, a positive reward is given when the output voltage is close to the set value and the fluctuation is within the allowable range, such as 10 points for every minute of stability. If the voltage exceeds the allowable range, a negative reward is given according to the degree of deviation, such as 5 points for every 0.1% deviation. At the same time, considering the system energy consumption, additional rewards are given when energy consumption is reduced to encourage the agent to optimize energy utilization while ensuring the quality of power supply;

[0059] S2. Construction and training of long short-term memory neural network. Collect historical operation data of the power control cabinet system and pre-process the data. Then, according to the complexity of the system and the characteristics of the data, confirm the number of layers and neuron data of the long short-term memory neural network. Finally, use the pre-processed long short-term memory neural network for training.

[0060] It should be noted that the steps for constructing a long short-term memory neural network are as follows:

[0061] S2.1. Determine the number of network layers and neurons. The number of neurons in the input layer depends on the dimension of the input features. The number of LSTM layers and the number of neurons in each layer need to be adjusted according to the complexity of the problem. The number of neurons in the output layer depends on the prediction or decision-making goal.

[0062] S2.2. Design the LSTM unit structure. The LSTM unit is the core of the LSTM network. It contains the input gate, forget gate, output gate and cell state. When constructing the LSTM network, it is necessary to select a suitable activation function to implement these gating mechanisms.

[0063] S2.3, network architecture construction, using deep learning framework to build LSTM network;

[0064] S2.4, Long Short-Term Memory Neural Network Training, firstly, data preparation is performed, and historical operation data is collected from the power control cabinet system, including input features (such as voltage, current) and corresponding target values ​​(such as future voltage prediction value, working status label), and then data preprocessing is performed, the collected data is cleaned, outliers and missing values ​​are removed, and normalization is performed to scale the data to a suitable range. Secondly, a suitable loss function is selected according to the specific task, and then the LSTM model is trained through PyTorch. Finally, the trained model is evaluated using the validation set, and the loss function value and other evaluation indicators are observed;

[0065] S3, reinforcement learning algorithm selection and design, according to the characteristics and requirements of the problem, select the reinforcement learning algorithm, and then carry out the interaction design between the intelligent agent and the environment, define the interaction process between the intelligent agent and the power control cabinet system environment, the intelligent agent selects actions according to the current state, and the environment updates the state according to the action of the intelligent agent and returns the reward;

[0066] It should be noted that the reinforcement learning algorithm design uses the output after LSTM processing as the partial state input of the reinforcement learning agent. LSTM can extract and memorize the historical data of the power control cabinet system to obtain a feature vector that can reflect the dynamic changes of the system;

[0067] Action design, design the action space according to the actual control requirements of the power control cabinet system. For continuous action space, the action can be represented as a vector, and each dimension corresponds to an adjustable parameter. For discrete action space, the action can be encoded with integers;

[0068] Reward function design,The reward function design is divided into stability reward, energy saving reward and fault avoidance reward;

[0069] Stability reward encourages the agent to maintain the stability of the power output. If the output voltage fluctuates within a certain range of the set value, a positive reward is given. The greater the fluctuation, the smaller the reward. If it exceeds the allowable range, a negative reward is given. The expression is:

[0070] r stability = -α*|VV set |

[0071] Where V is the actual output voltage, V set is the set voltage, α is a positive weight coefficient;

[0072] The energy-saving function encourages the agent to reduce the system energy consumption while ensuring stable power output. When the system energy consumption is lower than a certain threshold, an additional positive reward is given. The higher the energy consumption, the lower the reward. Its expression is:

[0073] r energy =-β*P

[0074] In the formula, P is the power consumption of the system, and β is a positive weight coefficient;

[0075] Fault avoidance rewards, which give positive rewards when the agent avoids actions that could cause system failure, and larger negative rewards if a failure occurs;

[0076] Algorithm fusion and training: LSTM is integrated with reinforcement learning algorithm. During the training process, the experience replay mechanism is used to improve data utilization and training stability. The experience data (state, action, reward, next state) obtained by the interaction between the agent and the environment is stored in the experience replay buffer, and then a batch of data is randomly sampled for training.

[0077] S4. Fusion architecture design: First, the trained LSTM neural network is used as a feature extractor to extract features from the input sequence data, and the output of the LSTM neural network is used as the input of the reinforcement learning agent. During the training process, the parameters of the LSTM neural network and the reinforcement learning agent are optimized at the same time. The loss of the LSTM neural network and the loss of the reinforcement learning agent can be combined by designing a joint loss function, and the parameters of the two parts can be updated simultaneously using the gradient descent method, and then the fused overall architecture is built;

[0078] It should be noted that the design idea of ​​the serial fusion architecture uses LSTM as the front-end feature extractor to first process the historical data of the power control cabinet system and extract feature representations containing time series information. The reinforcement learning agent makes decisions based on these feature representations to achieve control of the power system. This architecture takes advantage of LSTM's ability to process sequence data and provides richer and more valuable state information for reinforcement learning.

[0079] Specific architecture,LSTM layer receives historical monitoring data of the power control cabinet system, such as time series data of voltage, current, power, etc. in the past period of time. LSTM processes these sequence data through its gating mechanism to capture the long-term dependencies and dynamic change characteristics in the data;

[0080] The reinforcement learning agent uses the feature vector output by LSTM as the state input of the reinforcement learning agent. The agent selects the appropriate action based on the current state and the reinforcement learning algorithm adopted.

[0081] S5, training and optimization, let the fused intelligent agent and the power control cabinet system environment conduct a lot of interactive training. At each time step, the intelligent agent selects an action based on the current state, and the environment updates the state and returns the reward based on the action. At the same time, during the training process, the hyperparameters of the long short-term memory neural network and the reinforcement learning algorithm need to be adjusted, and the performance of the fusion model should be evaluated regularly;

[0082] It should be noted that the training process includes the following steps:

[0083] S5.1. Data collection and preprocessing: A large amount of historical operating data is collected from the power control cabinet system. These data form the basis of training and reflect the operating status of the power system under different working conditions. Then, the collected data is cleaned to remove outliers and missing values; normalization is performed to scale the data to a suitable range;

[0084] S5.2. Model initialization: First, initialize LSTM to determine the network structure of LSTM, including the number of layers, the number of neurons in each layer, etc., and randomly initialize the weight parameters of LSTM. Then, initialize the reinforcement learning agent: according to the selected reinforcement learning algorithm, initialize the parameters of the agent's policy network or value function network.

[0085] S5.3, training loop, input the preprocessed historical data into LSTM, LSTM processes the data, extracts the feature vector containing time series information, the reinforcement learning agent selects an action based on its policy network or value function network according to the current state, applies the selected action to the power control cabinet system, the system responds according to the action, returns a new state and reward signal, and then stores the current state, action, reward and next state as a piece of experience data in the experience playback buffer, and finally randomly samples a batch of experience data from the experience playback buffer, and uses these data to update the model parameters of LSTM and reinforcement learning agent;

[0086] S5.4, training termination condition, set the maximum number of training steps or training rounds, when the set number of steps or rounds is reached, the training is terminated;

[0087] S6, deployment and application, deploy the trained fusion model to the control module of the power control cabinet system, and in actual use, monitor the operating status of the system in real time and collect new data;

[0088] It should be noted that the deployment environment preparation should select appropriate computing devices according to the complexity of the model and the real-time requirements of the system. For small power control cabinet systems or scenarios with low real-time requirements, ordinary industrial computers can be used. For large and complex systems or scenarios that require rapid decision-making, GPU servers may be required to accelerate the reasoning process of the model, and sufficient storage space should be prepared to store model parameters, historical data, and log information generated during operation.

[0089] Software environment: Choose a stable, reliable and well-compatible operating system. Then install the same deep learning framework used in the training process and ensure that its version is compatible with the training environment. Finally, install the corresponding communication protocol stack and middleware according to the communication requirements of the system.

[0090] Model deployment and model conversion: convert the trained LSTM-reinforcement learning fusion model into a format suitable for the deployment environment, load the converted model in the deployment environment, initialize the model parameters, set the model input and output interfaces and related configuration parameters according to the actual situation of the system, and establish communication connections with sensors in the power control cabinet system to obtain system operation data in real time;

[0091] The main control unit includes a core processor and a storage circuit. The core processor is used for complex logic control and high-speed data processing, and the storage circuit is used to ensure that the instructions and data required for system startup and operation are not lost;

[0092] It should be noted that the core processor is the core of the main control unit and is responsible for performing various calculation and control tasks. Common processors include single-chip microcomputers, embedded microprocessors (such as ARM series), and digital signal processors (DSP);

[0093] The instruction generation unit is responsible for generating accurate and effective instructions based on the system's operating status and control requirements to ensure that the power control cabinet can operate stably and reliably;

[0094] It should be noted that the decisions made by the main control unit based on data analysis and strategy formulation are converted into specific executable control instructions;

[0095] The sampling module is used to sample the voltage, current and temperature parameters of the power control cabinet in real time and transmit the data to the control module;

[0096] It should be noted that the sampling module includes a sensor, a signal conditioning circuit, an analog-to-digital converter, and a communication interface;

[0097] The sensor includes a voltage sensor and a current sensor. The voltage sensor is used to measure the voltage value of the power supply system, and the current sensor is used to measure the current size.

[0098] The signal conditioning circuit includes an amplifier, a filter and an isolation circuit. The amplifier is used to amplify the weak signal output by the sensor. The filter is used to remove noise and interference in the signal. Commonly used filters include low-pass filters, high-pass filters and band-pass filters. The isolation circuit is used to achieve electrical isolation between the input signal and the subsequent circuit, thereby improving the anti-interference ability and safety of the system.

[0099] The analog-to-digital converter is used to convert the conditioned analog signal into a digital signal for processing by a computer or microcontroller, and the resolution and sampling rate of the analog-to-digital converter are important indicators that affect sampling accuracy and real-time performance;

[0100] The communication interface is responsible for transmitting the converted digital signal to the control module or other equipment;

[0101] The communication unit is responsible for building a bridge for information interaction between the components within the system and between the system and external devices, so as to ensure that the system can operate accurately, efficiently and stably. The communication unit includes a serial communication interface, Ethernet communication and CAN bus communication. The serial communication interface is used for short-distance, point-to-point communication, Ethernet communication is used for application scenarios with high requirements for data transmission rate and real-time performance, and CAN bus communication is used for power control cabinet systems with high requirements for communication reliability and real-time performance.

[0102] Input and output modules, the input module is used to receive signals from external devices, while the output module is used to send instructions to the control module;

[0103] It should be noted that the input module includes a sensor interface, a signal conditioning circuit, and an analog-to-digital converter (ADC);

[0104] The sensor interface is used to connect various analog sensors, such as voltage sensor, current sensor, temperature sensor, and humidity sensor;

[0105] The signal conditioning circuit is used to amplify, filter, isolate and process the weak analog signal output by the sensor to improve the signal quality and anti-interference ability;

[0106] Analog-to-digital converters are used to convert conditioned analog signals into digital signals for processing by computers or microcontrollers;

[0107] The output module includes a digital-to-analog converter (DAC), a signal amplifier, and an actuator interface;

[0108] The digital-to-analog converter (DAC) is used to convert the digital signal output by the main control unit into an analog signal, usually a voltage or current signal;

[0109] The signal amplifier is used to amplify the weak analog signal output by the digital-to-analog converter;

[0110] The actuator interface is used to connect various analog execution devices;

[0111] Communication module, used to realize data communication between the power control cabinet system and external equipment;

[0112] The display and operation panel module is used to provide an interface for users to interact with the power control cabinet and can display the operating status of the equipment in real time;

[0113] It should be noted that the display and operation panel module includes a display part, an operation part and a communication interface. The display part includes a display screen and a display driving circuit. The display screen can be selected from different types according to system requirements and cost considerations. The display driving circuit is responsible for driving the display screen to display various information and converting the data sent by the main control unit into a signal that can be recognized by the display screen.

[0114] The operation part includes input devices and operation detection circuits. Input devices and common input devices include buttons, knobs, touch screens, etc. Buttons and knobs are suitable for simple operations, such as parameter setting, mode switching, etc. The touch screen provides a more intuitive and convenient operation method. Users can perform various operations by touching icons and buttons on the screen. The operation detection circuit detects the user's operation signal and converts it into a digital signal to send to the main control unit;

[0115] Communication interface, the display and operation panel module needs to communicate with the main control unit to obtain the system's operating data and send user operation instructions;

[0116] The protection module is used to cut off the power supply or take other protective measures in time when an abnormal situation occurs in the system;

[0117] It should be noted that the main functions of the protection module include overcurrent protection, overvoltage protection, undervoltage protection, short circuit protection, overload protection, overheat protection and leakage protection. For overcurrent protection, when the current in the system exceeds the set rated value, the protection module can detect it in time and quickly cut off the circuit to prevent overheating and burning of the equipment due to overcurrent;

[0118] Overvoltage protection: monitor the voltage of the power supply system. Once the voltage exceeds the upper limit allowed, take immediate measures to reduce the voltage or cut off the power supply to prevent high voltage from damaging the equipment.

[0119] Undervoltage protection: when the power supply voltage is lower than the set lower limit, the protection module will be activated to prevent the device from running under low voltage, causing unstable operation or damage;

[0120] Short-circuit protection: when a short-circuit fault occurs, the short-circuit current will increase sharply. The protection module can cut off the circuit in a very short time to prevent the short-circuit current from causing serious damage to the system.

[0121] Overload protection: long-term overload operation will cause the device to heat up, accelerate device aging or even damage;

[0122] Overheat protection: monitor the temperature of key equipment in the power control cabinet. When the temperature exceeds the safe range, take cooling measures or cut off the power supply to prevent the equipment from being damaged due to overheating.

[0123] Leakage protection detects whether there is leakage in the system. Once it is detected that the leakage current exceeds the set value, the power supply will be cut off immediately to ensure the safety of personnel and equipment.

[0124] Embodiment 2

[0125] refer to Figure 5 As shown, a method for using a precisely controlled power control cabinet system provided by the second aspect of the present invention is applicable to a precisely controlled power control cabinet system, and is characterized in that it includes the following steps:

[0126] S10, system initialization, comprehensive inspection of various hardware devices in the power control cabinet, start the control system software, initialize the system parameters, and calibrate various sensors installed in the power control cabinet;

[0127] S20, data acquisition, the data acquisition system reads the environmental parameters and electrical parameters inside the power control cabinet from each sensor according to the set sampling period, and transmits the collected data to the data storage module through the communication interface for storage;

[0128] S30, data analysis and processing, cleaning the collected data, removing outliers, noise and erroneous data, filling in missing data, and extracting useful features from the cleaned and preprocessed data;

[0129] S40, formulate corresponding control strategies according to the results of data analysis and fault diagnosis, combined with the control objectives and constraints of the system, and then convert the formulated control strategies into specific control instructions, monitor the execution effect of the control instructions in real time, collect the feedback data of the system through sensors, and compare them with the control objectives;

[0130] S50. Develop a regular inspection plan to conduct comprehensive inspections and maintenance on the power control cabinet system. When a system failure occurs, respond to fault alarms in a timely manner and organize maintenance personnel to troubleshoot and repair the fault.

[0131] The circuits and controls involved in the present invention are all prior art and will not be described in detail here.

[0132] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A precisely controlled power supply control cabinet system, characterized in that: A power module is included for converting input AC power into a stable DC power; The control module is responsible for receiving operation instructions, processing data, and adjusting the system operation state according to the set parameters, wherein the control module includes a main control unit, an AI algorithm unit, an instruction generation unit and a communication unit. The main control unit is used to receive the real-time data of voltage, current and temperature transmitted by the sampling module for calculation and analysis. The AI ​​algorithm unit is used to fuse the long short-term memory neural network with the reinforcement learning algorithm, and learn and analyze the long-term operation data of the power control cabinet through the fused algorithm, and automatically adjust the control strategy according to the real-time monitored voltage, current and temperature real-time data. The instruction generation unit is used to generate mode switching instructions according to different work requirements and scenarios. The communication unit is used to transmit the real-time data of voltage, current and temperature obtained by the sampling module in the power control cabinet; The sampling module is used to sample the voltage, current and temperature parameters of the power control cabinet in real time and transmit the data to the control module; Input and output modules, the input module is used to receive signals from external devices, while the output module is used to send instructions to the control module; Communication module, used to realize data communication between the power control cabinet system and external equipment; The display and operation panel module is used to provide an interface for users to interact with the power control cabinet and can display the operating status of the equipment in real time; The protection module is used to cut off the power supply or take other protective measures in time when an abnormal situation occurs in the system.

2. The precise control power supply control cabinet system according to claim 1 is characterized in that: The AI ​​algorithm unit specifically includes the following steps when integrating the long short-term memory neural network with the reinforcement learning algorithm: S1. Problem definition and environment modeling, determine the system objectives and define the state space. At the same time, according to the control method of the system, determine the set of executable actions, and then set the reward function; S2. Construction and training of long short-term memory neural network. Collect historical operation data of the power control cabinet system and pre-process the data. Then, according to the complexity of the power control cabinet system and the data characteristics, confirm the number of layers and neuron data of the long short-term memory neural network. Finally, use the pre-processed long short-term memory neural network for training. S3, reinforcement learning algorithm selection and design, according to the characteristics and requirements of the problem, select the reinforcement learning algorithm, and then carry out the interaction design between the intelligent agent and the environment, define the interaction process between the intelligent agent and the power control cabinet system environment, the intelligent agent selects actions according to the current state, and the environment updates the state according to the action of the intelligent agent and returns the reward; S4. Fusion architecture design: First, the trained LSTM neural network is used as a feature extractor to extract features from the input sequence data, and the output of the LSTM neural network is used as the input of the reinforcement learning agent. During the training process, the loss of the LSTM neural network and the loss of reinforcement learning can be combined by designing a joint loss function, and the parameters of the two parts can be updated simultaneously using the gradient descent method, and then the fused overall architecture is built; S5, training and optimization, let the fused intelligent agent and the power control cabinet system environment conduct a lot of interactive training. At each time step, the intelligent agent selects an action based on the current state, and the environment updates the state and returns the reward based on the action. At the same time, during the training process, the hyperparameters of the long short-term memory neural network and the reinforcement learning algorithm need to be adjusted, and the performance of the fusion model should be evaluated regularly; S6. Deployment and application: deploy the trained fusion model to the control module of the power control cabinet system, and during actual use, monitor the operating status of the power control cabinet system in real time and collect new data.

3. The precise control power supply control cabinet system according to claim 2 is characterized in that: The main control unit includes a core processor and a storage circuit. The core processor is used for complex logic control and high-speed data processing.

4. The precise control power supply control cabinet system according to claim 3 is characterized in that: The instruction generation unit is responsible for generating accurate and effective instructions according to the operating status and control requirements of the system.

5. The precise control power supply control cabinet system according to claim 4, characterized in that: The communication unit is responsible for building a bridge for information interaction between the components within the system and between the system and external devices, and the communication unit includes a serial communication interface, Ethernet communication and CAN bus communication. The serial communication interface is used for point-to-point communication between a sending device and a receiving device, Ethernet communication is used for application scenarios with high requirements for data transmission rate and real-time performance, and CAN bus communication is used for power control cabinet systems with high requirements for communication reliability and real-time performance.

6. The precise control power supply control cabinet system according to claim 5, characterized in that: The power module includes a rectifier, a filter, a voltage stabilizer and a protection circuit. The rectifier is used to convert alternating current into direct current. The filter is used to smooth the rectified direct current and reduce voltage and current fluctuations. The voltage stabilizer is used to maintain the stability of the output voltage under various conditions and provide a stable and reliable power supply for the electronic components and equipment in the power control cabinet. The protection circuit is used to take protective measures in time when an abnormal situation occurs in the power module to prevent damage to the power module itself and other equipment and circuits connected to it.

7. A method for using a precisely controlled power supply control cabinet system, applicable to the precisely controlled power supply control cabinet system according to any one of claims 1 to 6, characterized in that: The following steps are involved: S10, system initialization, comprehensive inspection of various hardware devices in the power control cabinet, start the control system software, initialize the system parameters, and calibrate various sensors installed in the power control cabinet; S20, data acquisition, the data acquisition system reads the environmental parameters and electrical parameters inside the power control cabinet from each sensor according to the set sampling period, and transmits the collected data to the data storage module through the communication interface for storage; S30, data analysis and processing, cleaning the collected data, removing outliers, noise and erroneous data, filling in missing data, and extracting useful features from the cleaned and preprocessed data; S40, formulate corresponding control strategies according to the results of data analysis and fault diagnosis, combined with the control objectives and constraints of the system, and then convert the formulated control strategies into specific control instructions, monitor the execution effect of the control instructions in real time, collect the feedback data of the system through sensors, and compare them with the control objectives; S50. Develop a regular inspection plan to conduct comprehensive inspections and maintenance on the power control cabinet system. When a system failure occurs, respond to fault alarms in a timely manner and organize maintenance personnel to troubleshoot and repair the fault.

Images