Overload protection system and method for artificial power supply network
By introducing data acquisition, control, overload judgment and protection execution modules into the artificial power network, using neural networks and deepening intensity learning technology for real-time overload detection and protection, the problem of overload detection lag in the existing technology is solved, and the stability and safety of the test are improved.
Patent Information
- Application Number
- CN202510191810.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-21
- Publication Date
- 2025-05-02
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing artificial power networks usually rely on fuses or traditional circuit breakers for protection. These methods react laggingly and cannot accurately detect overload conditions, affecting the stability and safety of the test.
Provides an artificial power network overload protection system, including a data acquisition module, a control module, an overload judgment module and a protection execution module. By detecting input/output current and voltage parameters in real time, load data can be quickly obtained and overload judgment and protection execution can be used using neural networks and deepening intensity learning technology.
It realizes fast and accurate overload judgment and protective execution, avoids equipment damage and service interruption, and significantly improves the stability and security of testing.
Smart Images
Figure CN119921261A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of artificial power supply, and in particular to an artificial power supply network overload protection system and method. Background Art
[0002] Artificial power network, also known as power impedance stabilization network, is an important electromagnetic compatibility test equipment, mainly used to measure the continuous disturbance voltage emitted by the switching power supply under test along the power line to the power grid. The artificial power network provides a stable impedance to the switching power supply under test within the radio frequency range, isolates the switching power supply under test from the high-frequency interference on the power grid, and then couples the interference voltage to the receiver.
[0003] However, the artificial power supply network is located between the input power supply and the product under test. It is necessary to connect the input power supply and input the stable power supply to the product under test to provide a stable impedance to the power line of the product under test within the entire conducted emission measurement frequency range. At the same time, the measuring equipment can stably monitor the conducted interference of the product under test. During the test, the product under test may have abnormal conditions, such as short circuit, overload or power exceeding the tolerance range of the artificial power supply network, which may cause damage to the internal circuit of the artificial power supply network, affect the test accuracy, and even endanger operational safety. However, existing artificial power supply networks usually rely on fuses or traditional circuit breakers for protection. These methods often have a delayed response and cannot accurately detect overload conditions, affecting the stability and safety of the test. Summary of the invention
[0004] The object of the present invention is to provide an artificial power network overload protection system and method to solve the problems described in the background technology.
[0005] The technical solution of the present invention is achieved in this way:
[0006] In one aspect, the present invention provides an artificial power network overload protection system, comprising:
[0007] Data acquisition module, used to detect the input / output current parameters and voltage parameters of AMN in real time;
[0008] A control module, which obtains load data based on the current parameter and the voltage parameter;
[0009] An overload judgment module is used to judge whether the load exceeds a set load threshold according to the current load data;
[0010] The protection execution module is used to compare the load data with the load threshold value, and if the load data is greater than the load threshold value, cut off the power supply or limit the power output.
[0011] A further technical solution is that the data acquisition module includes a first detection unit and a second detection unit, the first detection unit is used to obtain input / output current parameters of the AMN, and the second detection unit is used to obtain input / output voltage parameters of the AMN.
[0012] A further technical solution is that the control module includes an analysis unit, and the analysis unit uses a preset algorithm to analyze the current parameters and voltage parameters to generate the load data.
[0013] A further technical solution is that the analysis unit uses a preset algorithm to analyze the current parameter and the voltage parameter, including:
[0014] The preset algorithm generates the load data by multiplying the current parameter by the voltage parameter. The preset algorithm calculates the load data using the following formula:
[0015] P=VIcosα
[0016] Among them, P is power, V is voltage, I is current, and α is power factor.
[0017] A further technical solution is that the overload judgment module includes a neural network unit and a deepening strength learning unit, the neural network unit is used to adopt the load data for training and predict the overload risk, and the deepening strength learning unit is used to automatically optimize the current limiting and power-off thresholds according to the load characteristics in the AMN.
[0018] A further technical solution is that the neural network unit includes an LSTM prediction model, the LSTM prediction model is trained using the load data, and PyTorch is used for model training to predict overload risks.
[0019] A further technical solution is that the deepening strength learning unit includes a DQN learning algorithm, and the DQN learning algorithm will adopt a reward mechanism according to the different load characteristics in the AMN, complete iterative training for different load characteristics, and automatically optimize the current limiting and power-off thresholds for different load characteristics.
[0020] A further technical solution is that the reward mechanism specifically includes:
[0021] If the system operates normally without triggering current limiting or power outages, a positive reward will be given. If the system frequently triggers current limiting or power outages by mistake, a first negative reward will be given. If the system is damaged due to failure to take protective measures in time, a second negative reward will be given.
[0022] A further technical solution is that the protection execution module includes a first-level execution unit, a second-level execution unit and a third-level execution unit. If the load data is greater than the load threshold and is in a first-level state, the first-level execution unit is triggered; if the load data is greater than the load threshold and is in a second-level state, the second-level execution unit is triggered; and if the load data is greater than the load threshold and is in a third-level state, the third-level execution unit is triggered.
[0023] Another aspect of the present invention provides an artificial power network overload protection method, comprising:
[0024] Step S101, real-time detection of input / output current parameters and voltage parameters of the AMN;
[0025] Step S201, obtaining load data based on the current parameter and the voltage parameter;
[0026] Step S301: judging whether the load exceeds a set load threshold according to the current load data;
[0027] Step S401: based on a comparison value between the load data and the load threshold, if the load data is greater than the load threshold, cutting off power or limiting power output.
[0028] The beneficial effects of the present invention are:
[0029] The present invention provides an artificial power network overload protection system and method. The data acquisition module detects the input / output current parameters and voltage parameters of the AMN in real time to improve the timeliness of data detection. The control module quickly obtains load data based on the current parameters and voltage parameters to facilitate subsequent use. The overload judgment module is used to quickly and accurately judge whether the set load threshold is exceeded according to the current load data to ensure the stability of the judgment process. The protection execution module is used to compare the load data with the load threshold. If the load data is greater than the load threshold, the power supply is cut off or the power output is limited, thereby solving the technical problems that the existing artificial power network usually relies on fuses or traditional circuit breakers for protection, these methods often have a delayed response, cannot accurately detect overload conditions, and affect the stability and safety of the test. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 It is a system module diagram of the present invention;
[0031] Figure 2 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION
[0032] In order to better understand the technical content of the present invention, specific embodiments are provided below, and the present invention is further described in conjunction with the accompanying drawings.
[0033] See also Figure 1 On the one hand, the present invention provides an artificial power network overload protection system, including: a data acquisition module, used to detect the input / output current parameters and voltage parameters of the AMN in real time; a control module, based on the current parameters and voltage parameters, to obtain load data; an overload judgment module, used to judge whether the load data exceeds the set load threshold according to the current load data; a protection execution module, used to compare the load data with the load threshold, and if the load data is greater than the load threshold, cut off the power supply or limit the power output.
[0034] Specifically, the data acquisition module is used to detect the input / output current parameters and voltage parameters of the AMN in real time to improve the timeliness of data detection. The control module quickly obtains the load data based on the current parameters and voltage parameters to facilitate subsequent use. The overload judgment module is used to quickly and accurately judge whether the set load threshold is exceeded according to the current load data to ensure the stability of the judgment process. The protection execution module is used to compare the load data with the load threshold. If the load data is greater than the load threshold, the power supply is cut off or the power output is limited, thereby solving the technical problem that the existing artificial power supply network usually relies on fuses or traditional circuit breakers for protection. These methods often have a delayed response and cannot accurately detect overload conditions, affecting the stability and safety of the test.
[0035] In this embodiment, the data acquisition module includes a first detection unit and a second detection unit, the first detection unit is used to obtain input / output current parameters of the AMN, and the second detection unit is used to obtain input / output voltage parameters of the AMN.
[0036] The first detection unit uses an ACS712 Hall current sensor, which can measure ±30A AC / DC current. It detects the current between the AMN and the DUT and the AMN and the input power supply, and converts the detected current signal into an analog voltage signal, where the signal conversion is performed by the ADC of the STM32 microcontroller. The second detection unit uses a voltage divider circuit to detect the terminal voltage of the AMN to ensure that the high-voltage signal is safely measured by the ADC, and combines a filter circuit to reduce noise interference. The accuracy of the voltage divider circuit and the Hall current sensor for collecting data reaches ±0.1V and ±0.1A, meeting the high-precision requirements of EMC testing. The low-power sensor reduces the power consumption of the data acquisition module to less than 0.5W, ensuring long-term stable operation and stable operation in complex electromagnetic environments.
[0037] In addition, during the entire test process, environmental factors can easily affect the accuracy of the measurement. Therefore, the data acquisition module also uses a DHT22 temperature and humidity sensor to monitor the ambient temperature and humidity around the AMN to avoid false triggering of the protection circuit due to excessive temperature.
[0038] In this embodiment, the control module includes an analysis unit, and the analysis unit uses a preset algorithm to analyze the current parameter and the voltage parameter to generate the load data.
[0039] During the operation of the AMN, the load is directly generated by the current and voltage. Therefore, when calculating the power load, it is necessary to determine the power load by detecting the load of the current and voltage in real time. Specifically, a preset algorithm is used to calculate the load directly generated by the current and voltage. The generated load data not only contains the specific parameters of the current and voltage, but also reflects the real-time operating status of the AMN, providing an important decision-making basis for the subsequent overload judgment module.
[0040] Further, the analysis unit uses a preset algorithm to analyze the current parameter and the voltage parameter, including: the preset algorithm generates the load data by multiplying the current parameter and the voltage parameter, and the preset algorithm calculates the load data using the following formula:
[0041] P=VIcosα
[0042] Among them, P is power, V is voltage, I is current, and cosα is power factor.
[0043] The preset algorithm application formula is convenient for multiplying the current parameter and the voltage parameter to obtain the energy consumed by the AMN in actual operation. At the same time, the consumed energy is transmitted to the overload judgment module in the form of a digital signal, and the result of the transmission can be quickly further analyzed. For example, through the specific voltage V=220V, current I=5A, power factor cosα=0.85 data, the actual load of the equipment can be calculated to be 935W. The use of such accurate load data not only helps to monitor the operating status of the system in real time, but also provides key decision-making basis for the subsequent overload judgment module.
[0044] In this embodiment, the overload judgment module includes a neural network unit and a deepening strength learning unit. The neural network unit is used to use the load data for training and predict the overload risk. The deepening strength learning unit is used to automatically optimize the current limiting and power-off thresholds according to the load characteristics in the AMN.
[0045] The overload judgment module ensures the safety and stability of the system through the collaborative work of the neural network unit and the in-depth strength learning unit. First, the neural network unit collects and analyzes historical load data, monitors the current load status in real time, predicts possible overload risks and issues early warnings. Next, the in-depth strength learning unit dynamically adjusts the current limiting and power-off thresholds based on the current and historical load characteristics, and optimizes the protection mechanism to reduce false triggering and improve the accuracy of the response. The neural network unit and the in-depth strength learning unit work together to enable the system to identify potential problems in advance and automatically optimize the protection strategy, thereby ensuring that the AMN operates in an efficient and safe state to avoid equipment damage and service interruptions.
[0046] Preferably, the neural network unit includes an LSTM prediction model, which is trained using the load data and PyTorch for model training to predict overload risks.
[0047] First, historical load data, including long-term monitoring data such as current, voltage, power, and temperature, are collected, and whether an overload occurs at each time point is marked. After normalization, these data are divided into time windows of fixed length for input into the LSTM model. Using the PyTorch framework, a neural network model with two layers of LSTM units is designed and optimized through a binary cross entropy loss function. During the training process, the model gradually learns how to identify potential overload patterns from historical data and predict possible overload risks in the future. After multiple iterations, the model can achieve high-precision prediction effects on the test set, reduce the number of false alarms, and ensure that the system can make accurate judgments even when the load fluctuates greatly. Ultimately, this trained LSTM model can monitor the current load status in real time, issue early warning signals in advance, and take preventive measures through the protection execution module, such as adjusting the current limiting threshold or starting the backup power supply, thereby effectively avoiding equipment damage and service interruption, and significantly improving the reliability and safety of the entire system.
[0048] Preferably, the deepening intensity learning unit includes a DQN learning algorithm, and the DQN learning algorithm will adopt a reward mechanism according to the different load characteristics in the AMN, complete iterative training for different load characteristics, and automatically optimize the current limiting and power-off thresholds for different load characteristics.
[0049] The DQN learning algorithm analyzes different load characteristics in the AMN (such as current, voltage, and power time series data) and performs iterative training according to a preset reward mechanism. The reward mechanism includes: if the system operates normally without triggering current limiting or power outages, a positive reward is given; if the current limiting or power outages are frequently triggered by mistake, a first negative reward is given; if the equipment is damaged due to failure to take protective measures in time, a second negative reward is given. By continuously adjusting the current limiting and power outage thresholds and obtaining corresponding reward scores based on the actual operating results, the DQN algorithm can gradually learn the optimal strategy. This process enables the system to dynamically optimize the current limiting and power outage thresholds under different load conditions, thereby improving the safety and stability of the system, ensuring efficient operation while effectively preventing overload risks and reducing misoperation.
[0050] Furthermore, the reward mechanism specifically includes: if the system operates normally without triggering current limiting or power outages, a positive reward is given; if the system frequently triggers current limiting or power outages by mistake, a first negative reward is given; if the system is damaged due to failure to take protective measures in time, a second negative reward is given.
[0051] For example, if the system operates normally without triggering current limiting or power outage, a positive reward (e.g., +10 points) is given to encourage this ideal state; if the system frequently triggers current limiting or power outage by mistake, a first negative reward (e.g., -5 points) is given to reduce unnecessary protective measures; if the equipment is damaged due to failure to take protective measures in time, a second negative reward (e.g., -20 points) is given to emphasize the importance of timely protection. Through this reward mechanism, the system can continuously adjust and optimize its current limiting and power outage strategies according to actual operating performance, thereby improving overall safety and reliability.
[0052] In this embodiment, the protection execution module includes a first-level execution unit, a second-level execution unit and a third-level execution unit. If the load data is greater than the load threshold and is in a first-level state, the first-level execution unit is triggered; if the load data is greater than the load threshold and is in a second-level state, the second-level execution unit is triggered; if the load data is greater than the load threshold and is in a third-level state, the third-level execution unit is triggered.
[0053] Specifically, when the load data exceeds the preset load threshold, the system will enter different states according to the degree of overload and trigger the corresponding execution unit: if the load data is slightly greater than the load threshold (level one), the first-level execution unit is triggered, such as triggering an alarm and recording data, and the load is switched to a redundant circuit; if the load data is significantly greater than the load threshold (level two), the second-level execution unit is triggered, such as adjusting the output current to a safe range through PWM; if the load data far exceeds the load threshold (level three), the third-level execution unit is triggered, such as immediately executing a full power-off protection to prevent damage to the equipment due to overload. Through this hierarchical protection mechanism, the system can dynamically adjust the protection strategy according to the actual load conditions, thereby effectively preventing overload risks and reducing misoperation, ensuring the safety and reliability of the EMC test process.
[0054] See also Figure 2 On the other hand, the present invention provides a method for an artificial power network overload protection system, including: step S101, real-time detection of input / output current parameters and voltage parameters of the AMN; step S201, based on the current parameters and voltage parameters, obtaining load data; step S301, judging whether it exceeds a set load threshold according to the current load data; step S401, based on a comparison value between the load data and the load threshold, if the load data is greater than the load threshold, cutting off the power supply or limiting the power output.
[0055] It should be noted that, if the load data is less than the load threshold value according to the load data and load threshold comparison value, the AMN works normally to ensure that the input power cannot superimpose the DUT emission interference and ensure that the EMC test is carried out normally.
[0056] Specifically, the input / output current parameters and voltage parameters of the AMN are detected in real time through the data acquisition module to improve the timeliness of data detection. The control module quickly obtains the load data based on the current parameters and voltage parameters to facilitate subsequent use. The overload judgment module is used to quickly and accurately judge whether the set load threshold is exceeded according to the current load data to ensure the stability of the judgment process. The protection execution module is used to compare the load data with the load threshold. If the load data is greater than the load threshold, the power supply is cut off or the power output is limited, thereby solving the technical problem that the existing artificial power supply network usually relies on fuses or traditional circuit breakers for protection. These methods often have a delayed response and cannot accurately detect overload conditions, affecting the stability and safety of the test.
[0057] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. An artificial power network overload protection system, characterized in that: include: Data acquisition module, used to detect the input / output current parameters and voltage parameters of AMN in real time; A control module, which obtains load data based on the current parameter and the voltage parameter; An overload judgment module is used to judge whether the load exceeds a set load threshold according to the current load data; The protection execution module is used to compare the load data with the load threshold value, and if the load data is greater than the load threshold value, cut off the power supply or limit the power output.
2. The artificial power network overload protection system according to claim 1, characterized in that: The data acquisition module includes a first detection unit and a second detection unit, the first detection unit is used to obtain input / output current parameters of the AMN, and the second detection unit is used to obtain input / output voltage parameters of the AMN.
3. The artificial power network overload protection system according to claim 1, characterized in that: The control module includes an analysis unit, which uses a preset algorithm to analyze the current parameter and the voltage parameter to generate the load data.
4. The artificial power network overload protection system according to claim 3, characterized in that: The analyzing unit uses a preset algorithm to analyze the current parameter and the voltage parameter, including: The preset algorithm generates the load data by multiplying the current parameter by the voltage parameter. The preset algorithm calculates the load data using the following formula: P=VIcosα Among them, P is power, V is voltage, I is current, and α is power factor.
5. The artificial power network overload protection system according to claim 1, characterized in that: The overload judgment module includes a neural network unit and a deepening strength learning unit. The neural network unit is used to adopt the load data for training and predict the overload risk. The deepening strength learning unit is used to automatically optimize the current limiting and power-off thresholds according to the load characteristics in the AMN.
6. An artificial power network overload protection system according to claim 5, characterized in that: The neural network unit includes an LSTM prediction model, which is trained using the load data and PyTorch for model training to predict overload risks.
7. The artificial power network overload protection system according to claim 5, characterized in that: The deepening intensity learning unit includes a DQN learning algorithm. The DQN learning algorithm will adopt a reward mechanism according to the different load characteristics in the AMN to complete iterative training for different load characteristics and automatically optimize the current limiting and power-off thresholds for different load characteristics.
8. An artificial power network overload protection system according to claim 7, characterized in that: The reward mechanism specifically includes: If the system operates normally without triggering current limiting or power outages, a positive reward will be given. If the system frequently triggers current limiting or power outages by mistake, a first negative reward will be given. If the system is damaged due to failure to take protective measures in time, a second negative reward will be given.
9. The artificial power network overload protection system according to claim 1, characterized in that: The protection execution module includes a first-level execution unit, a second-level execution unit and a third-level execution unit. If the load data is greater than the load threshold value, which is a first-level state, the first-level execution unit is triggered. If the load data is greater than the load threshold value, which is a second-level state, the second-level execution unit is triggered. If the load data is greater than the load threshold value, which is a third-level state, the third-level execution unit is triggered.
10. A method according to any one of claims 1 to 6 for an artificial power network overload protection system, characterized in that: include: Step S101, real-time detection of input / output current parameters and voltage parameters of the AMN; Step S201, obtaining load data based on the current parameter and the voltage parameter; Step S301: judging whether the load exceeds a set load threshold according to the current load data; Step S401: based on a comparison value between the load data and the load threshold, if the load data is greater than the load threshold, cutting off power or limiting power output.