Optimization design method and system for hybrid direct current circuit breaker based on machine learning

Abstract

The invention discloses a hybrid DC circuit breaker optimization design method and system based on machine learning, and belongs to the technical field of low-voltage power distribution. According to the method, feature vectors are constructed based on mechanical circuit breaker arc waveforms, circuit feature parameters and product parameters under different short-circuit working conditions, and an LSTM model introducing an attention mechanism is adopted to predict the arc waveforms; key parameters such as power electronic device switching loss and energy consumption branch working time are extracted, and the quantitative relation between the on-off time and device characteristics is established; by taking the minimum breaking time as an optimization target, combining nonlinear constraints such as temperature rise, withstand voltage and a latching effect, performing multi-parameter global optimization by utilizing a genetic algorithm, outputting optimal design parameters, and determining a design scheme of the hybrid circuit breaker according to the optimal design parameters, the development cost of the hybrid circuit breaker is reduced, the development time is shortened, and the development efficiency is improved. And a high-performance hybrid circuit breaker can be formed.

Description

Technical Field

[0001] The present invention relates to the field of low - voltage power distribution, and specifically to an optimized design method and system of a hybrid DC circuit breaker based on machine learning. Background Art

[0002] With the rapid development of low - voltage DC systems such as photovoltaic, energy storage, and electric vehicles, people reduce power transmission losses by increasing the voltage level. The voltage in the low - voltage field is gradually approaching the 1500V upper limit specified by IEC. Therefore, requirements for protection devices such as miniaturization, easy installation, and fast breaking are put forward. Compared with traditional AC systems, DC systems lack natural zero - crossing points and have small line impedance, resulting in a fast rising speed of fault current, which increases the difficulty of protection. Therefore, researching the breaking protection technology of low - voltage DC systems is very important for their development.

[0003] As a key protection device, circuit breakers can be classified into mechanical circuit breakers, solid - state circuit breakers, and hybrid DC circuit breakers according to their structures. Mechanical circuit breakers are the most mature, and can be divided into thermal - magnetic and electronic circuit breakers according to their operating principles. Their arc - extinguishing methods mainly include increasing arc - extinguishing grids, stretching arcs, and adding permanent magnets. However, with the development of DC systems, traditional arc - extinguishing methods are difficult to quickly clear faults, and the arc - extinguishing time is usually dozens of milliseconds. To meet the requirements of high - voltage levels, it is necessary to increase the volume of the arc - extinguishing chamber, gas - generating materials, and multi - stage series connection, which puts higher requirements on volume and operation consistency. With the progress of semiconductor technology, solid - state circuit breakers have received wide attention. Solid - state circuit breakers use fast detection and control modules to quickly clear short - circuit currents, and the fault - clearing time can be shortened to the microsecond level. Under normal operating conditions, rated current is conducted by power electronic devices, and during faults, the devices are turned off. Although they have the characteristics of arc - less breaking, their conduction losses are large, and higher requirements are put forward for heat - dissipation design.

[0004] The hybrid breaking technology developed by combining mechanical circuit breakers and solid - state circuit breakers is considered to be one of the important development directions in the future low - voltage DC field. This solution has the advantages of low conduction loss of mechanical circuit breakers and fast breaking speed of solid - state circuit breakers while making full use of the current - limiting function of traditional circuit breakers. However, existing design methods of hybrid DC circuit breakers often ignore the influence of arc parameters and power - electronic device parameters on current commutation. In practical engineering applications, different short - circuit conditions and mechanical circuit breaker products will affect the arc behavior during the breaking process, thus affecting the commutation efficiency. In addition, characteristics such as the switching speed, voltage withstand capacity, and conduction loss of power - electronic devices will further affect the success of current commutation. Therefore, how to comprehensively consider the influence of the above parameters has become a difficulty in the optimized design of hybrid DC circuit breakers at present. Summary of the Invention

[0005] Aiming at the problem that the design of the hybrid DC circuit breaker does not consider the influence of arc behavior and the parameters of power electronic devices, an optimization design method and system of the hybrid DC circuit breaker based on machine learning are proposed, which predict the arc waveform during the opening process of the circuit breaker, and then optimize the design of the hybrid circuit breaker.

[0006] The present invention is realized by the following technical solutions: An optimization design method of a hybrid DC circuit breaker based on machine learning, comprising the following steps: Step 1, according to the arc waveforms during the opening process of the mechanical circuit breaker under different short-circuit conditions, and combining the circuit characteristic parameters and product parameters corresponding to the working conditions to construct a feature vector, using the feature vector to train the prediction model, and the trained prediction model outputs the arc waveform of the mechanical circuit breaker; Step 2, obtain the key parameters affecting the opening characteristics of the hybrid circuit breaker, determine the process loss of the power electronic device during the switching stage according to the key parameters; according to the working time of the energy-consuming branch of the hybrid circuit breaker, determine the minimum opening time of the hybrid circuit breaker according to the working time; Step 3, establish an optimization model based on the genetic algorithm according to the arc waveform output by the prediction model, the key parameters affecting the opening characteristics of the hybrid circuit breaker, the process loss of the power electronic device during the switching stage, and the opening time of the hybrid circuit breaker, use the opening time of the hybrid circuit breaker as the optimization objective function, and add non-linear constraints based on the device failure principle to the optimization model; Step 4, taking the minimum opening time as the optimization objective, calculate the fitness of the individuals in the population of the optimization model, and perform selection, crossover, and mutation operations to generate the next generation of population. Determine whether the convergence condition is satisfied according to the next generation of population. When the convergence condition is satisfied, output the optimization parameters, and determine the design scheme of the hybrid circuit breaker according to the optimization parameters.

[0007] Preferably, the determination method of the circuit characteristic parameters in step 1 is as follows: According to the arc waveform and combining with the equivalent circuit during the short-circuit opening process, determine the circuit characteristic parameters under each short-circuit condition.

[0008] Preferably, the equivalent circuit is as follows:

[0009] In the formula, i f is the short-circuit current, R 0 is the line resistance, L 0 is the line inductance, u arc is the arc voltage, V s is the system power supply voltage.

[0010] Preferably, the prediction model described in step 1 is an LSTM model, and an attention mechanism is introduced into the LSTM model to learn the time series data of the arc waveform.

[0011] Preferably, the key parameters described in step 2 are the key parameters affecting the breaking characteristics in the hybrid circuit breaker, and the process losses of the power electronic devices in the switching stage are determined according to the key parameters. The process losses are as follows:

[0012]

[0013]

[0014]

[0015] In the formula: E on is the turn-on loss, t on is the turn-on time, P on is the turn-on power, E d is the conduction loss, t d is the conduction time, P d is the conduction power, E off is the turn-off loss, t off is the turn-off time, P off is the turn-off power, E IGBT is the total loss of the switching process.

[0016] Preferably, the method for determining the working time of the energy-consuming branch of the hybrid circuit breaker in step 2 is as follows:

[0017] In the formula, t mov is the energy-consuming time of this branch, V clamp is the clamping voltage value of the metal oxide varistor; The method for determining the minimum breaking time of the hybrid circuit breaker according to the working time is as follows: .

[0018] Preferably, the non-linear constraints based on the device failure principle in step 3 include the temperature rise, the latching effect and the withstand voltage of the power electronic device.

[0019] Preferably, the calculation method of the temperature rise is as follows: The 4th-order Foster thermal network model is used to evaluate the temperature rise during the transient process of power electronic devices. The temperature rise calculation formula is as follows:

[0020]

[0021]

[0022] In the formula, Z th is the thermal impedance of the selected device; R i is the i-th order thermal resistance; t i is the thermal time constant; C i is the i-th order heat capacity of the device; T j is the junction temperature of the device; T e is the ambient temperature.

[0023] A hybrid DC circuit breaker optimization design system based on machine learning includes: An arc prediction module, which is used to construct a feature vector according to the arc waveform during the opening process of the mechanical circuit breaker under different short-circuit conditions, combined with the circuit characteristic parameters and product parameters of the corresponding conditions, and train the prediction model with the feature vector. The trained prediction model outputs the arc waveform of the mechanical circuit breaker; A joint parameter acquisition module, which is used to obtain the key parameters affecting the opening characteristics of the hybrid circuit breaker, determine the process loss of the power electronic device during the switching stage according to the key parameters; determine the minimum opening time of the hybrid circuit breaker according to the working time of the energy-consuming branch of the hybrid circuit breaker; A genetic module, which is used to establish an optimization model based on the genetic algorithm according to the arc waveform output by the prediction model, the key parameters affecting the opening characteristics of the hybrid circuit breaker, the process loss of the power electronic device during the switching stage, and the opening time of the hybrid circuit breaker. Taking the opening time of the hybrid circuit breaker as the optimization objective function, add non-linear constraints based on the device failure principle to the optimization model; An optimization module, which is used to take the minimum opening time as the optimization objective, calculate the fitness of the individuals in the population of the optimization model, and perform selection, crossover, and mutation operations to generate the next generation of population. Determine whether the convergence condition is satisfied according to the next generation of population. When the convergence condition is satisfied, output the optimization parameters, and determine the design scheme of the hybrid circuit breaker according to the optimization parameters.

[0024] An electronic device includes: A memory, which is used to store computer programs; A processor, which is configured to implement the steps of the machine learning-based hybrid DC circuit breaker optimization design method when executing the computer program.

[0025] Compared with the prior art, the present invention has the following beneficial technical effects: For a machine learning-based hybrid DC circuit breaker optimization design method of the present application, firstly, machine learning technology is used to predict the arc waveforms of mechanical circuit breakers under different short-circuit conditions. By constructing a feature vector containing circuit characteristic parameters and product parameters, and using an LSTM model combined with an attention mechanism for training, the temporal characteristics and non-linear relationships of arc waveforms can be effectively captured. Secondly, key parameters of power electronic devices (such as turn-on loss, turn-off loss, etc.) are extracted, a quantitative relationship between the loss of power electronic devices and the opening time is established, and combined with the working time of the energy-consuming branch, the minimum opening time of the hybrid circuit breaker is accurately deduced. This process directly correlates the dynamic characteristics of power electronic devices with the overall performance of the circuit breaker, avoiding the performance bottleneck caused by separate parameter analysis in traditional designs. The genetic algorithm is introduced as an optimization framework, with the minimum opening time as the objective function, and at the same time incorporating non-linear constraint conditions such as temperature rise, hold effect, and withstand voltage, to achieve multi-parameter collaborative optimization. The opening device is designed according to the optimized parameters to maximize the overall opening performance of the hybrid circuit breaker.

[0026] The present application also proposes a machine learning-based hybrid DC circuit breaker optimization design system, an electronic device, and a computer storage medium, which have all the advantages of the above-mentioned machine learning-based hybrid DC circuit breaker optimization design method. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0028] Figure 1 It is a schematic diagram of the machine learning-based hybrid DC circuit breaker optimization design method of the present invention; Figure 2 It is an equivalent circuit diagram of the mechanical circuit breaker opening process of the present invention; Figure 3 It is a flowchart of the arc prediction model of the present invention Figure 4 It is a flowchart of the machine learning-based hybrid DC circuit breaker optimization design method of the present invention DETAILED DESCRIPTION OF THE EMBODIMENTS To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. Components of the embodiments of this application generally described and illustrated in the drawings here can be arranged and designed in a variety of different configurations.

[0029] Therefore, the detailed description of the embodiments of this application provided in the drawings below is not intended to limit the scope of this application that is claimed, but is merely representative of selected embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of this application without creative efforts fall within the scope of protection of this application.

[0030] The hybrid DC circuit breaker adopts the natural commutation method, that is, after the contacts are opened, the arc is used to force the current to commutate to the commutation branch where the power electronic device is located. The arc characteristics (arc voltage, current) at the commutation moment and the characteristic parameters of the power electronic device are crucial for reliable commutation. Therefore, it is necessary to find the optimal commutation moment, that is, the conduction moment of the power electronic device. The hybrid circuit breaker consists of a main current-carrying branch (mechanical circuit breaker), a commutation branch (power electronic device), and a dissipative branch (metal oxide varistor).

[0031] Referring to Figure 1 , this application provides an optimization design method for a hybrid DC circuit breaker based on machine learning. First, a short-circuit opening waveform data set of the mechanical circuit breaker is obtained, that is, the historical opening and breaking arc waveforms under different products and different short-circuit conditions. The product parameters and circuit parameters (mechanical circuit breaker product parameters, external circuit parameters) are further obtained through the above arc waveforms.

[0032] Then, an arc prediction model based on the Long Short-Term Memory (LSTM) network is established. The above product parameters, circuit parameters, and arc waveforms are used as the training set to train the model, so as to predict the arc current under unknown short-circuit conditions and mechanical product conditions. Based on the arc extinction principle of the DC system, the arc voltage is further calculated, and the obtained arc current and voltage are used as the optimization input variables.

[0033] Secondly, according to the device characteristic parameters provided in the data sheet of the power electronic device, the key device parameters that affect the turn-off characteristics of the power electronic device are used as the optimization input variables.

[0034] Finally, taking the opening time as the optimization objective function and the device temperature rise, short - circuit protection time, and device withstand voltage as the constraint conditions, a multi - parameter optimization method based on the Genetic Algorithm (GA) is established. Using the above variables as inputs, the optimal parameter selection of the hybrid DC circuit breaker under the condition of the minimum opening time is finally obtained, and the design of a high - performance hybrid circuit breaker is guided according to the optimal parameters.

[0035] Embodiment 1 Refer to Figures 1-4 , a method for optimizing the design of a hybrid DC circuit breaker based on machine learning, including the following steps: Step 1: Obtain the arc waveforms during the opening process of the mechanical circuit breaker under different short - circuit conditions; Obtain the historical arc waveform data during the opening process of the mechanical circuit breaker under different short - circuit conditions, and then obtain the current waveform and voltage waveform of the arc during the opening process of the mechanical circuit breaker.

[0036] Step 2: Determine the circuit characteristic parameters under each short - circuit condition according to the obtained arc waveforms and in combination with the equivalent circuit during the short - circuit opening process, and at the same time obtain the product parameters of the circuit breaker.

[0037] Refer to Figure 2 , the equivalent circuit during the short - circuit opening process. The expression of the equivalent circuit is as follows:

[0038] In the formula, i f is the short - circuit current, R 0 is the line resistance, L 0 is the line inductance, u arc is the arc voltage, V s is the system power supply voltage.

[0039] The product parameters include the rated voltage U e of the mechanical circuit breaker, I e the rated current, h the contact spacing, P the series connection method, i f the short - circuit current, etc. The product parameters can be obtained from the relevant data manuals of the circuit breaker.

[0040] Step 3: Pre - process the circuit characteristic parameters and product parameters, construct a feature vector according to the pre - processed circuit characteristic parameters and product parameters, and train an LSTM model according to the feature vector. The trained LSTM model is used to predict the arc waveform of the circuit breaker.

[0041] Assume that a short - circuit current waveform has a total of n data points. The above - mentioned circuit characteristic parameters are concatenated into a feature vector of n×m, which is used as the input data for the LSTM training network. The feature vector is divided into n - L groups of data according to a time window of length L. Subsequently, the current value at the L + 1 moment is predicted based on the first L data of the current, and the above - mentioned predicted data is used as the input for the next iteration. In this way, multiple arc - current data are cyclically trained.

[0042] This model is used to predict the arc waveforms under different short - circuit conditions and different mechanical circuit breaker product conditions. At the same time, in order to improve its generalization ability, an attention mechanism is introduced to better learn the time dependence of the data in the model. The predicted current is brought into the equivalent circuit to calculate the arc voltage.

[0043] Step 4: Input the circuit characteristic parameters under the set conditions and the product parameters of the corresponding circuit breaker into the trained LSTM model. Predict the arc waveform under this condition, and calculate the arc voltage in combination with the arc - extinguishing principle of the DC system. Use the current and voltage of the arc as input variables for subsequent genetic - algorithm optimization input.

[0044] Step 5: Obtain the key parameters affecting the breaking characteristics of the hybrid circuit breaker and use them as optimization input variables.

[0045] The hybrid circuit breaker includes a mechanical circuit breaker and power - electronic devices. According to the data manuals of different types of power - electronic devices, key characteristic parameters of the power - electronic devices are extracted, including device transconductance g m , parasitic capacitance C ies , conduction threshold voltage V th , rated voltage U e etc. The power - electronic device is a thyristor or an IGBT.

[0046] Step 6: Determine the process loss of the power - electronic device during the switching stage according to the key parameters of the power - electronic device; determine the minimum breaking time of the hybrid circuit breaker according to the working time of the energy - consuming branch of the hybrid circuit breaker.

[0047] 1) Calculate the process loss of the circuit breaker during the switching stage according to the key parameters: (2) (3) (4) (5) In the formula: E on is the turn - on loss, ton is the turn-on time, P on is the turn-on power, E d is the conduction loss, t d is the conduction time, P d is the conduction power, E off is the turn-off loss, t off is the turn-off time, P off is the turn-off power, E IGBT is the total loss during the switching process.

[0048] 2) Calculate the working time of the energy-consuming branch of the hybrid circuit breaker, and determine the minimum opening time of the hybrid circuit breaker according to the working time.

[0049] Calculate the working time of the energy-consuming branch. After the power electronic device is turned off, the overvoltage caused by the line parasitic inductance forces the current to commutate to the energy-consuming branch, and then the fault current dissipates energy through this branch. The conduction time of this energy-consuming branch can be calculated by the following formula: (6) In the formula, t mov is the energy-consuming time of this branch, V clamp is the clamping voltage value of the metal oxide varistor.

[0050] Finally, the opening time of the hybrid circuit breaker t total is: (7) Through the above formula, the relationship between the characteristics of the power electronic device, the arc behavior and the opening time can be established.

[0051] Step 7: Establish an optimization model based on the genetic algorithm according to the arc current and voltage, the key parameters of the opening characteristics of the hybrid circuit breaker, the process loss and the opening time of the power electronic device during the switching stage. Take the opening time as the optimization objective function, and add non-linear constraints based on the device failure principle to the optimization model.

[0052] The non-linear constraints include the temperature rise, the latching effect and the withstand voltage of the power electronic device.

[0053] Power electronic devices have a rated withstand voltage. Therefore, the voltage spike after the power electronic device is turned off shall not exceed its rated withstand voltage. At the same time, the increase in the device junction temperature caused by the switching loss of the power electronic device shall be lower than 175 °C specified by the device manufacturer. The 4th-order Foster thermal network model is used to evaluate the temperature rise during the transient process of the device. The temperature rise calculation formula is as follows: (8) (9) (10) In the formula, Z th is the thermal impedance of the selected device; R i is the i-th order thermal resistance; t i is the thermal time constant; C i is the i-th order heat capacity of the device; T j is the device junction temperature; T e is the ambient temperature. Among them, the thermal impedance parameters of power electronic devices can be obtained through function fitting of the data sheet.

[0054] An excessive voltage rise rate caused by the device turn-off may lead to the device being held and failing. Therefore, the voltage rise rate should be less than the engineering standard value of 6×10 9 V / s. According to the above analysis, the optimization objective function is as follows: Min (11) The non-linear constraints are as follows: s. t. (12) In the formula, d u / d t is the voltage rise rate at the turn-off moment of the power electronic device; V clamp is the clamping voltage of the metal oxide varistor.

[0055] Step 8: Initialize the population size, calculate the fitness of the individuals in the population according to the objective function, and perform selection, crossover, and mutation operations to generate the next generation population. Determine whether the generated next generation population meets the convergence condition. When the convergence condition is met, output the selected optimization parameters as the output; otherwise, repeat the iteration until the required convergence condition is reached.

[0056] It should be noted that when using the genetic algorithm, it is necessary to discretize the selection of its various parameters for application in actual engineering design. Therefore, it is necessary to customize its mutation and crossover functions.

[0057] Step 9: Determine the design scheme of the hybrid circuit breaker according to the optimized parameters.

[0058] In the optimized design method of the hybrid DC circuit breaker based on machine learning of the present application, since the commutation process of the natural commutation type hybrid DC circuit breaker needs to use an arc to force the current to commutate to the power electronic device branch, the arc behavior and the characteristics of the power electronic devices are crucial for current commutation at this moment. However, the existing design methods of hybrid DC circuit breakers do not consider the influence of various arc parameters and power electronic device parameters on current commutation. Therefore, the present invention proposes an optimized design method for the commutation of a hybrid DC circuit breaker based on a machine learning algorithm, comprehensively considering the arc behavior and the characteristics of power electronic devices, and obtaining the optimal design parameters of the hybrid DC circuit breaker through an artificial intelligence algorithm, so as to ensure its reliable commutation under various working conditions. This method can be widely applied to a variety of low-voltage hybrid DC circuit breaker products and has good applicability.

[0059] The present invention has the following beneficial technical effects: 1) It provides an arc waveform prediction method based on a machine learning algorithm, which can realize the prediction of arc waveforms for different working conditions and mechanical circuit breaker products, effectively reducing the cost problem brought by short-circuit experiments and saving the time required for product research and development.

[0060] 2) This optimization method can not only achieve the adaptive optimization of different types of mechanical circuit breakers, but also meet the adaptive optimization of different power electronic devices. By screening different types of switch products and devices, the overall commutation performance of the hybrid circuit breaker is maximized, which is beneficial to shortening the product design cost and realizing the substitution of domestic products.

[0061] 3) Compared with the existing technology, this optimization method comprehensively considers the influence of the arc behavior and device characteristic parameters at the commutation moment, provides comprehensive guidance for the optimized design, and is of great significance for the research and development of high-performance products.

[0062] 4) This optimization method has good scalability and can be widely applied to switch products in low, medium, and high voltage fields.

[0063] Correspondingly, the present application also provides an optimized design system for a hybrid DC circuit breaker based on machine learning, which may include: An arc prediction module, which is used to construct a feature vector according to the arc waveform during the commutation process of the mechanical circuit breaker under different short-circuit working conditions, combined with the circuit characteristic parameters and product parameters of the corresponding working conditions, and train the prediction model with the feature vector. The trained prediction model outputs the arc waveform of the mechanical circuit breaker. The joint parameter acquisition module is used to obtain the key parameters affecting the breaking characteristics of the hybrid circuit breaker, determine the process loss of the power electronic device during the switching stage according to the key parameters; determine the minimum breaking time of the hybrid circuit breaker according to the working time of the energy-consuming branch of the hybrid circuit breaker and the working time. The genetic module is used to establish an optimization model based on the genetic algorithm according to the arc waveform output by the prediction model, the key parameters affecting the breaking characteristics of the hybrid circuit breaker, the process loss of the power electronic device during the switching stage, and the breaking time of the hybrid circuit breaker. Taking the breaking time of the hybrid circuit breaker as the optimization objective function, add non-linear constraints based on the device failure principle to the optimization model. The optimization module is used to take the minimum breaking time as the optimization objective, calculate the fitness of the individuals in the population of the optimization model, and perform selection, crossover, and mutation operations to generate the next generation of the population. Determine whether the convergence condition is met according to the next generation of the population. When the convergence condition is met, output the optimization parameters, and determine the design scheme of the hybrid circuit breaker according to the optimization parameters.

[0064] This system uses machine learning algorithms to predict the arc waveforms during the breaking processes of various working conditions and mechanical circuit breaker products, and at the same time considers the influence of power electronic devices on the breaking performance of the hybrid circuit breaker. An optimization method based on the genetic algorithm is established, taking the breaking time of the hybrid circuit breaker as the optimization objective and various breaking failure faults as non-linear constraint conditions, to optimize the design parameters and device selection of the hybrid DC circuit breaker, providing technical support for the development of high-performance prototypes. This method can be applied to the design of hybrid circuit breakers in various low, medium, and high voltage fields.

[0065] It should be noted that in several embodiments provided in the present application, it should be understood that the disclosed device and method can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of each module is only a logical function division. In actual implementation, there may be other division methods. For example, multiple modules can be combined or integrated into another device, or some features can be ignored or not executed. The modules described as separate components may or may not be physically separated. The components shown as modules can be one physical unit or multiple physical units, that is, they can be located in one place, or distributed to multiple different places. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0066] In addition, each module in various embodiments of the present invention can be integrated in a processing unit, or each module can exist physically alone, or two or more modules can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0067] An electronic device provided by an embodiment of the present application includes a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the steps of the optimized design method of the hybrid DC circuit breaker based on machine learning described in any of the above embodiments are implemented.

[0068] Another electronic device provided by an embodiment of the present application may further include: an input port connected to the processor, configured to transmit multi-modal data collected by an external acquisition device to the processor; and a display unit connected to the processor, configured to display the processing result of the processor to the outside; a communication module connected to the processor, configured to implement communication between the electronic device and the outside. The display unit may be a display panel, a laser scanning display, etc.; the communication methods adopted by the communication module include but are not limited to Mobile High-Definition Link technology (HML), Universal Serial Bus (USB), High-Definition Multimedia Interface (HDMI), wireless connection (including Wireless Fidelity technology (WiFi), Bluetooth communication technology, low-power Bluetooth communication technology, communication technology based on IEEE802.11s).

[0069] A computer-readable storage medium provided by an embodiment of the present application stores a computer program. When the computer program is executed by a processor, the steps of the optimized design method of the hybrid DC circuit breaker based on machine learning described in any of the above embodiments are implemented.

[0070] For the description of the relevant parts in the optimized design system of the hybrid DC circuit breaker based on machine learning, the electronic device, and the computer-readable storage medium provided by the embodiments of the present application, please refer to the detailed description of the corresponding parts in the optimized design method of the hybrid DC circuit breaker based on machine learning provided by the embodiments of the present application, and will not be elaborated here. In addition, for the parts of the above technical solutions provided by the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art, no detailed description is given to avoid excessive elaboration.

[0071] The above content is only to illustrate the technical idea of the present invention, and the protection scope of the present invention cannot be limited thereby. Any modification made on the basis of the technical solution according to the technical idea proposed by the present invention falls within the protection scope of the claims of the present invention.

Claims

1. A method for optimizing the design of a hybrid DC circuit breaker based on machine learning, characterized in that It includes the following steps: Step 1: Based on the arc waveforms during the opening process of the mechanical circuit breaker under different short - circuit conditions, combined with the circuit characteristic parameters and product parameters of the corresponding conditions, construct feature vectors, and use the feature vectors to train the prediction model. The trained prediction model outputs the arc waveform of the mechanical circuit breaker; Step 2: Obtain the key parameters affecting the opening characteristics of the hybrid circuit breaker, determine the process losses of the power electronic devices during the switching stage according to the key parameters; according to the working time of the energy - consuming branch of the hybrid circuit breaker, determine the minimum opening time of the hybrid circuit breaker according to the working time; Step 3: Establish an optimization model based on the genetic algorithm according to the arc waveform output by the prediction model, the key parameters affecting the opening characteristics of the hybrid circuit breaker, the process losses of the power electronic devices during the switching stage, and the opening time of the hybrid circuit breaker. Take the opening time of the hybrid circuit breaker as the optimization objective function, and add non - linear constraints based on the device failure principle to the optimization model; Step 4: Take the minimum opening time as the optimization objective, calculate the fitness of the individuals in the population of the optimization model, and perform selection, crossover, and mutation operations to generate the next - generation population. Determine whether the convergence condition is met according to the next - generation population. When the convergence condition is met, output the optimized parameters, and determine the design scheme of the hybrid circuit breaker according to the optimized parameters.

2. The optimized design method of a hybrid DC circuit breaker based on machine learning according to claim 1, wherein The determination method of the circuit characteristic parameters described in Step 1 is as follows: Based on the arc waveform and combined with the equivalent circuit during the short - circuit opening process, determine the circuit characteristic parameters under each short - circuit condition.

3. The optimized design method of a hybrid DC circuit breaker based on machine learning according to claim 2, wherein The equivalent circuit is as follows: Wherein, i f is the short-circuit current, R 0 is the line resistance, L 0 is the line inductance, u arc is the arc voltage, V s is the system power supply voltage.

4. A method for optimizing the design of a hybrid DC circuit breaker based on machine learning according to claim 1, characterized in that The prediction model described in Step 1 is an LSTM model, and an attention mechanism is introduced into the LSTM model to learn the time - series data of the arc waveform.

5. A method for optimizing the design of a hybrid DC circuit breaker based on machine learning according to claim 1, wherein, The key parameters described in Step 2 are the key parameters affecting the opening characteristics in the hybrid circuit breaker. Determine the process losses of the power electronic devices during the switching stage according to the key parameters. The process losses are as follows: Where: E on is the turn-on loss, t on is the turn-on time, P on is the turn-on power, E d is the conduction loss, t d is the conduction time, P d is the conduction power, E off is the turn-off loss, t off is the turn-off time, P off is the turn-off power, E IGBT is the total loss during the switching process.

6. The optimized design method of a hybrid DC circuit breaker based on machine learning according to claim 5, wherein The method of determining the working time of the energy - consuming branch of the hybrid circuit breaker in Step 2 is as follows: In the formula, t mov is the energy consumption time of this branch; V clamp is the clamping voltage value of the metal oxide varistor; The method of determining the minimum opening time of the hybrid circuit breaker according to the working time is as follows: 。 7. A method for optimizing the design of a hybrid DC circuit breaker based on machine learning according to claim 1, characterized in that, The non - linear constraints based on the device failure principle in Step 3 include the temperature rise, latching effect, and voltage withstand of the power electronic devices.

8. The optimized design method of a hybrid DC circuit breaker based on machine learning according to claim 7, characterized in that, The calculation method of the temperature rise is as follows: Use a 4 - order Foster thermal network model to evaluate the temperature rise of the transient process of the power electronic device. The temperature - rise calculation formula is: In the formula, Z th is the thermal impedance of the selected device; R i is the i-th order thermal resistance; t i is the thermal time constant; C i is the i-th order heat capacity of the device; T j is the device junction temperature; T e is the ambient temperature.

9. A hybrid DC circuit breaker optimization design system based on machine learning, characterized in that, It includes: An arc prediction module, which is used to construct feature vectors based on the arc waveforms during the opening process of the mechanical circuit breaker under different short - circuit conditions, combined with the circuit characteristic parameters and product parameters of the corresponding conditions, and use the feature vectors to train the prediction model. The trained prediction model outputs the arc waveform of the mechanical circuit breaker; A joint parameter acquisition module, which is used to obtain the key parameters affecting the opening characteristics of the hybrid circuit breaker, determine the process losses of the power electronic devices during the switching stage according to the key parameters; according to the working time of the energy - consuming branch of the hybrid circuit breaker, determine the minimum opening time of the hybrid circuit breaker according to the working time; A genetic module, which is used to establish an optimization model based on a genetic algorithm according to the arc waveform output by the prediction model, the key parameters affecting the breaking characteristics of the hybrid circuit breaker, the process loss of the power electronic device during the switching stage, and the breaking time of the hybrid circuit breaker. Taking the breaking time of the hybrid circuit breaker as the optimization objective function, a non-linear constraint based on the device failure principle is added to the optimization model; An optimization module, which is used to take the minimum breaking time as the optimization objective, calculate the fitness of the individuals in the population of the optimization model, and perform selection, crossover, and mutation operations to generate the next generation of the population. Determine whether the convergence condition is met according to the next generation of the population. When the convergence condition is met, output the optimization parameters, and determine the design scheme of the hybrid circuit breaker according to the optimization parameters.

10. An electronic device, characterized in that, Comprising: A memory, which is used to store computer programs; A processor, which is used to implement the steps of the optimization design method of the hybrid DC circuit breaker based on machine learning according to any one of claims 1-8 when executing the computer program.

Images