CN116914685B - Solid state circuit breaker control system and solid state circuit breaker control method - Google Patents

Solid state circuit breaker control system and solid state circuit breaker control method Download PDF

Info

Publication number
CN116914685B
CN116914685B CN202311189779.1A CN202311189779A CN116914685B CN 116914685 B CN116914685 B CN 116914685B CN 202311189779 A CN202311189779 A CN 202311189779A CN 116914685 B CN116914685 B CN 116914685B
Authority
CN
China
Prior art keywords
current
model
bus
fault
voltage
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202311189779.1A
Other languages
Chinese (zh)
Other versions
CN116914685A (en
Inventor
张亮
彭保基
谭金平
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhuhai Huizhong Energy Technology Co ltd
Original Assignee
Zhuhai Huizhong Energy Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhuhai Huizhong Energy Technology Co ltd filed Critical Zhuhai Huizhong Energy Technology Co ltd
Priority to CN202311189779.1A priority Critical patent/CN116914685B/en
Publication of CN116914685A publication Critical patent/CN116914685A/en
Application granted granted Critical
Publication of CN116914685B publication Critical patent/CN116914685B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02H—EMERGENCY PROTECTIVE CIRCUIT ARRANGEMENTS
    • H02H3/00—Emergency protective circuit arrangements for automatic disconnection directly responsive to an undesired change from normal electric working condition with or without subsequent reconnection ; integrated protection
    • H02H3/02—Details
    • H02H3/021—Details concerning the disconnection itself, e.g. at a particular instant, particularly at zero value of current, disconnection in a predetermined order
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • G05B13/027—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
    • G05B19/05—Programmable logic controllers, e.g. simulating logic interconnections of signals according to ladder diagrams or function charts
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02H—EMERGENCY PROTECTIVE CIRCUIT ARRANGEMENTS
    • H02H1/00—Details of emergency protective circuit arrangements
    • H02H1/0007—Details of emergency protective circuit arrangements concerning the detecting means
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02H—EMERGENCY PROTECTIVE CIRCUIT ARRANGEMENTS
    • H02H1/00—Details of emergency protective circuit arrangements
    • H02H1/0092—Details of emergency protective circuit arrangements concerning the data processing means, e.g. expert systems, neural networks
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03K—PULSE TECHNIQUE
    • H03K17/00—Electronic switching or gating, i.e. not by contact-making and –breaking
    • H03K17/04—Modifications for accelerating switching
    • H03K17/041—Modifications for accelerating switching without feedback from the output circuit to the control circuit
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03K—PULSE TECHNIQUE
    • H03K2217/00—Indexing scheme related to electronic switching or gating, i.e. not by contact-making or -breaking covered by H03K17/00
    • H03K2217/0081—Power supply means, e.g. to the switch driver

Landscapes

  • Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Automation & Control Theory (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Keying Circuit Devices (AREA)

Abstract

The application discloses a solid state circuit breaker control system and a method. The model predictive controller in the system can predict the fault predictive result to obtain the state change trend in a future period of time, and the fault predictive result is utilized to conduct real-time fault prediction and decision, so that the control of the solid state switch is optimized, the solid state switch can rapidly respond to the system state change of the solid state breaker control system, and the response speed and control precision of the solid state breaker control system are improved. That is, the model predictive controller is adopted to optimally control the solid-state switch, so that the problems of switch shake, vibration and the like in the traditional control mode can be avoided, the loss and heat of the system are reduced, and the stability and reliability of the system are improved. The system also monitors the anode current and the cathode current and the direct current voltage of the system in real time by adopting the current sensor and the voltage sensor, can timely find out abnormal conditions in the system, carries out corresponding treatment, and improves the safety and the reliability of the system.

Description

Solid state circuit breaker control system and solid state circuit breaker control method
Technical Field
The application relates to the technical field of circuit breakers, in particular to a solid-state circuit breaker control system and a solid-state circuit breaker control method.
Background
In recent years, with the development of renewable energy sources, direct current transmission and other technologies, the application of direct current power systems is increasing. The direct current power system has high-efficiency and stable power transmission characteristics, and is suitable for the fields of long-distance power transmission, large-scale energy storage, electric traffic and the like. Therefore, dc system fault protection is critical to ensure safe and reliable operation of the dc power system. Compared with an alternating current system, the direct current system has low impedance and small inertia, so that fault current rises fast and has high amplitude, and a serious challenge is provided for direct current fault protection. The design and implementation of dc system fault protection requires consideration of these particularities and ensures that the system can quickly respond to and cut off fault current when a fault occurs to protect the integrity of the equipment and system. The direct current breaker as a protection device can rapidly detect and cut off fault current, and protect equipment and personnel of a direct current power system from faults and dangers. At present, a traditional direct current breaker generally adopts a mechanical structure, and has the advantages of complex structure, long action time, difficult arc extinction, and easy influence of mechanical abrasion and contact resistance, thereby leading to system failure. In order to overcome the defects of the traditional direct current circuit breaker, the solid state circuit breaker adopts a power semiconductor device as a main switch, and can realize the rapid cutting off and arc-free turn-off of fault current. However, the accuracy and reliability of dc solid state circuit breaker control systems still have certain limitations. Control of a dc solid state circuit breaker requires the realization of fast and accurate detection of faults and cut-off of current. However, the current control method may have drawbacks in terms of accuracy and reliability. Dc solid state circuit breakers need to have good dynamic response capabilities in the face of rapidly changing current and voltage conditions. The current control method may not meet the requirement of quick response, so that the circuit breaker cannot cut off current in time when a fault occurs. For example, the algorithm that detects the fault may be inaccurate, resulting in erroneous decisions or delayed shut-off actions. Therefore, how to improve the accuracy and reliability of the control of the dc solid state circuit breaker and to improve the dynamic response performance of the dc solid state circuit breaker is important, so as to ensure timely and effective fault protection is a problem to be solved.
Disclosure of Invention
The application provides a solid state circuit breaker control system and a method, which can solve the problems of how to improve the accuracy and reliability of the control of a direct current solid state circuit breaker, how to improve the dynamic response performance of the direct current solid state circuit breaker and how to ensure timely and effective fault protection.
In a first aspect, the present application provides a solid state circuit breaker control system comprising a voltage sensor, a current sensor, a model predictive controller, a field programmable gate array FPGA, a driver, and a solid state switch;
the voltage sensor is used for detecting bus voltage of the system;
the current sensor is used for detecting the current of the anode and the cathode of a bus of the system;
the model prediction controller is used for determining a fault prediction result according to the bus voltage and the bus anode and cathode current, and sending a control instruction corresponding to the fault prediction result to the field programmable gate array FPGA according to the fault prediction result;
the FPGA is used for sending a control signal corresponding to the control instruction to the solid-state switch through the driver according to the control instruction so as to control the switch state of the solid-state switch;
The model predictive controller comprises a fault predictive model; the model predictive controller is specifically configured to: inputting the bus voltage and the bus positive and negative current into the fault prediction model to obtain a fault prediction result, wherein the fault prediction result comprises fault condition information and an occurrence probability value corresponding to the fault condition information; if the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, judging whether the bus voltage and/or the bus positive and negative current meet fault conditions or not; if yes, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off, and sending the control instruction to the FPGA; if not, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be closed, and sending the control instruction to the FPGA; the fault prediction model is an LSTM model based on an Adam optimization algorithm;
the LSTM model based on the Adam optimization algorithm is generated as follows:
step 1, initializing weight and bias items of an LSTM model;
step 2, initializing first moment estimation and second moment estimation of an Adam algorithm;
Step 3, training and iterating the fault prediction model; firstly, calculating forward propagation of an LSTM model by using the historical busbar voltage and the current of the positive electrode and the negative electrode of the historical busbar in the training set to obtain a fault prediction result and a loss function value; then, calculating the gradient of the loss function value to the model parameter of the LSTM model; next, the first moment estimate is updated: the first moment estimate is updated using an exponentially weighted average, calculated by the following formula:
wherein m is a first moment estimation, beta1 is an attenuation rate parameter in an Adam algorithm, and grad is a gradient of a loss function value to a model parameter of an LSTM model;
updating the second moment estimate: the second moment estimate is updated using an exponentially weighted average, calculated by the following formula:
where v is the second moment estimate, beta2 is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of LSTM;
step 4, deviation correction: deviation correction was performed, calculated by the following formula:
wherein m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
step 5, updating model parameters of the LSTM by using the corrected first moment estimation and second moment estimation, wherein the model parameters are calculated by the following formula:
Wherein θ is a model parameter; η is the learning rate which can be chosen by multiple experiments and with minimal training errors; epsilon is a constant for avoiding division by zero; m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
repeating the step 3: and training and iterating the fault prediction model until the training times of the LSTM model reach the preset iteration times or the model parameters of the LSTM model reach the convergence condition.
Optionally, the solid state breaker control system further comprises a capacitor and a resistor; wherein the current sensor comprises a first current sensor and a second current sensor, the solid state switch comprises a first solid state switch and a second solid state switch, and the driver comprises a first driver and a second driver;
the first end of the voltage sensor is connected with the positive bus of the system, the second end of the voltage sensor is connected with the negative bus of the system, and the third end of the voltage sensor is connected with the model predictive controller;
the first end of the first current sensor is connected with the positive bus through the capacitor, and the second end of the first current sensor is connected with the model predictive controller;
The first end of the second current sensor is connected with the negative bus, and the second end of the second current sensor is connected with the model predictive controller;
the model prediction controller is connected with a third end of the FPGA;
the first end of the FPGA is connected with the first end of the first driver, and the second end of the first driver is connected with the second end of the first solid-state switch; the first end of the first solid-state switch is connected with the positive electrode bus through the capacitor, and the third end of the first solid-state switch is connected with one end of the resistor;
the second end of the FPGA is connected with the first end of the second driver, and the second end of the second driver is connected with the second end of the second solid-state switch; and the first end of the second solid-state switch is connected with the negative bus, and the third end of the second solid-state switch is connected with the other end of the resistor.
Optionally, the fault condition includes a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend; the second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold.
In a second aspect, the present application provides a solid state circuit breaker control method, the method comprising:
detecting the bus voltage of a solid-state circuit breaker control system;
detecting the current of the anode and the cathode of a bus of the solid-state circuit breaker control system;
determining a fault prediction result according to the bus voltage and the bus anode and cathode current;
generating a control instruction corresponding to the fault prediction result according to the fault prediction result;
according to the control instruction, controlling the switching state of a solid state switch in the solid state breaker control system;
determining a fault prediction result according to the bus voltage and the bus anode and cathode current, including:
inputting the busbar voltage and the busbar anode and cathode current into a trained fault prediction model to obtain a fault prediction result, wherein the fault prediction result comprises fault condition information and an occurrence probability value corresponding to the fault condition information;
the fault prediction model is an LSTM model based on an Adam optimization algorithm;
the LSTM model based on the Adam optimization algorithm is generated as follows:
step 1, initializing weight and bias items of an LSTM model;
step 2, initializing first moment estimation and second moment estimation of an Adam algorithm;
Step 3, training and iterating the fault prediction model; firstly, calculating forward propagation of an LSTM model by using the historical busbar voltage and the current of the positive electrode and the negative electrode of the historical busbar in the training set to obtain a fault prediction result and a loss function value; then, calculating the gradient of the loss function value to the model parameter of the LSTM model; next, the first moment estimate is updated: the first moment estimate is updated using an exponentially weighted average, calculated by the following formula:
wherein m is a first moment estimation, beta1 is an attenuation rate parameter in an Adam algorithm, and grad is a gradient of a loss function value to a model parameter of an LSTM model;
updating the second moment estimate: the second moment estimate is updated using an exponentially weighted average, calculated by the following formula:
where v is the second moment estimate, beta2 is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of LSTM;
step 4, deviation correction: deviation correction was performed, calculated by the following formula:
wherein m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
step 5, updating model parameters of the LSTM by using the corrected first moment estimation and second moment estimation, wherein the model parameters are calculated by the following formula:
Wherein θ is a model parameter; η is the learning rate which can be chosen by multiple experiments and with minimal training errors; epsilon is a constant for avoiding division by zero; m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
repeating the step 3: and training and iterating the fault prediction model until the training times of the LSTM model reach the preset iteration times or the model parameters of the LSTM model reach the convergence condition.
Optionally, the generating, according to the failure prediction result, a control instruction corresponding to the failure prediction result includes:
if the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, judging whether the bus voltage and/or the bus positive and negative current meet fault conditions or not;
if yes, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off;
if not, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be closed.
Optionally, the fault condition includes a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend; the second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold.
According to the technical scheme, the solid-state circuit breaker control system comprises a voltage sensor, a current sensor, a model predictive controller, a Field Programmable Gate Array (FPGA), a driver and a solid-state switch; the voltage sensor is used for detecting bus voltage of the system; the current sensor is used for detecting the current of the anode and the cathode of a bus of the system; the model prediction controller is used for determining a fault prediction result according to the bus voltage and the bus anode and cathode current, and sending a control instruction corresponding to the fault prediction result to the field programmable gate array FPGA according to the fault prediction result; and the FPGA is used for sending a control signal corresponding to the control instruction to the solid-state switch through the driver according to the control instruction so as to control the switch state of the solid-state switch. In this embodiment, since the model predictive controller may perform system state prediction (i.e., prediction of a failure prediction result), obtain a state change trend in a future period of time, and perform real-time failure prediction and decision by using the failure prediction result, and optimize control of the solid state switch, so that the solid state switch can quickly respond to a system state change of the solid state breaker control system, and improve a response speed and control accuracy of the solid state breaker control system. That is, the embodiment adopts the model predictive controller to optimally control the solid-state switch, so that the problems of switch shake, oscillation and the like in the traditional control mode can be avoided, the loss and heat of the system are reduced, and the stability and reliability of the system are improved. In addition, the embodiment can also monitor the positive and negative electrode current and the direct current voltage of the system in real time by adopting the current sensor and the voltage sensor, so that abnormal conditions in the system can be found in time, corresponding processing can be carried out, and the safety and the reliability of the system are improved. By combining the advantages, the solid-state circuit breaker control system of the embodiment has higher practicability and economic benefit, and can be widely applied to the field of direct current protection, such as power regulation, energy storage, electric automobile charging and the like in a power system.
Further effects of the above-described non-conventional preferred embodiments will be described below in connection with the detailed description.
Drawings
In order to more clearly illustrate the embodiments or prior art solutions of the present application, the drawings that are required for the description of the embodiments or prior art will be briefly described below, it being apparent that the drawings in the following description are only some of the embodiments described in the present application, and that other drawings may be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a schematic system configuration diagram of a solid-state circuit breaker control system provided in the present application;
fig. 2 is a schematic flow chart of a method for controlling a solid-state circuit breaker provided in the present application;
fig. 3 is a flow chart of another method for controlling a solid state circuit breaker provided in the present application;
fig. 4 is a schematic diagram of a network architecture of an LSTM model provided in the present application.
Detailed Description
For the purposes, technical solutions and advantages of the present application, the technical solutions of the present application will be clearly and completely described below with reference to specific embodiments and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
Various non-limiting embodiments of the present application are described in detail below with reference to the attached drawing figures.
Referring to fig. 1, a solid state circuit breaker control system in an embodiment of the present application is shown, the system comprising a voltage sensor, a current sensor, a model predictive controller, a field programmable gate array (Field Programmable Gate Array, FPGA), a driver, and a solid state switch.
In this embodiment, the voltage sensor is used to detect the bus voltage of the system. The current sensor is configured to detect a bus positive and negative current of the system, and in one implementation, the current sensor may include a first current sensor configured to detect a bus positive current i+ of the system and a second current sensor configured to detect a bus negative current I-of the system.
The model prediction controller is used for determining a fault prediction result according to the bus voltage and the bus anode and cathode current, and sending a control instruction corresponding to the fault prediction result to the field programmable gate array FPGA according to the fault prediction result.
In one implementation, the model predictive controller may include a fault predictive model. In one implementation, the fault prediction model may be an LSTM model based on Adam optimization algorithm, where the network architecture of the LSTM model is shown in fig. 4. Specifically, the training set of the fault prediction model includes a historical busbar voltage, a historical busbar anode and cathode current, and a corresponding fault result. The model training of the fault prediction model is to use the historical bus voltage and the current of the positive electrode and the negative electrode of the historical bus in the training set as inputs, output the historical bus voltage and the current of the positive electrode and the current of the negative electrode of the historical bus as corresponding fault prediction results, calculate a loss function value by utilizing the fault prediction results and preset corresponding fault results, and adjust model parameters of the fault prediction model by utilizing the loss value. In this embodiment, the failure prediction model may adaptively adjust the learning rate through Adam optimization algorithm, and update parameters by combining first moment estimation and second moment estimation, so that the training process of the LSTM model is more efficient and stable, and the specific process is as follows: 1. the weights and bias terms of the LSTM model are initialized. 2. Initializing a first moment estimate (an exponentially weighted average of the gradient) and a second moment estimate (an exponentially weighted average of the square of the gradient) of the Adam algorithm; initializing first moment estimation: initializing a first moment estimation as a zero vector, wherein the dimension is the same as the corresponding gradient dimension, and initializing a second moment estimation: the second moment estimate is initialized to a zero vector, the dimension being the same as the corresponding gradient dimension. 3. Training and iterating the fault prediction model; specifically, forward propagation is performed first: calculating forward propagation of a fault prediction model LSTM by using current parameters (namely, the historical bus voltage and the current of the positive electrode and the negative electrode of the historical bus in the training set) to obtain a predicted value (namely, a fault prediction result) and a loss function value; then, back propagation is performed: the gradient of the loss function value to the model parameters of the failure prediction model LSTM is calculated. Next, the first moment estimate is updated: updating the first moment estimate using an exponentially weighted average can be calculated by the following formula:
Where m is the first moment estimate, beta 1 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of the failure prediction model LSTM;
updating the second moment estimate: using an exponentially weighted average to update the second moment estimate, it can be calculated by the following formula:
where v is the second moment estimate, beta 2 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of the failure prediction model LSTM.
4. Deviation correction: to eliminate the initial bias of the first moment estimate and the second moment estimate, bias correction is performed, which can be calculated by the following formula:
where m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations.
5. Parameter updating: using the corrected first and second moment estimates to update the model parameters, the model parameters can be calculated by the following formula:
wherein θ is a model parameter; η is the learning rate which can be chosen by multiple experiments and with minimal training errors; epsilon is a constant for avoiding division by zero; m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations.
Repeating the step 3: and training and iterating the fault prediction model until the training times of the fault prediction model reach the preset iteration times or the model parameters of the fault prediction model reach the convergence condition.
Model evaluation of the failure prediction model is performed, and this embodiment adopts an Accuracy (Accuracy), a recall (recovery), a precision (precision), and an F1 score (F) 1_socre ) The four points evaluate the fault prediction model. The accuracy is one of common indexes in model evaluation, is used for measuring the accuracy degree of model prediction, and is required to be 90-95%. The calculation formula is as follows:
the recall measure model is required to be 0.7-0.9 for its ability to identify positive samples, also known as sensitivity or recall. The calculation formula is as follows:
the accuracy measures the accuracy of the model prediction positive sample, and the accuracy is required to be about 0.9. The calculation formula is as follows:
the F1 score is an index that comprehensively considers the accuracy and recall for evaluating the performance of the model. The method is a harmonic average value of the recall rate of the precision rate, and the requirement is 0.8-09. The calculation formula is as follows:
；
in the above, T P Is a true example (model forecast isPositive samples and actually positive samples), T N Is the number of true negative examples (model predicted as negative and actually negative), F P Is the number of false positive examples (model predicted as positive but actually negative), F N Is the number of false negatives (negative but actually positive samples predicted by the model).
And finally, verifying the trained fault prediction model by using verification data, evaluating the prediction capability of the fault prediction model, adjusting the fault prediction model according to the evaluation result, and selecting the optimal state prediction model according to the verification result.
Model deployment: as shown in fig. 1, a trained Adam-optimized fault prediction model LSTM model is deployed into a model predictive controller of an actual solid state circuit breaker control system) for real-time fault prediction and decision control.
The model predictive controller is specifically configured to: and inputting the bus voltage and the bus positive and negative current into the fault prediction model to obtain a fault prediction result. The fault prediction result comprises fault condition information and an occurrence probability value corresponding to the fault condition information. It can be understood that after the data of the positive electrode current i+, the negative electrode current I-and the voltage U (i.e., the bus voltage and the bus positive and negative electrode current) of the direct current solid state circuit breaker are obtained in real time, the reliability of the fault and the prediction result (i.e., the fault condition information and the occurrence probability value corresponding to the fault condition information) is judged through calculation.
If the fault condition information is faulty and the occurrence probability value is larger than a preset threshold value, judging whether the bus voltage and/or the bus positive and negative current meet fault conditions or not. The fault conditions include a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend. The second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold.
In this embodiment, a preset threshold is set for determining the reliability of the prediction result. If the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, the occurrence probability value is predicted fault_prob Greater than or equal to a preset threshold, consider the predicted outcome (i.e., to the failure predicted outcome) to be useable prediction Reliable (i.e., correct, accurate), the expression is as follows:
If the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, the predicted result (i.e. the failure predicted result) is considered to be available prediction Reliable (i.e. correct, accurate), it is determined whether the bus voltage and the bus positive-negative current meet fault conditions. The fault conditions include a first fault condition and a second fault condition; the first fault condition is the voltage trend U of the bus voltage of the system trend And a preset voltage trend U expected_trend Non-conforming (i.e. U trend !=U expected_trend ) Alternatively, the current trend I of the busbar anode and cathode current of the system trend And preset current trend I expected _trend Non-conforming (i.e. I trend !=I expected _trend ) Such as current trend I+ of bus positive current of system trend And presetting the current trend I- expected _trend Non-conforming current trend I-t of bus negative current of system rend And preset negative current trend I + expected _trend Non-conforming. The second fault condition includes at least one of: bus voltage U of the system current Is greater than a preset voltage upper threshold U threshold_hig (i.e. U) current >U threshold_hig ) Bus voltage U of the system current Less than a threshold U at a preset voltage threshold_low (i.e. U) current <U threshold_low ) The positive and negative current of the bus of the system is larger than the preset upper current threshold (such as the positive current I+of the bus of the system current Is larger than a preset positive current upper threshold I + threshold_high I.e. I + current > I+ threshold_high Or bus negative current I- current Is greater than a preset anode current upper threshold I- threshold_high I.e.) current >I- threshold_high ). I.e. fault condition fault detected The method comprises the following steps:
wherein, the upper threshold value U of the voltage is preset threshold_high =K ref1 U r The method comprises the steps of carrying out a first treatment on the surface of the Threshold U under preset voltage threshold_low = K ref2 U r The method comprises the steps of carrying out a first treatment on the surface of the Presetting an upper threshold value I+of the positive current threshold_high = K ref3 I+ r The method comprises the steps of carrying out a first treatment on the surface of the Presetting a threshold under the positive current: i- threshold_high = K re4 I- r The method comprises the steps of carrying out a first treatment on the surface of the Wherein U is r I+for bus voltage rating in the system r And I- r Respectively rated bus anode and cathode currents in the system; setting K ref1 The general value is 1.3-1.5; k (K) re2 The general value is 0.5-0.7; k (K) ref3 And K ref4 The value is generally 2-3.
If the bus voltage or the bus positive-negative current meets the fault condition, a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off is generated and the control instruction is sent to the FPGA, it can be understood that if the fault is judged to be true (fault detected True), i.e. the busbar voltage or the busbar positive and negative current meets fault conditions, and the prediction result is reliable (i.e. the occurrence probability value is larger than a preset threshold value, and the busbar is available) prediction True), corresponding decision control operation is executed, and the FPGA is controlled to send a turn-off signal to drive the solid state switch to turn off so as to cut off the fault. If not, namely the bus voltage or the bus positive-negative current does not accord with the fault If the condition is that a control instruction for controlling the switch state of the solid state switch to be adjusted to be closed is generated and the control instruction is sent to the FPGA, it can be understood that if the fault is judged to be false (fault detected Fault), i.e. the busbar voltage or the busbar positive and negative current meets fault conditions, and the prediction result is reliable (i.e. the occurrence probability value is larger than a preset threshold value, and the busbar is available) prediction And if the solid state switch is fault), corresponding decision control operation is executed, and the FPGA is controlled to send a closing signal to drive the solid state switch to be closed so as to continue running.
And the FPGA is used for sending a control signal corresponding to the control instruction to the solid-state switch through the driver according to the control instruction. And if the control instruction is an off instruction, the FPGA sends a control signal (namely an off signal) corresponding to the control instruction to the solid-state switch through the driver so as to control the switch state of the solid-state switch to be the off state through the driver. If the control instruction is a closing instruction, the FPGA sends a control signal (namely a closing signal) corresponding to the control instruction to the solid-state switch through the driver so as to control the switch state of the solid-state switch to be a closing state. Wherein in one implementation, the solid state switches may include a first solid state switch and a second solid state switch, and the driver may include a first driver and a second driver; the FPGA can send a control signal corresponding to the control instruction to the first solid-state switch through the first driver according to the control instruction, and the FPGA can send a control signal corresponding to the control instruction to the second solid-state switch through the second driver according to the control instruction. In one implementation, the solid state switch may be a direct current solid state switch.
Specifically, as shown in fig. 1, the connection relationship of the solid state breaker control system is as follows: the first end of the voltage sensor is connected with the positive bus of the system, the second end of the voltage sensor is connected with the negative bus of the system, and the third end of the voltage sensor is connected with the model predictive controller; the first end of the first current sensor is connected with the positive bus, and the second end of the first current sensor is connected with the model predictive controller; the first end of the second current sensor is connected with the negative bus, and the second end of the second current sensor is connected with the model predictive controller; the model prediction controller is connected with a third end of the FPGA; the first end of the FPGA is connected with the first end of the first driver, and the second end of the first driver is connected with the second end of the first solid-state switch; the first end of the first solid-state switch is connected with the positive electrode bus; the second end of the FPGA is connected with the first end of the second driver, and the second end of the second driver is connected with the second end of the second solid-state switch; and the first end of the second solid-state switch is connected with the negative bus.
Specifically, as shown in fig. 1, the solid state circuit breaker control system further includes an inductor and a resistor R load The method comprises the steps of carrying out a first treatment on the surface of the Wherein the current sensor comprises a first current sensor and a second current sensor, the solid state switch comprises a first solid state switch and a second solid state switch, and the driver comprises a first driver and a second driver; the first end of the voltage sensor is connected with the positive bus of the system, the second end of the voltage sensor is connected with the negative bus of the system, and the third end of the voltage sensor is connected with the model predictive controller; the first end of the first current sensor is connected with the positive bus through the inductor, and the second end of the first current sensor is connected with the model predictive controller; the first end of the second current sensor is connected with the negative bus, and the second end of the second current sensor is connected with the model predictive controller; the model prediction controller is connected with a third end of the FPGA; the first end of the FPGA is connected with the first end of the first driver, and the second end of the first driver is connected with the second end of the first solid-state switch; the first end of the first solid-state switch is connected with the positive electrode bus through the inductor, and the third end of the first solid-state switch is connected with one end of the resistor; the second end of the FPGA is connected with the first end of the second driver, and the second end of the second driver is connected with the second end of the second solid-state switch Connecting; and the first end of the second solid-state switch is connected with the negative bus, and the third end of the second solid-state switch is connected with the other end of the resistor.
According to the technical scheme, the solid-state circuit breaker control system comprises a voltage sensor, a current sensor, a model predictive controller, a Field Programmable Gate Array (FPGA), a driver and a solid-state switch; the voltage sensor is used for detecting bus voltage of the system; the current sensor is used for detecting the current of the anode and the cathode of a bus of the system; the model prediction controller is used for determining a fault prediction result according to the bus voltage and the bus anode and cathode current, and sending a control instruction corresponding to the fault prediction result to the field programmable gate array FPGA according to the fault prediction result; and the FPGA is used for sending a control signal corresponding to the control instruction to the solid-state switch through the driver according to the control instruction so as to control the switch state of the solid-state switch. In this embodiment, since the model predictive controller may perform system state prediction (i.e., prediction of a failure prediction result), obtain a state change trend in a future period of time, and perform real-time failure prediction and decision by using the failure prediction result, and optimize control of the solid state switch, so that the solid state switch can quickly respond to a system state change of the solid state breaker control system, and improve a response speed and control accuracy of the solid state breaker control system. That is, the embodiment adopts the model predictive controller to optimally control the solid-state switch, so that the problems of switch shake, oscillation and the like in the traditional control mode can be avoided, the loss and heat of the system are reduced, and the stability and reliability of the system are improved. In addition, the embodiment can also monitor the positive and negative electrode current and the direct current voltage of the system in real time by adopting the current sensor and the voltage sensor, so that abnormal conditions in the system can be found in time, corresponding processing can be carried out, and the safety and the reliability of the system are improved. By combining the advantages, the solid-state circuit breaker control system of the embodiment has higher practicability and economic benefit, and can be widely applied to the field of direct current protection, such as power regulation, energy storage, electric automobile charging and the like in a power system. That is, the present embodiment can solve the problems existing in the conventional control method of the dc solid-state circuit breaker, such as slow fault response speed, switch jitter, oscillation, etc., and improve the stability and reliability of the system. Meanwhile, through model predictive control, the on-off control of the direct current solid state breaker can be optimally controlled, the system loss and heat are reduced, and the service life of the system is prolonged.
Fig. 2 is a flow chart of a control method of a solid-state circuit breaker according to an embodiment of the present application. The solid state circuit breaker control method of fig. 2 may be performed by the solid state circuit breaker control system of fig. 1. As shown in fig. 2, the solid state circuit breaker control method includes:
s201: the bus voltage of the solid state circuit breaker control system is detected.
For example, a voltage sensor in a solid state circuit breaker control system may be utilized for detecting a bus voltage of the solid state circuit breaker control system.
S202: and detecting the current of the anode and the cathode of the bus of the solid-state circuit breaker control system.
For example, a current sensor in a solid state circuit breaker control system may be utilized to detect bus positive and negative current in the solid state circuit breaker control system. In one implementation, the current sensor may include a first current sensor for detecting a bus positive current i+ of the system and a second current sensor for detecting a bus negative current I-of the system.
S203: and determining a fault prediction result according to the bus voltage and the bus anode and cathode current.
As an example, the bus voltage and the bus anode and cathode current may be input into a trained fault prediction model to obtain a fault prediction result, where the fault prediction result includes fault condition information and an occurrence probability value corresponding to the fault condition information. It can be understood that after the data of the positive electrode current i+, the negative electrode current I-and the voltage U (i.e., the bus voltage and the bus positive and negative electrode current) of the direct current solid state circuit breaker are obtained in real time, the reliability of the fault and the prediction result (i.e., the fault condition information and the occurrence probability value corresponding to the fault condition information) can be judged through calculation.
In one implementation, the fault prediction model may be an LSTM model based on Adam optimization algorithm, where the network architecture of the LSTM model is shown in fig. 4. Specifically, the training set of the fault prediction model includes a historical busbar voltage, a historical busbar anode and cathode current, and a corresponding fault result. The model training of the fault prediction model is to use the historical bus voltage and the current of the positive electrode and the negative electrode of the historical bus in the training set as inputs, output the historical bus voltage and the current of the positive electrode and the current of the negative electrode of the historical bus as corresponding fault prediction results, calculate a loss function value by utilizing the fault prediction results and preset corresponding fault results, and adjust model parameters of the fault prediction model by utilizing the loss value. In this embodiment, the failure prediction model may adaptively adjust the learning rate through Adam optimization algorithm, and update parameters by combining first moment estimation and second moment estimation, so that the training process of the LSTM model is more efficient and stable, and the specific process is as follows: 1. the weights and bias terms of the LSTM model are initialized. 2. Initializing a first moment estimate (an exponentially weighted average of the gradient) and a second moment estimate (an exponentially weighted average of the square of the gradient) of the Adam algorithm; initializing first moment estimation: initializing a first moment estimation as a zero vector, wherein the dimension is the same as the corresponding gradient dimension, and initializing a second moment estimation: the second moment estimate is initialized to a zero vector, the dimension being the same as the corresponding gradient dimension. 3. Training and iterating the fault prediction model; specifically, forward propagation is performed first: calculating forward propagation of a fault prediction model LSTM by using current parameters (namely, the historical bus voltage and the current of the positive electrode and the negative electrode of the historical bus in the training set) to obtain a predicted value (namely, a fault prediction result) and a loss function value; then, back propagation is performed: the gradient of the loss function value to the model parameters of the failure prediction model LSTM is calculated. Next, the first moment estimate is updated: updating the first moment estimate using an exponentially weighted average can be calculated by the following formula:
Where m is the first moment estimate, beta 1 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of the failure prediction model LSTM;
updating the second moment estimate: using an exponentially weighted average to update the second moment estimate, it can be calculated by the following formula:
where v is the second moment estimate, beta 2 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of the failure prediction model LSTM.
4. Deviation correction: to eliminate the initial bias of the first moment estimate and the second moment estimate, bias correction is performed, which can be calculated by the following formula:
where m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations.
5. Parameter updating: using the corrected first and second moment estimates to update the model parameters, the model parameters can be calculated by the following formula:
wherein θ is a model parameter; η is the learning rate which can be chosen by multiple experiments and with minimal training errors; epsilon is a constant for avoiding division by zero; m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations.
Repeating the step 3: and training and iterating the fault prediction model until the training times of the fault prediction model reach the preset iteration times or the model parameters of the fault prediction model reach the convergence condition.
Model evaluation of the failure prediction model is performed, and this embodiment adopts an Accuracy (Accuracy), a recall (recovery), a precision (precision), and an F1 score (F) 1_socre ) The four points evaluate the fault prediction model. The accuracy is one of common indexes in model evaluation, is used for measuring the accuracy degree of model prediction, and is required to be 90-95%. The calculation formula is as follows:
the recall measure model is required to be 0.7-0.9 for its ability to identify positive samples, also known as sensitivity or recall. The calculation formula is as follows:
the accuracy measures the accuracy of the model prediction positive sample, and the accuracy is required to be about 0.9. The calculation formula is as follows:
the F1 score is an index that comprehensively considers the accuracy and recall for evaluating the performance of the model. The method is a harmonic average value of the recall rate of the precision rate, and the requirement is 0.8-09. The calculation formula is as follows:
in the above, T P Is the number of real examples (model predicted as positive and actually positive), T N Is the number of true negative examples (model predicted as negative and actually negative), F P Is the number of false positive examples (model predicted as positive but actually negative), F N Is the number of false negatives (negative but actually positive samples predicted by the model).
And finally, verifying the trained fault prediction model by using verification data, evaluating the prediction capability of the fault prediction model, adjusting the fault prediction model according to the evaluation result, and selecting the optimal state prediction model according to the verification result.
S204: and generating a control instruction corresponding to the fault prediction result according to the fault prediction result.
In this embodiment, if the fault condition information is faulty and the occurrence probability value is greater than a preset threshold, determining whether the bus voltage and/or the bus positive-negative current meets a fault condition; the fault conditions include a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend; the second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold. If yes, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off; if not, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be closed.
It can be understood that if the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, whether the bus voltage and/or the bus positive and negative current meet a fault condition is determined. The fault conditions include a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend. The second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold.
In the present embodiment of the present invention,and setting a preset threshold value threshold for judging the reliability of the prediction result. If the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, the occurrence probability value is predicted fault_prob Greater than or equal to a preset threshold, consider the predicted outcome (i.e., to the failure predicted outcome) to be useable prediction Reliable (i.e., correct, accurate), the expression is as follows:
If the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, the predicted result (i.e. the failure predicted result) is considered to be available prediction Reliable (i.e. correct, accurate), it is determined whether the bus voltage and the bus positive-negative current meet fault conditions. The fault conditions include a first fault condition and a second fault condition; the first fault condition is the voltage trend U of the bus voltage of the system trend And a preset voltage trend U expected_trend Non-conforming (i.e. U trend !=U expected_trend ) Alternatively, the current trend I of the busbar anode and cathode current of the system trend And preset current trend I expected _trend Non-conforming (i.e. I trend !=I expected _trend ) Such as current trend I+ of bus positive current of system trend And preset positive current trend I + expected _trend Non-conforming current trend I-t of bus negative current of system rend And preset negative current trend I- expected _trend Non-conforming. The second fault condition includes at least one of: bus voltage U of the system current Is greater than a preset voltage upper threshold U threshold_hig (i.e. U) current >U threshold_hig ) Bus voltage U of the system current Less than a threshold U at a preset voltage threshold_low (i.e. U) current <U threshold_low ) The positive and negative current of the bus of the system is larger than the upper threshold value of the preset current (such as the positive current I of the bus of the system + current Is larger than a preset positive current upper threshold I + threshold_high I.e. I + current > I+ threshold_high Or bus negative current I- current Is greater than a preset anode current upper threshold I- threshold_high I.e.) current >I- threshold_high ). As shown in FIG. 3, i.e. fault condition fault detected The method comprises the following steps:
wherein, the upper threshold value U of the voltage is preset threshold_high =K ref1 U r The method comprises the steps of carrying out a first treatment on the surface of the Threshold U under preset voltage threshold_low = K ref2 U r The method comprises the steps of carrying out a first treatment on the surface of the Presetting an upper threshold value I+of the positive current threshold_high = K ref3 I+ r The method comprises the steps of carrying out a first treatment on the surface of the Presetting a threshold under the positive current: i- threshold_high = K re4 I- r The method comprises the steps of carrying out a first treatment on the surface of the Wherein U is r I+for bus voltage rating in the system r And I- r Respectively rated bus anode and cathode currents in the system; setting K ref1 The general value is 1.3-1.5; k (K) re2 The general value is 0.5-0.7; k (K) ref3 And K ref4 The value is generally 2-3.
If yes, namely the bus voltage or the bus positive and negative current accords with a fault condition, a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off is generated. It will be appreciated that if the fault determination is true (fault detected True), i.e. the busbar voltage or the busbar positive and negative current meets fault conditions, and the prediction result is reliable (i.e. the occurrence probability value is larger than a preset threshold value, and the busbar is available) prediction True), corresponding decision control operation is executed, and a turn-off signal is sent to drive the turn-off of the solid state switch so as to cut off the fault. And if not, namely the bus voltage or the bus positive and negative current does not accord with the fault condition, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be closed. It will be appreciated that if the fault is judged to be false (fault detected Is fault) is calledThe bus voltage or the bus positive and negative current accords with the fault condition, and the prediction result is reliable (namely the occurrence probability value is larger than the preset threshold value, and the fault is available) prediction And fault), corresponding decision control operation is executed, and a closing signal is sent to drive the solid state switch to be closed so as to continue operation.
S205: and controlling the switching state of a solid state switch in the solid state breaker control system according to the control instruction.
In this embodiment, according to the control instruction, a control signal corresponding to the control instruction may be sent to the solid-state switch. And if the control instruction is an off instruction, sending a control signal (namely an off signal) corresponding to the control instruction to the solid-state switch so as to control the switch state of the solid-state switch to be an off state. And if the control instruction is a closing instruction, sending a control signal (namely a closing signal) corresponding to the control instruction to the solid-state switch so as to control the switch state of the solid-state switch to be a closing state. In one implementation, the solid state switch may be a direct current solid state switch.
As can be seen from the above technical solution, the present application provides a solid state circuit breaker control method, which includes: detecting the bus voltage of a solid-state circuit breaker control system; detecting the current of the anode and the cathode of a bus of the solid-state circuit breaker control system; determining a fault prediction result according to the bus voltage and the bus anode and cathode current, and generating a control instruction corresponding to the fault prediction result according to the fault prediction result; and controlling the switching state of a solid state switch in the solid state breaker control system according to the control instruction. In this embodiment, since the system state prediction (i.e., the prediction of the fault prediction result) can be performed, the state change trend in a period of time in the future is obtained, and the fault prediction result is utilized to perform real-time fault prediction and decision, so as to optimize the control of the solid state switch, so that the solid state switch can quickly respond to the system state change of the solid state breaker control system, and the response speed and control accuracy of the solid state breaker control system are improved. That is, the embodiment can perform optimal control on the solid-state switch, so that the problems of switch shake, oscillation and the like in the traditional control mode can be avoided, the loss and heat of the system are reduced, and the stability and reliability of the system are improved. In addition, the embodiment can also monitor the positive and negative electrode current and the direct current voltage of the system in real time by adopting the current sensor and the voltage sensor, so that abnormal conditions in the system can be found in time, corresponding processing can be carried out, and the safety and the reliability of the system are improved. By combining the advantages, the solid-state circuit breaker control system of the embodiment has higher practicability and economic benefit, and can be widely applied to the field of direct current protection, such as power regulation, energy storage, electric automobile charging and the like in a power system. That is, the present embodiment can solve the problems existing in the conventional control method of the dc solid-state circuit breaker, such as slow fault response speed, switch jitter, oscillation, etc., and improve the stability and reliability of the system. Meanwhile, through model predictive control, the on-off control of the direct current solid state breaker can be optimally controlled, the system loss and heat are reduced, and the service life of the system is prolonged.
It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic of each process, and should not limit the implementation process of the embodiment of the present application in any way.
The integrated modules/units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application implements all or part of the flow in the methods of the above embodiments, or may be implemented by a computer program to instruct related hardware, and the computer program may be stored in a computer readable storage medium, where the computer program may implement the steps of the respective method embodiments described above when executed by a processor. The computer program may comprise computer program code, which may be in source code form, object code form, executable file or in some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth.
It should be noted that, in the present specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment is mainly described in a different point from other embodiments. The apparatus and system embodiments described above are merely illustrative, in which the units illustrated as separate components may or may not be physically separate. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
The foregoing is merely illustrative of the preferred embodiments, but the scope of the present application is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the technical scope of the present application should be included in the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (6)

1. A solid state circuit breaker control system, comprising a voltage sensor, a current sensor, a model predictive controller, a field programmable gate array FPGA, a driver and a solid state switch;
The voltage sensor is used for detecting bus voltage of the system;
the current sensor is used for detecting the current of the anode and the cathode of a bus of the system;
the model prediction controller is used for determining a fault prediction result according to the bus voltage and the bus anode and cathode current, and sending a control instruction corresponding to the fault prediction result to the field programmable gate array FPGA according to the fault prediction result;
the FPGA is used for sending a control signal corresponding to the control instruction to the solid-state switch through the driver according to the control instruction so as to control the switch state of the solid-state switch;
the model predictive controller comprises a fault predictive model; the model predictive controller is specifically configured to: inputting the bus voltage and the bus positive and negative current into the fault prediction model to obtain a fault prediction result, wherein the fault prediction result comprises fault condition information and an occurrence probability value corresponding to the fault condition information; if the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, judging whether the bus voltage and/or the bus positive and negative current meet fault conditions or not; if yes, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off, and sending the control instruction to the FPGA; if not, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be closed, and sending the control instruction to the FPGA; the fault prediction model is an LSTM model based on an Adam optimization algorithm;
The LSTM model based on the Adam optimization algorithm is generated as follows:
step 1, initializing weight and bias items of an LSTM model;
step 2, initializing first moment estimation and second moment estimation of an Adam algorithm;
step 3, training and iterating the fault prediction model; firstly, calculating forward propagation of an LSTM model by using the historical busbar voltage and the current of the positive electrode and the negative electrode of the historical busbar in the training set to obtain a fault prediction result and a loss function value; then, calculating the gradient of the loss function value to the model parameter of the LSTM model; next, the first moment estimate is updated: the first moment estimate is updated using an exponentially weighted average, calculated by the following formula:
where m is the first moment estimate, beta 1 Is the decay rate parameter in Adam algorithm, and grad is the loss function value to LSTM modelGradient of model parameters;
updating the second moment estimate: the second moment estimate is updated using an exponentially weighted average, calculated by the following formula:
where v is the second moment estimate, beta 2 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of LSTM;
step 4, deviation correction: deviation correction was performed, calculated by the following formula:
wherein m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
Step 5, updating model parameters of the LSTM by using the corrected first moment estimation and second moment estimation, wherein the model parameters are calculated by the following formula:
wherein θ is a model parameter; η is the learning rate which can be chosen by multiple experiments and with minimal training errors; epsilon is a constant for avoiding division by zero; m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
repeating the step 3: and training and iterating the fault prediction model until the training times of the LSTM model reach the preset iteration times or the model parameters of the LSTM model reach the convergence condition.
2. The solid state circuit breaker control system of claim 1, further comprising an inductance and a resistance; wherein the current sensor comprises a first current sensor and a second current sensor, the solid state switch comprises a first solid state switch and a second solid state switch, and the driver comprises a first driver and a second driver;
the first end of the voltage sensor is connected with the positive bus of the system, the second end of the voltage sensor is connected with the negative bus of the system, and the third end of the voltage sensor is connected with the model predictive controller;
The first end of the first current sensor is connected with the positive bus through the inductor, and the second end of the first current sensor is connected with the model predictive controller;
the first end of the second current sensor is connected with the negative bus, and the second end of the second current sensor is connected with the model predictive controller;
the model prediction controller is connected with a third end of the FPGA;
the first end of the FPGA is connected with the first end of the first driver, and the second end of the first driver is connected with the second end of the first solid-state switch; the first end of the first solid-state switch is connected with the positive electrode bus through the inductor, and the third end of the first solid-state switch is connected with one end of the resistor;
the second end of the FPGA is connected with the first end of the second driver, and the second end of the second driver is connected with the second end of the second solid-state switch; and the first end of the second solid-state switch is connected with the negative bus, and the third end of the second solid-state switch is connected with the other end of the resistor.
3. The solid state circuit breaker control system of claim 1, wherein the fault condition comprises a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend; the second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold.
4. A method of controlling a solid state circuit breaker, the method comprising:
detecting the bus voltage of a solid-state circuit breaker control system;
detecting the current of the anode and the cathode of a bus of the solid-state circuit breaker control system;
determining a fault prediction result according to the bus voltage and the bus anode and cathode current;
generating a control instruction corresponding to the fault prediction result according to the fault prediction result;
according to the control instruction, controlling the switching state of a solid state switch in the solid state breaker control system;
determining a fault prediction result according to the bus voltage and the bus anode and cathode current, including:
inputting the busbar voltage and the busbar anode and cathode current into a trained fault prediction model to obtain a fault prediction result, wherein the fault prediction result comprises fault condition information and an occurrence probability value corresponding to the fault condition information;
the fault prediction model is an LSTM model based on an Adam optimization algorithm;
the LSTM model based on the Adam optimization algorithm is generated as follows:
step 1, initializing weight and bias items of an LSTM model;
step 2, initializing first moment estimation and second moment estimation of an Adam algorithm;
Step 3, training and iterating the fault prediction model; firstly, calculating forward propagation of an LSTM model by using the historical busbar voltage and the current of the positive electrode and the negative electrode of the historical busbar in the training set to obtain a fault prediction result and a loss function value; then, calculating the gradient of the loss function value to the model parameter of the LSTM model; next, the first moment estimate is updated: the first moment estimate is updated using an exponentially weighted average, calculated by the following formula:
where m is the first moment estimate, beta 1 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of LSTM model;
updating the second moment estimate: the second moment estimate is updated using an exponentially weighted average, calculated by the following formula:
where v is the second moment estimate, beta 2 Is the decay rate parameter in Adam algorithm, grad is the gradient of the loss function value to the model parameter of LSTM;
step 4, deviation correction: deviation correction was performed, calculated by the following formula:
wherein m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
step 5, updating model parameters of the LSTM by using the corrected first moment estimation and second moment estimation, wherein the model parameters are calculated by the following formula:
Wherein θ is a model parameter; η is the learning rate which can be chosen by multiple experiments and with minimal training errors; epsilon is a constant for avoiding division by zero; m_hat is the corrected first moment estimate, v_hat is the corrected second moment estimate, and t represents the number of iterations;
repeating the step 3: and training and iterating the fault prediction model until the training times of the LSTM model reach the preset iteration times or the model parameters of the LSTM model reach the convergence condition.
5. The method according to claim 4, wherein generating the control command corresponding to the failure prediction result according to the failure prediction result comprises:
if the fault condition information is faulty and the occurrence probability value is greater than a preset threshold value, judging whether the bus voltage and/or the bus positive and negative current meet fault conditions or not;
if yes, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be turned off;
if not, generating a control instruction for controlling the switch state of the solid-state switch to be adjusted to be closed.
6. The solid state circuit breaker control method of claim 5, wherein the fault condition comprises a first fault condition and a second fault condition; the first fault condition is that the voltage trend of the bus voltage of the system does not accord with a preset voltage trend, or the current trend of the bus anode current and the bus cathode current of the system does not accord with a preset current trend; the second fault condition includes at least one of: the bus voltage of the system is larger than a preset voltage upper threshold, the bus voltage of the system is smaller than a preset voltage lower threshold, and the current of the positive electrode and the negative electrode of the bus of the system is larger than a preset current upper threshold.
CN202311189779.1A 2023-09-15 2023-09-15 Solid state circuit breaker control system and solid state circuit breaker control method Active CN116914685B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311189779.1A CN116914685B (en) 2023-09-15 2023-09-15 Solid state circuit breaker control system and solid state circuit breaker control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311189779.1A CN116914685B (en) 2023-09-15 2023-09-15 Solid state circuit breaker control system and solid state circuit breaker control method

Publications (2)

Publication Number Publication Date
CN116914685A CN116914685A (en) 2023-10-20
CN116914685B true CN116914685B (en) 2023-12-22

Family

ID=88353539

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311189779.1A Active CN116914685B (en) 2023-09-15 2023-09-15 Solid state circuit breaker control system and solid state circuit breaker control method

Country Status (1)

Country Link
CN (1) CN116914685B (en)

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108830745A (en) * 2018-06-29 2018-11-16 国网上海市电力公司 Power grid cascading fault diagnosis, early warning, assessment system based on monitoring information
CN110336270A (en) * 2019-04-22 2019-10-15 清华大学 A kind of update method of Power system transient stability prediction model
CN110829417A (en) * 2019-11-14 2020-02-21 电子科技大学 Electric power system transient stability prediction method based on LSTM double-structure model
CN112821354A (en) * 2020-12-31 2021-05-18 国网江苏省电力有限公司电力科学研究院 Direct-current transformer solid-state switch circuit, air-cooled solid-state switch module, and fault protection system and method
CN113376516A (en) * 2021-06-07 2021-09-10 科润智能控制股份有限公司 Medium-voltage vacuum circuit breaker operation fault self-diagnosis and early-warning method based on deep learning
CN113379279A (en) * 2021-06-23 2021-09-10 黑龙江大学 Deep reinforcement learning short-term voltage stability evaluation method based on incomplete data
CN113837443A (en) * 2021-08-26 2021-12-24 国网河北省电力有限公司邢台供电分公司 Transformer substation line load prediction method based on depth BilSTM
CN114839523A (en) * 2022-04-12 2022-08-02 国网黑龙江省电力有限公司电力科学研究院 Circuit breaker fault detection system and method based on CNN combined LSTM algorithm
CN116087667A (en) * 2023-03-09 2023-05-09 国网安徽省电力有限公司超高压分公司 LSTM-based relay protection fault detection method for extra-high voltage direct current transmission line
CN116436327A (en) * 2023-04-12 2023-07-14 山东大学 Adaptive prediction control method and system for converter based on neural network
CN116432006A (en) * 2023-05-04 2023-07-14 国网河南省电力公司安阳供电公司 Photovoltaic grid-connected inverter fault diagnosis method based on CEEMDAN-SE-IHHO-LSTM model

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11101641B2 (en) * 2018-09-24 2021-08-24 S&C Electric Company Fault protection devices and related reclosing methods

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108830745A (en) * 2018-06-29 2018-11-16 国网上海市电力公司 Power grid cascading fault diagnosis, early warning, assessment system based on monitoring information
CN110336270A (en) * 2019-04-22 2019-10-15 清华大学 A kind of update method of Power system transient stability prediction model
CN110829417A (en) * 2019-11-14 2020-02-21 电子科技大学 Electric power system transient stability prediction method based on LSTM double-structure model
CN112821354A (en) * 2020-12-31 2021-05-18 国网江苏省电力有限公司电力科学研究院 Direct-current transformer solid-state switch circuit, air-cooled solid-state switch module, and fault protection system and method
CN113376516A (en) * 2021-06-07 2021-09-10 科润智能控制股份有限公司 Medium-voltage vacuum circuit breaker operation fault self-diagnosis and early-warning method based on deep learning
CN113379279A (en) * 2021-06-23 2021-09-10 黑龙江大学 Deep reinforcement learning short-term voltage stability evaluation method based on incomplete data
CN113837443A (en) * 2021-08-26 2021-12-24 国网河北省电力有限公司邢台供电分公司 Transformer substation line load prediction method based on depth BilSTM
CN114839523A (en) * 2022-04-12 2022-08-02 国网黑龙江省电力有限公司电力科学研究院 Circuit breaker fault detection system and method based on CNN combined LSTM algorithm
CN116087667A (en) * 2023-03-09 2023-05-09 国网安徽省电力有限公司超高压分公司 LSTM-based relay protection fault detection method for extra-high voltage direct current transmission line
CN116436327A (en) * 2023-04-12 2023-07-14 山东大学 Adaptive prediction control method and system for converter based on neural network
CN116432006A (en) * 2023-05-04 2023-07-14 国网河南省电力公司安阳供电公司 Photovoltaic grid-connected inverter fault diagnosis method based on CEEMDAN-SE-IHHO-LSTM model

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于马尔科夫模型的直流断路器可靠性评估方法;赵书涛等;《电工技术学报》;第34卷;第126-130页 *

Also Published As

Publication number Publication date
CN116914685A (en) 2023-10-20

Similar Documents

Publication Publication Date Title
Bayati et al. Localized protection of radial DC microgrids with high penetration of constant power loads
CN101542662B (en) A method and an apparatus for predicting the future behavior of currents in current paths
KR101467249B1 (en) Apparatus and method for adaptive auto-reclosing based on transient stability
Mirshekali et al. Linear and nonlinear fault location in smart distribution network under line parameter uncertainty
JP2008529227A (en) Method and apparatus for determining the closing time of an electrical switchgear
Li et al. A permanent fault identification method for single-pole grounding fault of overhead transmission lines in VSC-HVDC grid based on fault line voltage
EP1854113B1 (en) An apparatus and a method for predicting a fault current
Yari et al. A comprehensive assessment to propose an improved line stability index
CN101521407B (en) Three-segment out-of-step self-adaptive disconnecting method
Zhang et al. An application‐oriented multistate estimation framework of lithium‐ion battery used in electric vehicles
CN116914685B (en) Solid state circuit breaker control system and solid state circuit breaker control method
Jarrahi et al. Fault detection in DC microgrid: A transient monitoring function-based method
Mobeen et al. A Generalized Modeling and Analysis for Transformer Parameterization
Xiong et al. Peak power estimation of vanadium redox flow batteries based on receding horizon control
JP2012170168A (en) Parameter estimation method for power system abridged model, and apparatus using the same
CN111240306A (en) Self-adaptive distribution transformer fault diagnosis system and diagnosis method thereof
KR101862029B1 (en) Adaptive Reclosing Method using Second-Order Difference of THD in Distribution System with BESS used as Uninterruptible Power Supply
Tejeswara Rao et al. Prevention of maloperation of distance relay under severe stressed conditions for series compensated transmission line considering optimal placement of phasor measurement units
Sebastian et al. A novel PSO based fuzzy controller for robust operation of solid-state transfer switch and fast load transfer in power systems
Li et al. A simultaneous phase sequence exchange control method based on transient energy function
Zonkoly et al. Optimal placement of PMUs using improved tabu search for complete observability and out-of-step prediction
CN109742845B (en) Rapid switching-on method of power supply rapid switching-off device based on optimal phase
CN115128442A (en) Dynamic evaluation method for electrical life of circuit breaker based on full-life operation information
CN115549454A (en) Converter bridge arm junction temperature balancing method and device, electronic equipment and storage medium
CN114552621A (en) Follow-up commutation failure early warning method and system for converter station considering relay protection action behavior

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant