Power converter real-time state monitoring method based on heterogeneous embedded digital twinning

Abstract

The invention discloses a power converter real-time state monitoring method based on heterogeneous embedded digital twinning. The method comprises the following steps: step 1, establishing a power converter embedded digital twinning control system; step 2, realizing real-time simulation of the power converter based on the FPGA; and step 3, realizing a processor-based power converter digital twinning mechanism model updating algorithm. A power converter control algorithm and a model updating algorithm are realized in a processor based on a processor + FPGA heterogeneous mode on the basis of not changing an original power converter circuit structure, the FPGA is responsible for real-time simulation tasks and virtual-real data interaction of a converter digital twin mechanism model, and the FPGA is responsible for real-time simulation tasks and virtual-real data interaction of the converter digital twin mechanism model. Therefore, real-time state monitoring of the real power converter is realized. According to the method, the performance advantages of the processor and the FPGA are brought into full play, real-time updating and dynamic evolution of production, management and operation and maintenance in the whole life cycle of the power converter are facilitated, and the digitization degree of the whole life cycle of the power converter is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment status monitoring, and in particular to a real-time status monitoring method for a power converter based on an embedded digital twin of a processor + FPGA heterogeneous mode. Background Art

[0002] With the continuous development of science and technology and the rapid innovation of information technology, modern power electronic devices have placed higher demands on system performance, functional completeness, and environmental adaptability. The integration, intelligence, and complexity of these systems are also increasing, and the probability of failure is also increasing. Therefore, as power converter devices play a key role in power electronic circuits, monitoring the health of power converter devices has become particularly important. In power electronic circuits, power converters not only ensure the stability and reliability of power supply, but also meet the different power requirements of various devices and systems in the circuit, improve energy utilization efficiency, and play a vital role in the safety, economic, and environmental performance of the circuit.

[0003] Currently, existing power converter condition monitoring methods have numerous shortcomings, making them difficult to meet the high stability requirements of current power electronic circuits. For example, regular inspections are cyclical, preventing real-time monitoring and creating blind spots. They rely on engineers' experience and judgment, which can easily lead to misjudgments and prevent accurate predictions of potential equipment failures and remaining lifespan. Furthermore, data from disparate devices and sensors lacks effective integration, making it difficult to form a holistic health assessment. With the advancement of industrial digital transformation, scholars and experts at home and abroad have conducted in-depth research, exploration, and discussion on digital twin technology and its applications.

[0004] The simplest understanding of a digital twin is the mapping of a physical entity into a virtual space, and the digital twin evolves in tandem with the physical entity's development, maturity, and degradation. With the help of digital twins, researchers can analyze, predict, and diagnose physical entities, and implement system optimization and intelligent decision-making. Traditional power electronic circuit fault diagnosis methods are constrained by the measurement and computing resources of the circuit itself, making it difficult to guarantee the accuracy of power converter condition monitoring results. However, digital twins can perceive real physical entities and construct high-fidelity models. This allows for real-time and accurate estimation of converter status without the need for excessive additional hardware circuitry. Furthermore, by combining historical converter health data, the converter's current health status can be estimated. Therefore, digital twins are particularly suitable for real-time power converter status monitoring scenarios.

[0005] To sum up, the existing technology has the technical problem that the power converter status monitoring process does not simulate the power converter, analyze and monitor the health status of the power converter in real time, resulting in low power converter monitoring efficiency and lag, resulting in poor accuracy of power converter health prediction. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to address the deficiencies in the above-mentioned prior art and provide a real-time status monitoring method for power converters based on heterogeneous embedded digital twins. The invention aims to provide an FPGA-based power converter digital twin control system for realizing complex and real-time power converter digital twin applications, including:

[0007] The physical space includes a power converter circuit, i.e., a physical twin. The power converter circuit includes an input power supply, an inductor, a capacitor, a switch, a freewheeling diode, and a load.

[0008] The digital space includes a processor and an FPGA processing unit. The processor is responsible for model algorithm updates and data synchronization. Model algorithm updates are based on historical data and adjust the mathematical model of the power converter to optimize control parameters. The FPGA processing unit is mainly responsible for building a high-precision model of the digital twin and parallel solving.

[0009] Human-machine interface (HMI), connected to the FPGA processing unit, used to display operating parameters, receive user input, and adjust control parameters;

[0010] Communication interface, used for FPGA to interact with real physical data of external devices. Used for data acquisition, signal processing and control output;

[0011] The power converter circuit is a Buck-Boost converter, the switching signal of the switch tube is provided by a processor, and its duty cycle is adjusted according to a control algorithm of a controller in the processor.

[0012] Among them, the FPGA processing unit includes a high-precision model of the digital twin and a model solving module.

[0013] The communication interface supports remote data transmission and provides monitoring and control instruction issuance functions.

[0014] The human-machine interface includes a display module and an input module. The display module is used to present the working status of the power converter in real time, and the input module is used to receive control parameters set by the user and transmit them to the FPGA processing unit.

[0015] The entire framework can be divided into physical and digital spaces, with the heterogeneous processor + FPGA architecture being the key architecture of the digital space. This architecture allows the power converter control algorithm and model update algorithm to be implemented in the processor without changing the original power converter circuit structure. The FPGA is responsible for the real-time simulation of the converter's digital twin mechanism model and the interaction between virtual and real data, thereby achieving real-time status monitoring of the actual power converter. The method is well-designed, easy to implement, and non-invasive, meaning that it can be implemented on the existing power converter physical platform without the introduction of additional hardware.

[0016] To solve the above problems, the technical solution adopted by the present invention includes the following steps:

[0017] Step 1: Design of heterogeneous embedded digital twin control system for power converters;

[0018] Step 2: FPGA-based real-time simulation of power converter;

[0019] Step 3: Implementing a processor-based power converter digital twin mechanism model update algorithm;

[0020] In the above-mentioned method for real-time status monitoring of power converters based on heterogeneous embedded digital twins, the steps of establishing the embedded digital twin control system of the power converter in step 1 are as follows:

[0021] Step 101: In the physical space, the power converter hardware platform of the state monitoring object is used as a physical twin, and the drive control signal of the physical twin is provided by the processor. At the same time, the physical twin reserves the inductor current i L and the output voltage v o Sampling interface for FPGA sampling input.

[0022] Step 102: The inductor current i obtained by sampling the physical twin of the power converter is L Measured value and output voltage v o The measured values are used as matrix input I. Let the physical twin of the power converter operate under steady-state conditions. Each sampling period t s The sampled inductor current and output voltage are respectively represented by i L,0 ,i L,1 ,…i L,N and v o,0 ,v o,1 ,…v o,N N represents the number of sampled data, that is, the sample size of the measured data, which is expressed as follows:

[0023]

[0024] Among them, h is the discrete solution step size of the digital twin mechanism model; n is the total number of iterations required to execute the digital twin parameter identification and update algorithm once.

[0025] Step 103: In the digital space, refer to the power converter hardware platform design indicators, reserve resource space in the FPGA, and build a real-time simulation model of the power converter, namely the digital twin. In addition, the FPGA interface is also used to establish a virtual-real data interaction channel to receive the inductor current i collected in the physical space. L and the output voltage v o data sets and transmits the driving control signals of the processor in the digital space to the physical twin.

[0026] Step 104: The calculated values of the inductor current and the output voltage obtained by the real-time simulation model (digital twin) of the power converter are used as matrix input quantities II. The matrix input quantities II are represented by i Lm,0 ,i Lm,1 ,…i Lm,N and v om,0 ,v om,1 ,…v om,N Matrix input quantities I and II are used as inputs to the real-time update algorithm of the processor model to perform real-time status monitoring of the power converter hardware platform (physical twin).

[0027] In the above-mentioned real-time state monitoring method of power converter based on heterogeneous embedded digital twin, the inductor current measurement value i in step 102 is Lm,n and the output voltage measurement value v om,n It is necessary to sample analog quantities through an FPGA interface with highly parallel and pipelined structure to compare it with the inductor current calculated by the power converter digital twin mechanism model. L,n And the output voltage calculated value v o,n Maintain time synchronization. Where n = 0,…,N. N represents the sample size of the measurement data. The higher the value, the more accurate the condition monitoring effect.

[0028] In the above-mentioned real-time status monitoring method for power converters based on heterogeneous embedded digital twins, the steps for implementing the real-time simulation of power converters based on FPGA in step 2 are as follows:

[0029] Step 201: Establish a digital twin mechanism model of the power converter. The modeling object is the BUCK-BOOST power converter circuit. When operating in the BUCK operating mode, the digital twin mechanism model is represented in state space form and can be expressed as:

[0030]

[0031] When working in BOOST mode, the digital twin mechanism model can be expressed as:

[0032]

[0033] Among them, R dson 、r L 、R C are the parasitic resistances of power MOSFET, inductor, and capacitor respectively; i L is the inductor current; V in is the input voltage; R is the equivalent load resistance; L and C are the capacitance and inductance values respectively; D is the duty cycle, generated by the controller in the processor; v o is the output voltage; v c is the capacitor voltage; V f is the forward conduction voltage of the diode. The following analysis and discussion will take the BOOST operating mode as an example.

[0034] Step 202: Use the implicit Euler method to solve the BOOST power converter digital twin mechanism model. The implicit Euler method solution formula can be expressed as:

[0035]

[0036] Among them, h is the discrete solution step of the digital twin mechanism model; i L,n+1 and i L,n is the exact solution L (t n+1 ) and i L (t n ) approximate value; t n Represents the nth th Calculated position; t n+1 =t n +h represents the next calculation position.

[0037] Output voltage v o Using discrete time steps, this can be expressed as:

[0038]

[0039] So, the (n+1)th th The output voltage value v o,n+1 You can pass the current nth th The inductor current i L,n and capacitor voltage v c,n Calculate and get it. Discretize the above formula and it can be expressed as:

[0040]

[0041] Therefore, the matrices are defined as follows:

[0042]

[0043] Where, is defined as the observation matrix; θ1 and θ2 are defined as parameter estimation matrices, and the system parameters P1(L, C, R L ,R c , R) have an algebraic relationship. Based on the algebraic relationship shown in the above formula, parameter identification can be performed through linear regression analysis methods, and finally the estimated value of the system parameter P1 can be obtained through iterative calculation.

[0044] Step 203: Use the values of the inductor current, capacitor voltage, output voltage, and voltage error signal of the next discrete time step calculated by the BOOST power converter digital twin mechanism model as the input of the current discrete time step to implement closed-loop control. The implementation of closed-loop control can be expressed as:

[0045]

[0046] Among them, v e,n+1 Shows voltage error signal, V ref is the voltage reference value;

[0047] The above-mentioned real-time status monitoring method for power converters based on heterogeneous embedded digital twins is implemented in a processor and is real-time monitoring. Step 3 of the processor-based power converter digital twin mechanism model update algorithm is implemented as follows:

[0048] Step 301: Initialize the model update algorithm.

[0049] Initialize the inverse of the covariance matrix P(0) = α 2 I, where α is an infinite integer, usually 10 5 ~10 8 ; I is the identity matrix;

[0050] Initialize the parameter estimation matrix Usually it can be a very small value or a zero vector; the initial value P(0) and The effect of decreases as the number of recursions increases.

[0051] Step 302: Update sample data x n (t) and the observation matrix Where n = 1, 2; x1(t) = i L (t), x2(t)=u0(t), inductor current i L The output voltage u0 is obtained through FPGA embedded sampling; the observation matrix Calculated according to the formula in step 202, the entire process is implemented in the digital space.

[0052] Step 303: Calculate the objective function. L 、v c The dataset i obtained by sampling with the physical twin Lm 、v cm As two sets of matrix input quantities I and II, and calculate the objective function f obj , as the evaluation basis for the iterative results of the model update algorithm. The objective function f obj The calculation formula is:

[0053]

[0054] Among them, i L,j and v o,j are the inductor current and output voltage calculated from the BOOST power converter digital twin mechanism model, defined as matrix I. Lm,j and v om,j is the real measurement data from the BOOST power converter hardware platform, defined as matrix II, and N is the sample size of the measurement data.

[0055] Step 304: Optimize the parameter estimation matrices θ1 and θ2. After calculating the objective function, perform crossover and mutation operations on the parameter estimation matrices θ1 and θ2.

[0056] The crossover operation uses single-point and two-point crossover operations. Before the crossover operation, according to the objective function f obj The size of the parameter estimation matrix is divided into three tribes. The crossover rules are as follows: the third-class tribe uses a two-point crossover with other tribes (first and second class); the second-class tribe uses a single-point crossover with tribes other than the third-class (second and first class); and the first-class tribe uses a single-point crossover within itself.

[0057] The mutation operation multiplies the matrix elements in the tribe by a reasonable random number to improve the tribe's diversity. The mutation rules are: for third-class tribes, a global mutation method is used; for second-class tribes, a single-point mutation method is used; and for first-class tribes, no mutation operation is performed.

[0058] Step 305: Model parameter adaptation. If the objective function f obj If the value is greater than the threshold, the model parameter adaptive process (i.e., recursive least squares genetic optimization algorithm) is iteratively updated, and the digital twin model parameter P1 is updated. Otherwise, the optimization process ends and the parameter estimation matrix θ is output. n , after decoupling, we get the system parameter P1. The parameter adaptation process is shown in the following formula:

[0059]

[0060] Where k(t) is defined as the gain matrix; λ is a very small integer used to ensure the reversibility of the matrix, and is set to 1. n = 1, 2; x1(t) = i L (t), x2(t)=v0(t). Assume that the process parameter matrix obtained by identification is:

[0061]

[0062] According to step 202, the process parameter matrices θ1, θ2 and the system parameters P1 (L, C, R L ,R c ,R), we can get the decoupling result of the above formula, that is, the system parameters P1(L,C,R L ,R c ,R) is:

[0063]

[0064] Step 306: Feedback of status monitoring results. The final parameter identification result P1(L, C, R L ,R c ,R) is transmitted to the host computer monitoring platform as one of the real-time status monitoring elements of the BOOST power converter hardware platform (physical twin) for the decision-making link to make corresponding decisions.

[0065] The aforementioned method for real-time power converter status monitoring based on heterogeneous embedded digital twins is a virtual-physical fusion system designed specifically for power converters and integrated on a dedicated computing platform. This system requires a high-precision digital twin model that runs in real time and has the ability to be dynamically updated. A heterogeneous approach based on a processor and FPGA is key to the implementation of this complex system. The processor's excellent algorithm processing capabilities enable the implementation of the power converter's physical twin control algorithm and the digital twin model update algorithm, while the FPGA's parallel computing capabilities provide the foundation for small-step, high-precision real-time simulation of power converters.

[0066] Compared with the prior art, the present invention has the following advantages:

[0067] The technical solution of the present invention addresses the shortcomings of existing technologies for real-time status monitoring of power converters. Existing methods suffer from significant deficiencies in monitoring accuracy, real-time fault response, and model solution efficiency, making them incapable of meeting the requirements for high-precision and real-time monitoring of power converters in complex dynamic environments. These deficiencies make it difficult to achieve stable and reliable operation in practical applications, impacting overall performance and safety.

[0068] To address the above technical issues, the present invention proposes a real-time state monitoring method for power converters based on heterogeneous embedded digital twins. This method constructs a high-precision digital twin control system by combining a heterogeneous computing architecture of a processor and a field-programmable gate array (FPGA), enabling real-time simulation and state monitoring of power converters. The digital twin can efficiently interact with the physical twin, dynamically updating the digital twin mechanism model using a recursive least squares genetic optimization algorithm to ensure the accuracy and synchronization of the physical twin's monitoring data, thereby significantly improving the stability of the entire system.

[0069] This invention has achieved significant technological advancements, significantly improving the accuracy and response speed of real-time status monitoring of power converters while effectively reducing computational complexity. Through a heterogeneous embedded digital twin system, efficient monitoring and precise management of power converters in dynamic and complex environments are achieved, enhancing the stability and reliability of the system. Furthermore, the recursive least squares genetic optimization algorithm employed enhances the model's adaptive capabilities, enabling the monitoring system to rapidly respond to environmental changes, meeting the demand for high-performance monitoring in fields such as intelligent power systems and industrial automation, and promoting the integrated development of the vehicle-mounted metaverse and edge computing technologies.

[0070] The proposed method for real-time status monitoring of power converters based on heterogeneous embedded digital twins is rationally designed, easy to implement, and non-invasive. This method, based on the existing physical platform of the power converter, does not require the introduction of additional hardware and achieves efficient real-time monitoring through the collaborative work of the processor and FPGA. Furthermore, the heterogeneous computing architecture of the processor and FPGA employed not only ensures high model accuracy and rapid solution capabilities, but also reduces the real-time simulation step size to the "hundreds of nanoseconds" level, making the status monitoring effect even more significant.

[0071] The embedded digital twin architecture of this invention enables efficient information exchange between the physical characteristics and degradation information of the physical model and the virtual power converter model. By evaluating changes in power converter fault widths and degradation trends of key components, the method facilitates real-time updates and dynamic evolution of production, management, and maintenance throughout the power converter's lifecycle. This not only improves the digitization of the power converter's lifecycle but also enhances the system's ability to predict and prevent faults, improving the reliability and stability of overall operation.

[0072] Furthermore, the present invention can be effectively applied to various real-time status monitoring scenarios for power converters, achieving a high degree of consistency between real-time monitoring results and actual operating results, resulting in significant results and facilitating widespread adoption and application. Through its high-precision digital twin model and optimized monitoring algorithm, the system demonstrates broad application prospects in fields such as intelligent power systems and industrial automation. It can meet the demand for high-precision and real-time monitoring of power converters in complex dynamic environments, promoting the further development and integration of related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 Flowchart of the real-time status monitoring method of power converter based on heterogeneous embedded digital twins of the present invention;

[0074] Figure 2 A block diagram for building an embedded digital twin architecture for power converters;

[0075] Figure 3 The equivalent circuit diagram of the on and off states of the buck-boost power converter object;

[0076] Figure 4 Updated algorithm flow chart for processor-based power converter digital twin mechanism model;

[0077] Figure 5 Schematic diagram of parameter estimation matrix classification and crossover operation;

[0078] Figure 6 Schematic diagram of parameter estimation matrix mutation operation;

[0079] Figure 7 It is the waveform diagram of the objective function convergence process;

[0080] Figure 8 The waveform diagram of the online identification result of the parameter set;

[0081] Figure 9 The figure compares the output results of the digital twin mechanism model of the power converter and the physical power converter. DETAILED DESCRIPTION

[0082] Specific implementation method 1: Combination Figures 1 to 9 The detailed structure and working principle of the embodiment of the present invention include:

[0083] (1) Overview of the overall architecture

[0084] This system consists of physical and digital spaces, with processors and FPGAs (field programmable gate arrays) as its core, combined with power converter circuits and real-time simulation platforms to achieve modeling, optimization, monitoring, and decision-making for power electronic converters. It mainly includes the following modules:

[0085] - Physical space (Buck-Boost converter): Contains inductors, capacitors, switches, and loads, which are the actual circuit components of the power converter.

[0086] -Digital space (processor and FPGA control unit): The FPGA is mainly used to monitor the physical circuit status in real time and perform real-time simulation calculations of the digital model; the processor is mainly responsible for the power converter control algorithm, data calculation module, and model update module for decision-making and optimization.

[0087] -Human-machine interface: Provides interaction, displays system status, and allows users to adjust parameters.

[0088] -Communication interface, used for FPGA to interact with real physical data of external devices. Used for data acquisition, signal processing and control output;

[0089] (2) Physical Space (Buck-Boost Converter)

[0090] In the physical space, the core is the Buck-Boost power converter, which is used for voltage step-up and step-down conversion. The main components include:

[0091] -Input voltage source: provides DC input to the system.

[0092] -Inductor L: used for energy storage and current smoothing.

[0093] -Switches (Q1, Q2): Controlled by the processor, they determine the circuit's boost or buck mode.

[0094] -Diode: Provides a freewheeling path to prevent current from flowing in the reverse direction.

[0095] -Output capacitor C2: used to smooth the output voltage and reduce ripple.

[0096] -Load R_L: represents the target device that the system ultimately powers.

[0097] The Buck-Boost converter can provide different output voltages under different duty cycle control and is suitable for a variety of power electronic applications.

[0098] (3) FPGA control unit (digital space)

[0099] As a secondary processing unit, FPGA mainly undertakes the following tasks:

[0100] -Data acquisition: Through analog-to-digital conversion (ADC), state parameters such as voltage and current are acquired in real time.

[0101] -Power Converter Digital Twin Model: Through mathematical modeling and real-time simulation, a highly accurate digital model of the converter is constructed to facilitate real-time simulation and further solutions.

[0102] -Power converter model solution: By performing electrical network decomposition and parallel simulation calculations on the converter's high-precision digital model, real-time simulation calculations of the digital model are achieved.

[0103] -Communication interface (FPGA I / O): Provides the ability to interact with external devices and supports remote monitoring and data storage.

[0104] (4) Processor (digital space)

[0105] As the core processing unit, the processor is mainly responsible for internal data processing and algorithm optimization, and mainly undertakes the following tasks:

[0106] -Digital power converter control: Calculates optimal switching control signals to optimize energy transfer efficiency.

[0107] -Data analysis and storage: Real-time recording of converter operating status for subsequent analysis.

[0108] -Model update module: Optimize the digital twin model based on historical data to improve simulation accuracy.

[0109] -Optimization, monitoring and decision-making: Integrate physical and digital space data to adjust control parameters and improve system performance.

[0110] (5) Human-machine interface (HMI)

[0111] -Display circuit operating status through augmented reality (AR) or visual dashboards.

[0112] - Allows users to modify control parameters such as duty cycle, switching frequency, etc., affecting output characteristics.

[0113] - Real-time monitoring and alarm, when abnormal conditions (such as overcurrent, short circuit, etc.) are detected, the user will be automatically prompted or protective measures will be triggered.

[0114] This system adopts the approach of physical simulation + digital twin, and through the intelligent collaboration of processor + FPGA, it monitors the operating status of the power converter in real time, realizing efficient, safe and intelligent power conversion management. It is suitable for various application scenarios such as electric vehicle charging and renewable energy management.

[0115] The method for real-time status monitoring of a power converter based on heterogeneous embedded digital twins described in this embodiment includes the following steps:

[0116] Step 1: Design of heterogeneous embedded digital twin control system for power converters;

[0117] Step 2: FPGA-based real-time simulation of power converter;

[0118] Step 3: Implementation of the processor-based power converter digital twin mechanism model update algorithm.

[0119] The heterogeneous embedded digital twin control system for power converters is designed through a combination of hardware and software, comprising a field-programmable gate array (FPGA), a processor, and an embedded control module. The FPGA is used for high-precision real-time simulation, the processor is responsible for computing complex algorithms and updating the twin model, and the embedded control module coordinates data exchange between the actual system and the digital twin. This heterogeneous architecture enables efficient parallel computing and multi-threaded processing, ensuring real-time performance and accuracy.

[0120] The FPGA, serving as the system's real-time simulation core, implements the power converter circuit model using a hardware description language and accelerates the simulation process through parallel computing. Key power converter state parameters (such as current, voltage, and temperature) are collected in real time via sensors and input into the FPGA for rapid processing. The FPGA compares the real-time simulation results with actual system operating data to verify the accuracy of the simulation model and provide a basis for updating the twin model.

[0121] The processor is responsible for dynamically updating the digital twin model. It uses algorithms to analyze the errors between the actual system and the digital twin and adjust the twin model's parameters (such as inductance, capacitance, and load characteristics) to improve model accuracy. The model update algorithm can utilize techniques such as recursive least squares, neural networks, and particle swarm optimization, combining historical and real-time data to continuously optimize the twin model and ensure it accurately reflects the actual operating status of the power converter.

[0122] The digital twin system combines real-time simulation with mechanistic models to monitor the condition of power converters. Real-time collected system operating data is compared with the output of the twin model to identify abnormal conditions and potential faults. Monitoring results are transmitted to a host computer or cloud platform via an embedded control module, providing decision-making support for operations and maintenance personnel. The system also supports dynamic adjustment of control strategies to adapt to complex operating conditions and ensure efficient and stable operation of the power converter.

[0123] The real-time status monitoring method of power converter based on heterogeneous embedded digital twins described in this embodiment is as follows: Figure 1 As shown in Figure 1, it includes three main steps. Step 1 is to establish an embedded digital twin control system for the power converter, see Figure 2 The system mainly consists of physical twins and digital twins. The physical twin includes the actual power converter hardware platform, whose drive control signal is provided by the processor, and the inductor current and output voltage data are collected through the reserved sampling interface, such as Figure 3 The equivalent circuit diagrams for the power converter object in the on and off states are shown in the figure. These sampled data, serving as matrix input quantity I, are sampled at each sampling cycle under steady-state conditions to ensure data continuity and accuracy. Simultaneously, in the digital space, the FPGA reserves resources to build a real-time simulation model of the power converter and establish a virtual-real data exchange channel to enable data transmission and synchronization between the physical twin and the digital twin.

[0124] Step 2 is to implement real-time simulation of power converter based on FPGA, such as Figures 4 to 6 As shown. Based on the design indicators of the BUCK power converter circuit, FPGA uses the state space form to build a digital twin mechanism model, and uses the implicit Euler method for real-time solution. FPGA calculates the inductor current and output voltage values in real time, and uses these calculation results as matrix input quantities II. Through highly parallel and pipeline structures, FPGA ensures high-speed sampling and calculation of analog quantities, enabling the digital twin to complete simulation steps within the "hundred nanoseconds" level, significantly improving the accuracy and response speed of real-time simulation. Figures 5 and 6 As shown in Figure 3, parameter estimation matrix classification, crossover operations, and mutation operations further optimize the computational efficiency and accuracy of the simulation model. The establishment of a virtual-real data interaction channel enables strict temporal synchronization between the physical twin’s sampled data and the digital twin’s calculated data, ensuring the consistency and reliability of the monitoring data.

[0125] Step 3 is to implement the processor-based power converter digital twin mechanism model update algorithm, such as Figure 4 The processor executes the recursive least squares genetic optimization algorithm to update the digital twin mechanism model in real time, including initializing the model update algorithm, updating the sample data x n (t) and the observation matrix There are six sub-steps: calculating the objective function, optimizing the parameter estimation matrices θ1 and θ2, adapting the model parameters, and feeding back the status monitoring results. Figure 7 As shown in the figure, the convergence process of the objective function is displayed through the waveform diagram to ensure the effectiveness of the optimization process. Figure 8 The waveform of the parameter set online identification result is shown to verify the accuracy of the model update algorithm. Figure 9As shown in the figure, the output results of the digital twin mechanism model of the power converter are compared with those of the physical power converter, which proves the accuracy and high reliability of this method in real-time state monitoring. The optimized parameter identification results are transmitted to the host computer monitoring platform through the communication interface to complete the real-time state monitoring of the power converter and ensure the stable operation of the system in a dynamic and complex environment. In order to realize the embedded digital twin architecture based on the processor + FPGA heterogeneous mode proposed in the present invention, taking the BOOST converter as an example, the NI cRIO-9038 multi-core controller is selected as the basis to build an embedded digital twin heterogeneous platform. The platform configuration is 1.33GHz dual-core CPU, 2GB DRAM, Kintex-7160T FPGA, and 8-slot Compact-RIO external board. At the same time, the model update algorithm in the processor is implemented with the help of recursive least squares genetic optimization algorithm.

[0126] Specific implementation method 2: Combination Figures 1 to 9 This embodiment describes a real-time state monitoring method for a power converter based on heterogeneous embedded digital twins. Figure 2 As shown in the figure, a CPU+FPGA heterogeneous approach is used to perform real-time status monitoring of the power converter. The steps for establishing an embedded digital twin control system for the power converter in step 1 are as follows:

[0127] Step 101: In the physical space, the power converter hardware platform of the state monitoring object is used as a physical twin, and the drive control signal of the physical twin is provided by the processor. At the same time, the physical twin reserves the inductor current i L and the output voltage v o Sampling interface for FPGA sampling input.

[0128] Step 102: The inductor current i obtained by sampling the physical twin of the power converter is L Measured value and output voltage v o The measured values are used as matrix inputs I. Let the physical twin of the power converter operate under steady-state conditions. The inductor current and output voltage sampled in each sampling period ts are represented by i L,0 ,i L,1 ,…i L,N and v o,0 ,v o,1 ,…v o,N N represents the number of sampled data, that is, the sample size of the measured data, which is expressed as follows:

[0129]

[0130] Among them, h is the discrete solution step size of the digital twin mechanism model; n is the total number of iterations required to execute the digital twin parameter identification and update algorithm once.

[0131] Step 103: In the digital space, refer to the power converter hardware platform design indicators, reserve resource space in the FPGA, and build a real-time simulation model of the power converter, namely the digital twin. In addition, the FPGA interface is also used to establish a virtual-real data interaction channel to receive the inductor current i collected in the physical space. L and the output voltage v o data sets and transmits the driving control signals of the processor in the digital space to the physical twin.

[0132] Step 104: The calculated values of the inductor current and the output voltage obtained by the real-time simulation model (digital twin) of the power converter are used as matrix input quantities II. The matrix input quantities II are represented by i Lm,0 ,i Lm,1 ,…i Lm,N and v om,0 ,v om,1 ,…v om,N Matrix input quantities I and II are used as inputs to the real-time update algorithm of the processor model to perform real-time status monitoring of the power converter hardware platform (physical twin).

[0133] Specific implementation method three: Combination Figures 1 to 9 This embodiment describes a real-time state monitoring method for a power converter based on heterogeneous embedded digital twins. Figure 2 As shown, the inductor current measurement value i in step 102 Lm,n and the output voltage measurement value v om,n It is necessary to sample analog quantities through an FPGA interface with highly parallel and pipelined structure to compare it with the inductor current calculated by the power converter digital twin mechanism model. L,n And the output voltage calculated value v o,n Maintain time synchronization. Where n = 0,…,N. N represents the sample size of the measurement data. The higher the value, the more accurate the condition monitoring effect.

[0134] Specific implementation method four: Combination Figures 1 to 9 This embodiment describes a method for real-time status monitoring of a power converter based on heterogeneous embedded digital twins. Step 2 of this embodiment implements the real-time simulation of the power converter based on FPGA as follows:

[0135] Step 201: Establish a digital twin mechanism model of the power converter. The modeling object is the BUCK-BOOST power converter circuit. When operating in the BUCK operating mode, the digital twin mechanism model is represented in state space form and can be expressed as:

[0136]

[0137] When working in BOOST mode, the digital twin mechanism model can be expressed as:

[0138]

[0139] Among them, R dson 、R L 、R C are the parasitic resistances of power MOSFET, inductor, and capacitor respectively; i L is the inductor current; V in is the input voltage; R is the equivalent load resistance; L and C are the capacitance and inductance values respectively; D is the duty cycle, generated by the controller in the processor; v o is the output voltage; v c is the capacitor voltage; V f is the forward conduction voltage of the diode. The following analysis and discussion will take the BOOST operating mode as an example.

[0140] Step 202: Use the implicit Euler method to solve the power converter digital twin mechanism model. The implicit Euler method solution formula can be expressed as:

[0141]

[0142] Among them, h is the discrete solution step of the digital twin mechanism model; i L,n+1 and i L,n is the exact solution L (t n+1 ) and i L (t n ) approximate value; t n Represents the nth th Calculated position; t n+1 =t n +h represents the next calculation position.

[0143] Output voltage v o Using discrete time steps, this can be expressed as:

[0144]

[0145] So, the (n+1)th th The output voltage value v o,n+1 You can pass the current nth th The inductor current i L,n and capacitor voltage v c,n Calculate and get it. Discretize the above formula and it can be expressed as:

[0146]

[0147] Therefore, the matrices are defined as follows:

[0148]

[0149] Where, is defined as the observation matrix; θ1 and θ2 are defined as parameter estimation matrices, and the system parameters P1(L, C, R L ,R c , R) have an algebraic relationship. Based on the algebraic relationship shown in the above formula, parameter identification can be performed through linear regression analysis methods, and finally the estimated value of the system parameter P1 can be obtained through iterative calculation.

[0150] Step 203: Use the values of the inductor current, capacitor voltage, output voltage, and voltage error signal of the next discrete time step calculated by the BOOST power converter digital twin mechanism model as the input of the current discrete time step to implement closed-loop control. The implementation of closed-loop control can be expressed as:

[0151]

[0152] Among them, v e,n+1 Shows voltage error signal, V ref is the voltage reference value;

[0153] Specific implementation method five: Combination Figures 1 to 9 To illustrate this real-time method, the embodiment of the present invention is a real-time state monitoring method for a power converter based on heterogeneous embedded digital twins. The state monitoring method is implemented in the CPU and is real-time monitoring, such as Figure 4 As shown in FIG, the steps for implementing the processor-based power converter digital twin mechanism model update algorithm in step three are as follows:

[0154] Step 301: Initialize the model update algorithm.

[0155] Initialize the inverse of the covariance matrix P(0) = α 2 I, where α is an infinite integer, usually 10 5 ~10 8 ; I is the identity matrix;

[0156] Initialize the parameter estimation matrix Usually it can be a very small value or a zero vector; the initial value P(0) and The effect of decreases as the number of recursions increases.

[0157] Step 302: Update sample data x n (t) and the observation matrix Where n = 1, 2; x1(t) = i L (t), x2(t)=u0(t), inductor current iL The output voltage u0 is obtained through FPGA embedded sampling; the observation matrix Calculated according to the formula in step 202, the entire process is implemented in the digital space.

[0158] Step 303: Calculate the objective function. L 、v c The dataset i obtained by sampling with the physical twin Lm 、v cm As two sets of matrix input quantities I and II, and calculate the objective function f obj , as the evaluation basis for the iterative results of the model update algorithm. The objective function f obj The calculation formula is:

[0159]

[0160] Among them, i L,j and v o,j are the inductor current and output voltage calculated from the BOOST power converter digital twin mechanism model, defined as matrix I. Lm,j and v om,j is the real measurement data from the BOOST power converter hardware platform, defined as matrix II, and N is the sample size of the measurement data.

[0161] Step 304: Optimize the parameter estimation matrices θ1 and θ2. After calculating the objective function, perform crossover and mutation operations on the parameter estimation matrices θ1 and θ2.

[0162] The crossover operation uses single-point and two-point crossover operations. Before the crossover operation, according to the objective function f obj The size of the parameter estimation matrix is divided into three tribes. The crossover rules are as follows: the third-class tribe uses a two-point crossover with other tribes (first and second class); the second-class tribe uses a single-point crossover with tribes other than the third-class (second and first class); and the first-class tribe uses a single-point crossover within itself.

[0163] The mutation operation multiplies the matrix elements in the tribe by a reasonable random number to improve the tribe's diversity. The mutation rules are: for third-class tribes, a global mutation method is used; for second-class tribes, a single-point mutation method is used; and for first-class tribes, no mutation operation is performed.

[0164] Step 305: Model parameter adaptation. If the objective function f obj If the value is greater than the threshold, the model parameter adaptive process (i.e., recursive least squares genetic optimization algorithm) is iteratively updated, and the digital twin model parameter P1 is updated. Otherwise, the optimization process ends and the parameter estimation matrix θ is output. n, after decoupling, we get the system parameter P1. The parameter adaptation process is shown in the following formula:

[0165]

[0166] Where k(t) is defined as the gain matrix; λ is a very small integer used to ensure the reversibility of the matrix, and is set to 1. n = 1, 2; x1(t) = i L (t), x2(t)=v0(t). Assume that the process parameter matrix obtained by identification is:

[0167]

[0168] According to step 202, the process parameter matrices θ1, θ2 and the system parameters P1 (L, C, R L ,R c ,R), we can get the decoupling result of the above formula, that is, the system parameters P1(L,C,R L ,R c ,R) is:

[0169]

[0170] Step 306: Feedback of status monitoring results. The final parameter identification result P1(L, C, R L ,R c ,R) is transmitted to the host computer monitoring platform as one of the real-time status monitoring elements of the BOOST power converter hardware platform (physical twin) for the decision-making link to make corresponding decisions.

[0171] Specific implementation method six: combination Figures 1 to 9 This real-time method is described. This embodiment describes a real-time state monitoring method for power converters based on heterogeneous embedded digital twins. This method is a virtual-physical fusion system designed specifically for power converters and integrated on a dedicated computing platform. The system requires a high-precision digital twin model that runs in real time and has dynamic update capabilities. The heterogeneous method based on processors and FPGAs is the key to the implementation of this complex system. The processor's excellent algorithm processing capabilities provide the conditions for implementing the power converter's physical twin control algorithm and the digital twin model update algorithm, while the FPGA's parallel computing capabilities provide the basis for achieving small-step, high-precision real-time simulation of power converters.

[0172] In order to verify the technical effect of the real-time status monitoring method of power converters based on heterogeneous embedded digital twins of the present invention, a verification test was carried out.

[0173] An embedded digital twin heterogeneous platform was built based on the NI cRIO-9038 multicore controller. The platform features a 1.33GHz dual-core CPU, 2GB of DRAM, a Kintex-7160T FPGA, and an 8-slot Compact-RIO add-on card. The BOOST physical power converter circuit was modeled using the implicit Euler method to establish its linear discrete model and build a digital twin of the BOOST converter. A recursive least squares genetic optimization algorithm was used as the interface between the BOOST converter's physical twin and its digital twin. A digital twin system of the BOOST converter was constructed on the NI-9038 multicore controller, verifying sensorless real-time condition monitoring capabilities.

[0174] Figure 7 is the objective function f obj It can be seen that after about 20 iterations, the objective function finally converges.

[0175] Figure 8 The digital twin parameter set P1 (L, C, R L ,R c ,R) after 0.01 seconds of iteration. It can be seen that the parameter set P1 (L, C, R L ,R c After 0.01 seconds of iteration, R) reaches a stable value, indicating that the error between the output waveforms of the BOOST converter's digital twin and physical twin is less than the set threshold. At this point, the parameter values identified by the digital twin can be accurately mapped to the real-time state of the physical twin.

[0176] In order to further verify the identified parameter set P1(L, C, R L ,R c ,R) is correct, the calculated estimated value i of the digital twin L 、v o and the actual sampling value i of the physical twin Lm 、v om For comparison, the results are as follows Figure 9 As shown. It can be seen that whether in the initial dynamic process or the final steady-state process, the inductor current i L and the output voltage v o The errors between the calculated estimated values and the actual sampled values are very small, which further confirms the accuracy of the method proposed in this invention.

[0177] In summary, the proposed method for real-time power converter condition monitoring based on heterogeneous embedded digital twins utilizes a recursive least squares genetic optimization algorithm to self-calibrate the parameters of the BOOST converter discretization model, thereby achieving sensorless real-time condition monitoring without requiring hardware modifications. The real-time condition monitoring results maintain consistency with actual operating results, demonstrating significant effectiveness and facilitating widespread adoption.

[0178] Example 1: Power Management System in a Smart Manufacturing Factory

[0179] In smart manufacturing plants, power management systems are crucial to the stable operation of production equipment. In this embodiment, the present invention's real-time status monitoring method for power converters based on heterogeneous embedded digital twins is used to monitor and manage the power modules of key production equipment within the factory in real time.

[0180] Each key production equipment in the factory is equipped with a power converter to regulate and stabilize the power supply. In step one, an embedded digital twin control system is established to synchronize the state evolution of each power converter's physical twin with its corresponding digital twin in the FPGA. The processor and FPGA work together to ensure real-time data acquisition and synchronization of inductor current and output voltage.

[0181] In step two, the FPGA performs real-time simulation of the power converter, using the implicit Euler method to rapidly calculate the inductor current and output voltage, ensuring simulation step sizes of hundreds of nanoseconds. The processor then executes a recursive least-squares genetic optimization algorithm. Based on the comparison of collected real-world data with simulated data, the parameters of the digital twin mechanism model are updated in real time, improving the accuracy of condition monitoring.

[0182] Through step three, the system can promptly detect abnormal power converter conditions, such as unusual fluctuations in inductor current and output voltage, as well as degradation trends in capacitors, inductors, and switches, providing early warning of potential equipment failures. Based on real-time monitoring data, maintenance personnel can proactively schedule equipment inspections and maintenance, avoiding production downtime caused by unstable power supplies and improving overall factory operational efficiency and equipment lifespan.

[0183] Example 2: Power Management System for Electric Vehicle Charging Station

[0184] At electric vehicle charging stations, a stable and efficient power management system is key to ensuring the reliable operation of charging equipment. This embodiment applies the present invention's real-time status monitoring method for power converters based on heterogeneous embedded digital twins to perform real-time monitoring and optimized management of multiple charging piles within the charging station.

[0185] Each charging station is equipped with a power converter that converts AC power into DC suitable for EV charging. In step one, an embedded digital twin control system is established, synchronizing the state evolution of each charging station's power converter physical twin with its corresponding digital twin real-time simulation model in the FPGA. The processor and FPGA work together to collect inductor current and output voltage data in real time, ensuring data accuracy and real-time performance.

[0186] In step two, the FPGA performs real-time simulation of the power converter, rapidly calculating the inductor current and output voltage using the highly accurate implicit Euler method, ensuring high simulation step accuracy and low latency. The processor uses a recursive least-squares genetic optimization algorithm to compare actual sampled data with simulated data, updating the parameters of the digital twin model in real time to ensure high accuracy and synchronization of the monitoring system.

[0187] Through step three, the system monitors the operating status of charging piles in real time, promptly detecting and responding to power anomalies such as voltage fluctuations or abnormal currents, as well as degradation trends in capacitors, inductors, and switches. Based on this real-time monitoring data, the system dynamically adjusts power distribution strategies to optimize load distribution among charging piles, ensuring efficient and safe operation of each charging pile. Furthermore, the system predicts maintenance needs for charging piles, pre-scheduling inspections to minimize downtime and improve overall service quality and user satisfaction at charging stations.

[0188] The two examples above demonstrate the specific application of this invention in smart manufacturing plants and electric vehicle charging stations, demonstrating its adaptability and efficiency in different scenarios. This real-time status monitoring method based on heterogeneous embedded digital twins significantly improves the monitoring accuracy and response speed of power converters, ensuring stable operation and optimized management of the system.

[0189] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any technician familiar with the profession can make some changes or modifications using the technical content disclosed above without departing from the scope of the technical solution of the present invention, which are regarded as equivalent embodiments of equivalent changes. Any simple modifications, changes and equivalent structural changes made to the above embodiments based on the technical essence of the present invention without departing from the technical essence of the present invention are still within the scope of protection of the technical solution of the present invention.

Claims

1. A real-time status monitoring method for power converters based on heterogeneous embedded digital twins, characterized in that: The following steps are involved: Step 1: Design of heterogeneous embedded digital twin control system for power converters; Step 2: FPGA-based real-time simulation of power converter; Step 3: Implementation of processor-based power converter digital twin mechanism model update algorithm.

2. The method for real-time status monitoring of power converters based on heterogeneous embedded digital twins according to claim 1, characterized in that: The step one comprises: In the physical space, the power converter hardware platform is used as a physical twin, and the processor provides driving control signals; In the digital space, based on the power converter hardware platform design indicators, a real-time simulation model is built in the FPGA, and a virtual-real data interaction channel is established; The inductor current and output voltage calculated by the power converter simulation model and the measured data are input into the processor model for real-time status monitoring.

3. The method for real-time status monitoring of power converters based on heterogeneous embedded digital twins according to claim 2, characterized in that: In step 1, the sampling of the inductor current and the output voltage is implemented through the FPGA interface, and the time synchronization of the sampling value and the calculated value of the digital twin mechanism model is achieved through real-time scheduling of the embedded system.

4. The method for real-time status monitoring of power converters based on heterogeneous embedded digital twins according to claim 1, characterized in that: The second step includes: Establish a digital twin mechanism model of power converter; The implicit Euler method is used to solve the mechanism model; Closed-loop control is achieved through data such as inductor current, capacitor voltage and output voltage.

5. The method for real-time status monitoring of power converters based on heterogeneous embedded digital twins according to claim 4, characterized in that: In the second step, the inductor current, capacitor voltage, voltage error and modulation signal of the next time step are calculated by the control model, and are assigned to the relevant parameters of the current calculation time step.

6. The method for real-time status monitoring of power converters based on heterogeneous embedded digital twins according to claim 1, characterized in that: The step three includes: Update sample data and observation matrix; Calculate the objective function based on the input data of the model update algorithm; Based on the objective function value, perform crossover and mutation operations on the parameter estimation matrix; Based on the model update algorithm, perform adaptive update of model parameters; Output parameter estimation matrix, decouple the matrix to obtain parameter identification results; The parameter identification results are fed back to the host computer platform for real-time status monitoring.

7. A real-time status monitoring system for power converters based on heterogeneous embedded digital twins, characterized in that: include: The physical twin, i.e. the power converter hardware platform, has its drive control signals provided by the processor and reserves sampling interfaces for the power converter inductor current and output voltage; The digital twin, or the power converter's mechanism model, is implemented in parallel by the FPGA for model solving and real-time simulation. It primarily receives model simulation and identification parameters from the processor and simultaneously outputs the calculated inductor current and output voltage. The accuracy of the power converter mechanism model in the FPGA is verified by comparing its offline simulation results with a reference model built using the Simulink SPS model library. The FPGA-based real-time simulation requires verification through experimental results. Heterogeneous embedded systems, that is, a heterogeneous computing platform consisting of processors and FPGAs, are the key to realizing this complex and real-time system. The powerful control and algorithm processing performance of the processor provides the conditions for implementing complex control algorithms and model update algorithms for power converters, while the parallel computing capability and pipeline structure of the FPGA lay the foundation for realizing small-step, high-precision real-time simulation of power converters.

8. The real-time status monitoring system for power converters based on heterogeneous embedded digital twins according to claim 7, characterized in that: The model update algorithm in the processor is further configured to: Compare the objective function with the threshold. If the objective function is greater than the threshold, perform the recursive least squares (GA-RLS) genetic optimization model update algorithm. Otherwise, end the algorithm optimization process and output the parameter estimation matrix. Calculate the gain vector; Update the inverse of the covariance matrix; Update the parameter estimation matrix; The parameter estimation matrix is transferred to the digital twin mechanism model as the input for the next iterative optimization of the model update algorithm.

9. The real-time status monitoring system for power converters based on heterogeneous embedded digital twins according to claim 7, characterized in that: The controller in the processor is further configured to: The real power converter controller performs the generation of power converter switching signals, the processing of sampling signals, and the implementation of PI double closed-loop control, etc., mainly scheduled in the form of a state machine; The digital twin model controller performs feedback and correction of digital twin model parameters, as well as data synchronization with the real power converter controller.

10. The real-time status monitoring system for power converters based on heterogeneous embedded digital twins according to claim 7, characterized in that: The FPGA is further configured as follows: According to the design indicators of the power converter hardware platform, reserve resource space to build a real-time simulation model of the power converter; Establish a virtual-real data interaction channel to receive sampled data from the physical twin and transmit the drive control signals generated by the processor to the physical twin; Realize real-time solution and simulation of the digital twin mechanism model of the power converter, including the calculation of inductor current and output voltage.

Images