Impedance online identification method and device of resonant converter based on model fine-tuning technology

Abstract

The present application provides an impedance online identification method and device for a resonant converter based on model fine-tuning technology, including: deriving a theoretical impedance model according to the structure of an input series output parallel LLC converter and constructing a BPNN network; obtaining theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; obtaining measurement reference data of the LLC converter under multiple working points by a frequency sweep method to form a second data set; training the pre-trained model according to the second data set to obtain an online impedance identification model; inputting input parameters under the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series output parallel LLC converter. It can effectively improve the impedance identification efficiency and accuracy of the submarine LLC converter and facilitate the online analysis of system stability.

Description

Technical Field

[0001] The present application belongs to the technical field of impedance identification, and in particular, relates to an impedance online identification method and device for a resonant converter based on a model fine-tuning technology. Background Art

[0002] The stable operation of the submarine medium and low voltage DC power supply system is the guarantee for the normal operation of the submarine scientific observation network. In the submarine medium and low voltage DC power supply system, the Modular Multilevel Converter (MMC) of the shore base station is connected to the main junction box by a photoelectric composite cable with a large resistance. The voltage drop of the cable is large when transmitting power over long distances, which significantly affects the efficiency of power transmission. Therefore, the main junction box often uses an input-series output-parallel (ISOP) type LLC converter that can adapt to a wide range of voltage inputs and has high power conversion efficiency, and adjusts the working state of each LLC sub-module according to different input voltages to ensure the stable operation of the entire system.

[0003] In actual engineering, submarine LLC converters are mostly black boxes, and the internal control parameters and circuit structure are unknown. Therefore, the modeling method of deriving equivalent impedance based on converter structure and parameters cannot be directly used for stability analysis of submarine systems. The frequency sweeping method can continuously inject disturbance signals into the running system and calculate the impedance value by measuring the voltage and current response of the disturbance signal. However, data collection in the submarine environment is more difficult, and the data samples that can be collected are small. Therefore, the frequency sweeping method is not suitable for impedance identification of submarine LLC converters under multiple working points. In recent years, data-driven impedance modeling methods have been widely used in stability analysis. By inputting a large amount of relevant data into neural networks such as artificial neural networks (ANN), a nonlinear mapping relationship between the converter operating point and the equivalent impedance value is constructed to achieve online identification of equivalent impedance. The volume of such data samples will significantly affect the effect of identification. Summary of the invention

[0004] The main purpose of the embodiments of the present invention is to provide an impedance online identification method and device for a resonant converter based on model fine-tuning technology, which solves the problem of insufficient training data samples, can effectively improve the impedance identification efficiency and accuracy of the seabed LLC converter, and facilitates online analysis of system stability.

[0005] In a first aspect, a method for online impedance identification of a resonant converter based on a model fine-tuning technology is provided, the method comprising: deriving a theoretical impedance model of an input series-output parallel type LLC converter according to the structure of the input series-output parallel type LLC converter, and constructing a BPNN network according to the theoretical impedance model; acquiring theoretical parameter data of the theoretical impedance model at multiple operating points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; acquiring measured reference data of the input series-output parallel type LLC converter at multiple operating points by a sweep frequency method to form a second data set, and performing model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; inputting input parameters at the operating point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series-output parallel type LLC converter.

[0006] In a possible implementation, a theoretical impedance model of an input series-output parallel type LLC converter is derived according to the structure of the input series-output parallel type LLC converter, including: constructing a state space equation of a single LLC converter according to the structure of the input series-output parallel type LLC converter using an extended description function method; adding a small disturbance signal to each variable in the state space equation to obtain an open-loop small signal state space model of the single LLC converter; constructing a small signal control structure of the single LLC converter according to the open-loop small signal state space model of the single LLC converter; and deriving a theoretical equivalent output impedance of the input series-output parallel type LLC converter according to the small signal control structure of the single LLC converter in combination with an input series-output parallel circuit cascade method to obtain a theoretical impedance model of the input series-output parallel type LLC converter.

[0007] In another possible implementation, the small signal control structure includes: an output voltage outer loop, an output current inner loop and a voltage balancing control loop. The voltage balancing control loop is used to output the control current of each sub-module in the input series output parallel LLC converter. The output voltage outer loop is used to output the outer loop regulation current of each sub-module in the input series output parallel LLC converter. The control current and the outer loop regulation current are superimposed to obtain the reference current of each sub-module. The output current of each sub-module is output through the output current inner loop control according to the reference current. The output current of each sub-module is superimposed and transformed by the load impedance to obtain the output voltage of the input series output parallel LLC converter.

[0008] In another possible implementation, the small signal control structure is expressed as: , ,in , , , , They are the sine and cosine of the resonant inductor current with a small disturbance signal added. , are the sine and cosine of the resonant capacitor voltage with a small disturbance signal added, , are the sine and cosine values ​​of the excitation inductor current with a small disturbance signal added, is the output voltage of a single LLC converter with a small disturbance signal added, is the input voltage of a single LLC converter with a small disturbance signal added, is the switching duty cycle, A, B, C are the coefficient matrices of each vector respectively.

[0009] In another possible implementation, the theoretical impedance model is:

[0010] ;

[0011] is the theoretical equivalent output impedance of the input series output parallel LLC converter, represents the output capacitance, s is the complex domain of Laplace transform, N is the number of submodules, represents the gain of the output voltage outer loop, represents the gain of the output current inner loop, Represents the output transfer function of the output current inner loop.

[0012] In another possible implementation, theoretical parameter data of a theoretical impedance model at multiple operating points are obtained to form a first data set, including: obtaining a first output power and a first frequency at multiple operating points; calculating theoretical equivalent impedance amplitudes and phase angles corresponding to the first output power and the first frequency at multiple operating points according to the theoretical impedance model; and combining the first output power, the first frequency, and the theoretical equivalent impedance amplitudes and phase angles at multiple operating points to form a first data set.

[0013] In another possible implementation, measurement reference data of the input series output parallel type LLC converter at multiple operating points is obtained by a frequency sweeping method to form a second data set, including: under different second output power conditions, injecting a disturbance current signal into the parallel connection of the output ports of each sub-module in the input series output parallel type LLC converter; changing the frequency of the disturbance current signal at equal logarithmic intervals within a preset frequency range, and obtaining the measured equivalent impedance amplitude and phase angle of the input series output parallel type LLC converter corresponding to different second frequencies by a frequency sweeping method; combining different second output powers and second frequencies, and the corresponding measured equivalent impedance amplitude and phase angle, to form a second data set of real measurements of a small sample.

[0014] In a second aspect, an impedance online identification device for a resonant converter based on a model fine-tuning technology is provided, comprising: a network construction unit, used to derive a theoretical impedance model of an input series-output parallel type LLC converter according to the structure of the input series-output parallel type LLC converter, and to construct a BPNN network according to the theoretical impedance model; a first model training unit, used to obtain theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and to train the BPNN network according to the first data set to obtain a pre-trained model; a second model training unit, used to obtain measured reference data of the input series-output parallel type LLC converter under multiple working points by a sweep frequency method to form a second data set, and to perform model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; an impedance identification unit, used to input input parameters under the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series-output parallel type LLC converter.

[0015] In a third aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, an impedance online identification method for a resonant converter based on a model fine-tuning technology as provided in the first aspect is implemented.

[0016] In a fourth aspect, a non-transitory computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, an impedance online identification method of a resonant converter based on a model fine-tuning technology as provided in the first aspect is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0018] Figure 1 A schematic flow chart of an impedance online identification method for a resonant converter based on a model fine-tuning technology provided by an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of the structure of an input series output parallel type LLC converter provided by one embodiment of the present invention;

[0020] Figure 3 A small signal control block diagram of an ISOP type LLC converter provided by an embodiment of the present invention;

[0021] Figure 4 A schematic diagram of the structure of a BPNN network provided in an embodiment of the present invention;

[0022] Figure 5A schematic diagram comparing the theoretical value and measured value of equivalent impedance of an ISOP type LLC converter provided in an embodiment of the present invention, wherein (a) is the amplitude and (b) is the phase;

[0023] Figure 6 The figure is a schematic diagram of the comparison between the theoretical value and the measured value of the equivalent impedance at the working point of 50 kW output power using the impedance online identification method according to the embodiment of the present invention, wherein figure (a) is the amplitude and figure (b) is the phase;

[0024] Figure 7 The figure is a schematic diagram of the comparison between the theoretical value and the measured value of the equivalent impedance at the working point of 100 kW output power using the impedance online identification method of the embodiment of the present invention, wherein figure (a) is the amplitude and figure (b) is the phase;

[0025] Figure 8 A structural diagram of an impedance online identification device for a resonant converter based on a model fine-tuning technology provided by an embodiment of the present invention;

[0026] Fig. 9 A schematic diagram of the physical structure of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0027] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar modules or modules with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.

[0028] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, modules and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, modules, components and / or groups thereof. It should be understood that when we refer to a module as being "connected" or "coupled" to another module, it may be directly connected or coupled to the other modules, or there may be intermediate modules. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any modules and all combinations of one or more associated listed items.

[0029] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation method of the present application will be further described in detail below in conjunction with the accompanying drawings.

[0030] The technical solution of the present application and how to solve the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0031] like Figure 1 The flowchart of an impedance online identification method of a resonant converter based on a model fine-tuning technology provided by an embodiment of the present invention is shown. The method is applied to a server, and the method includes:

[0032] Step S11, deriving a theoretical impedance model of the input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter, and constructing a BPNN network based on the theoretical impedance model.

[0033] Among them, the input series output parallel type LLC converter is preferably an ISOP type submarine LLC converter.

[0034] Step S12, obtaining theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model.

[0035] Step S13, obtaining measurement reference data of the input series output parallel LLC converter at multiple operating points by a frequency sweep method to form a second data set, and performing model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model.

[0036] Step S14, inputting the input parameters of the working point to be identified into the online impedance identification model, and obtaining the equivalent impedance amplitude and phase angle of the input series output parallel LLC converter.

[0037] The embodiment of the present invention first establishes a general theoretical impedance model of an ISOP-type submarine LLC converter, then collects theoretical equivalent impedance data to train a BP neural network (BPNN, Back Propagation Neural Network) to obtain a pre-trained model, and then combines fine-tuning transfer learning technology and uses a small amount of real measurement data to adjust the pre-trained model to obtain an online impedance identification model of the ISOP-type submarine LLC converter, thereby applying the online impedance identification model to online identify the impedance at the working point to be identified, solving the problem of insufficient training data samples, and being able to effectively improve the impedance identification efficiency and accuracy of the ISOP-type submarine LLC converter, and facilitating online analysis of system stability.

[0038] In the embodiments of the present invention, see Figure 2, the input series output parallel type LLC converter includes: a plurality of sub-modules connected in series on the input side and in parallel on the output side. Any sub-module represents a single LLC converter, and the structure and circuit parameters of each sub-module are the same. Any sub-module includes: an input capacitor, a switch full bridge, an LLC resonant cavity, a transformer, an uncontrolled rectifier bridge, and an output capacitor. The LLC resonant cavity includes a resonant inductor, an excitation inductor, and a resonant capacitor. The input capacitor is connected in parallel to the input end of the switch full bridge, one end of the resonant inductor is connected to the midpoint of the first bridge arm of the switch full bridge, and the other end is connected to one end of the excitation inductor, one end of the resonant capacitor is connected to the midpoint of the second bridge arm of the switch full bridge, and the other end is connected to the other end of the excitation inductor, the input end of the transformer is connected in parallel to the two ends of the excitation inductor, the output end of the transformer is connected between the midpoint of the first bridge arm and the midpoint of the second bridge arm of the uncontrolled rectifier bridge, and the output capacitor is connected in parallel to the output end of the uncontrolled rectifier bridge. As Figure 2 Included N submodules, v in_i For the i The input voltage of each submodule, i outi For the i The output current of each submodule is i=1,2,…,N . L r is the resonant inductor, L m is the magnetizing inductance, C r is the resonant capacitor, C in is the input capacitance, C o is the output capacitance, v in is the total input voltage, v out is the output voltage, i o is the output current, Z L The output voltage of each submodule is v out S1, S2, S3, and S4 are the four switch tubes that make up the switch full bridge, and D5, D6, D7, and D8 are the four diodes that make up the uncontrolled rectifier bridge. The voltage ratio between the input and output ends of the transformer is n:1.

[0039] In step S11, optionally, the state space equation of a single LLC converter is constructed by using an extended description function method according to the structure of the input series output parallel type LLC converter. Different from the previous modeling research on LLC converters, the ISOP type LLC converter adopts a phase shift control method to adapt to the application scenario with a wide input voltage range. The state space equation of a single LLC converter in the input series output parallel type LLC converter can be constructed by using the extended description function method, as shown in formula (1). In the LLC with phase shift control, w s is a constant, w s =2πf s , f s is the switching frequency, and the duty cycle d is a variable.

[0040] (1)

[0041] in, L r is the resonant inductor, L m is the magnetizing inductance, C r is the resonant capacitor, C in is the input capacitance, C o is the output capacitance, v in is the input voltage, v out is the output voltage, i in is the input current, i o is the output current, i rs 、i rc are the sine and cosine of the resonant inductor current respectively, i ms 、i mc are the sine and cosine of the magnetizing inductor current, v rs 、v rc are the sine and cosine of the resonant capacitor voltage respectively, d is the switching duty cycle. I p is the composite amplitude, which is calculated as (2):

[0042] (2)

[0043] Furthermore, by adding small disturbance signals to each variable in the state space equation, the open-loop small signal state space model of a single LLC converter is obtained, as shown in equation (3):

[0044] (3)

[0045] in, , They are the sine and cosine of the resonant inductor current with a small disturbance signal added. , are the sine and cosine of the resonant capacitor voltage with a small disturbance signal added, , are the sine and cosine values ​​of the excitation inductor current with a small disturbance signal added, is the output voltage of a single LLC converter with a small disturbance signal added, For any submodule The input voltage of the small disturbance signal is added, Any submodule that adds a small disturbance signal The input voltage, For any submodule The steady-state value of the input voltage, V out is the steady-state value of the output voltage, I o is the steady-state value of the output current, I in is the steady-state value of the input current, I rs 、I rc are the steady-state values ​​of the sine and cosine of the resonant inductor current, I ms 、I mc are the steady-state values ​​of the sine and cosine of the magnetizing inductor current, V rs 、V rc are the steady-state values ​​of the sine and cosine of the resonant capacitor voltage, respectively. D y is the steady-state value of the switch duty cycle.

[0046] Then, the small signal control structure of a single LLC converter is constructed based on the open-loop small signal state space model of a single LLC converter. Considering the stability and dynamic response performance of the control structure, a three-loop control structure of output voltage outer loop-output current inner loop-voltage control loop is adopted to ensure the voltage balancing on the input side and current balancing on the output side of the LLC converter, and to achieve the stable voltage output of the ISOP type LLC converter. Combined with the internal structure of the submodule and the circuit connection relationship between the submodules, the following can be constructed: Figure 3 The small signal control block diagram of the ISOP type LLC converter is shown. The small signal control structure includes: an output voltage outer loop, an output current inner loop and a voltage balancing control loop. The voltage balancing control loop is used to output the control current of each submodule in the input series output parallel type LLC converter. The output voltage outer loop is used to output the outer loop regulation current of each submodule in the input series output parallel type LLC converter. The control current and the outer loop regulation current are superimposed to obtain the reference current of each submodule. The output current of each submodule is output through the output current inner loop control according to the reference current. The output current of each submodule is superimposed and converted by the load impedance to obtain the output voltage of the input series output parallel type LLC converter. It should be noted that the control structure, control parameters and circuit parameters of each submodule are the same.

[0047] Σ i out Represents the sum of the output currents of the rectifier bridges of each submodule. Combined with the Laplace transform, Figure 3 available v out With Σ i out Relationship:

[0048] (4)

[0049] in, s is a complex variable in the frequency domain, used to represent frequency, N is the number of submodules, .

[0050] Since the ISOP cascade model can achieve input voltage equalization and output current equalization, therefore:

[0051] (5)

[0052] in, N is the number of submodules.

[0053] Further v out The output current of each submodule i out The relationship is shown in formula (6):

[0054] (6)

[0055] In this way, based on the small signal control structure of a single LLC converter and the circuit cascade mode of input series and output parallel, the theoretical equivalent output impedance of the input series and output parallel LLC converter is derived, and the theoretical impedance model of the input series and output parallel LLC converter is obtained. Figure 3 The theoretical equivalent output impedance of the ISOP type LLC converter can be derived as:

[0056] (7)

[0057] in, is the theoretical equivalent output impedance of the input series output parallel LLC converter, Represents the output capacitance, s is the complex field of the Laplace transform, represents the gain of the output voltage outer loop, represents the gain of the output current inner loop, The output transfer function of the output current inner loop is represented by arrive The theoretical impedance model can provide technical support for the online impedance identification of ISOP submarine LLC converter.

[0058] In the embodiment of the present invention, the small signal control structure can be expressed as: , ,in , for The differential of , for The differential of , , They are the sine and cosine of the resonant inductor current with a small disturbance signal added. , are the sine and cosine of the resonant capacitor voltage with a small disturbance signal added, , are the sine and cosine values ​​of the excitation inductor current with a small disturbance signal added, is the output voltage of a single LLC converter with a small disturbance signal added, is the input voltage of a single LLC converter with a small disturbance signal added, is the switching duty cycle, A, B, and C are the coefficient matrices of each vector. Further deduction: G if The expression is shown in formula (8):

[0059] (8)

[0060] After deriving the theoretical impedance model of the input series and output parallel LLC converter, a BPNN network is constructed based on the theoretical impedance model.

[0061] In step S12, the theoretical equivalent impedance data of the ISOP type LLC at multiple working points can be obtained according to formula (7), forming a first output power P L , first frequency f , and the first frequency f The corresponding equivalent impedance phase angle , equivalent impedance amplitude Specifically, the first output power and the first frequency at multiple operating points are obtained; the theoretical equivalent impedance amplitude and phase angle corresponding to the first output power and the first frequency at multiple operating points are calculated according to the theoretical impedance model; the first output power, the first frequency, and the theoretical equivalent impedance amplitude and phase angle at multiple operating points are combined to form the first data set.

[0062] The structure of BPNN network can be found in Figure 4 , including input layer, output layer and hidden layer. Among them, X i Indicates the input quantity, i Indicates the serial number of the input layer node, H j,k represents the hidden layer nodes, j represents the number of hidden layers, k represents the hidden layer node number, Y o Indicates the output quantity, subscript o represents the node number of the output layer, w ih and b ih represents the input weight and input bias, w ho and b ho Indicates the output weight and output deviation. The input quantity in this application is the first output power P L and the first frequency f .

[0063] In the BPNN network, the linear transformation function is first used to perform linear transformation on the input data, then the ReLU activation function is used for nonlinear processing, and finally the linear transformation function is used to directly output the result. Figure 4 For example, the third hidden layer m The output of neuron No. is shown in formula (9):

[0064] (9)

[0065] in, h 3,m Indicates the third layer m The output of neuron number, R represents the ReLU activation function, r is the neuron number of the second hidden layer, h 2,r Represents the second hidden layer r The output of neuron number, Represents the third hidden layer m The input weight of neuron number, b 3,m Represents the third hidden layer m The input bias of neuron number.

[0066] The BPNN network is trained according to the first data set to obtain a pre-trained model Model1.

[0067] In step S13, the impedance measurement value of the black box ISOP type LLC converter at some operating points can be obtained to form A second data set composed of. Specifically, under different second output power conditions, a disturbance current signal is injected into the parallel connection of the output ports of each submodule in the input series output parallel type LLC converter; the frequency of the disturbance current signal is changed at equal logarithmic intervals within a preset frequency range, and the measurement amplitude and phase angle of the equivalent impedance of the input series output parallel type LLC converter corresponding to different second frequencies are obtained by a frequency sweep method, wherein the preset frequency range can be set as needed, preferably between 1 and 10000 Hz; different second output powers and second frequencies, as well as the corresponding measurement amplitudes and phase angles, are combined to form a second data set of real measurements of a small sample. For example, the output power of the black box ISOP type LLC converter is changed, thereby changing the operating point of the converter. Under different second output powers of the converter, a disturbance current signal is injected into the parallel connection of the output ports of each submodule, and the second frequency of the injected disturbance current signal is changed at equal logarithmic intervals between 1 and 10000 Hz. f m , use the frequency sweep method to obtain different second frequencies f m The corresponding measured equivalent impedance phase angle of the converter Z om_pha ( f m ) and impedance magnitude Z om_amp ( f m ) ，A second data set that constitutes a real measurement of a small sample. It should be noted that the number of values ​​of the second output power is less than the number of values ​​of the first output power, and the number of values ​​of the second frequency is much smaller than the number of values ​​of the first frequency. The specific value of the second output power can be a part of the value of the first output power. Of course, the specific value of the second output power may also be different from the value of the first output power. Similarly, the specific value of the second frequency can be a part of the value of the first frequency. Of course, the specific value of the second frequency may also be different from the value of the first frequency, and no specific limitation is made here.

[0068] The embodiment of the present invention can use a fine-tuned transfer learning method to obtain an online impedance identification model. Specifically, the pre-trained model is fine-tuned according to the second data set to obtain an online impedance identification model. Specifically, part of the weights of the pre-trained model Model1 are frozen, the second data set is input into the pre-trained model Model1 for training, and the weights and thresholds of Model1 are adjusted to obtain the online impedance identification model Model2.

[0069] By integrating the general theoretical impedance model, BPNN network and fine-tuning transfer learning technology, the limitations of traditional methods such as sweep frequency method and equivalent impedance modeling, which cannot be accurately modeled due to the LLC converter black box, unknown internal parameters and difficulty in obtaining measurement data, are successfully solved. This provides a new and effective way for converter impedance analysis and successfully solves the problem of low efficiency of traditional data-driven impedance identification methods when faced with a shortage of training data samples. This provides a more optimized solution for online impedance identification, thereby greatly improving the efficiency of the entire impedance identification process while ensuring accuracy.

[0070] In step S14, the output power at any working point to be identified is input into the online impedance identification model Model2. P L_p and frequency f p , the equivalent impedance amplitude of ISOP type LLC converter with unknown internal parameters can be obtained Z or_pha ( f p ) and phase angle Z or_amp ( f p ) to realize the online identification of equivalent impedance of ISOP type LLC converter.

[0071] In order to verify the correctness and feasibility of the above method, we first built the following Figure 2 The simulation of ISOP type LLC converter is shown in Figure 1, and the measured value is compared with the theoretical value. The results are as follows Figure 5As shown in the figure, (a) is the amplitude and (b) is the phase, which verifies the rationality of the theoretical impedance model of the proposed ISOP type LLC converter.

[0072] In the output power range [0.1pu~1.0pu], It refers to the ratio of the actual value of the voltage in the power system to its reference value, also known as per unit. The output power of the converter is changed in intervals of 0.1 pu. P L , changing in equal logarithmic intervals between 1 and 10000 Hz f According to formula (7), the theoretical equivalent impedance data of ISOP type LLC at multiple working points are obtained, forming a set of 4000 P L , f , , The first dataset is constructed. Training with the first dataset Figure 4 The BPNN network shown in the figure obtains the pre-trained model Model1.

[0073] Next, for the LLC converter whose internal parameters are unknown, the converter output power is changed at intervals of 0.3 pu. P L , changing in equal logarithmic intervals between 1 and 10000 Hz f m , 400 sets of data were collected by frequency sweeping method , forming the second data set. Freeze some weights of the pre-trained model Model1, input the second data set to train Model1, adjust the weights and thresholds of Model1, and obtain the online impedance identification model Model2. Input the online impedance identification model Model2 P L_p , f p , the equivalent impedance amplitude of ISOP type LLC converter with unknown internal parameters can be obtained Z or_pha ( f p ) and phase angle Z or_amp ( f p ), the comparison results of the impedance prediction value and the measured value of the online impedance identification model Model2 at different working points are as follows Figure 6 and Figure 7 As shown, Figure 6 The figure is a schematic diagram of the comparison of the impedance prediction and measured values ​​of the ISOP type LLC converter with an output power of 50kW at different frequency points, where (a) is the amplitude and (b) is the phase. Figure 7 The figure is a schematic diagram of the comparison of the predicted and measured impedance values ​​of an ISOP type LLC converter with an output power of 100kW at different frequency points, where (a) is the amplitude and (b) is the phase.

[0074] Model3 is the impedance identification model obtained by training the BPNN network using only the second data set. The determination coefficient of the output results of Model1, Model2, and Model3 ( R 2 ) and the root mean square error ( RMSD ) are shown in Table 1 below:

[0075] Table 1 Schematic diagram of the comparison of the output results of the three models

[0076]

[0077] When only a small number of measured samples are available, Model 2 significantly improves the accuracy of impedance identification, and this result verifies the effectiveness of the method proposed in this application.

[0078] The embodiment of the present invention first establishes a general theoretical impedance model of an ISOP-type submarine LLC converter, combines the general theoretical impedance model, BPNN network and fine-tuning transfer learning technology, and obtains an online impedance identification model of an ISOP-type submarine LLC converter. The online impedance identification model is used to identify the equivalent impedance amplitude and phase angle under the working point to be identified. The problem of insufficient training data samples is solved, the efficiency of obtaining the equivalent impedance of the submarine LLC converter is improved, and high-accuracy online impedance identification is achieved, which is convenient for online analysis of system stability.

[0079] In summary, the embodiment of the present invention derives the theoretical impedance model of the input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter, and constructs a BPNN network according to the theoretical impedance model; obtains theoretical parameter data of the theoretical impedance model at multiple working points to form a first data set, and trains the BPNN network according to the first data set to obtain a pre-trained model; obtains measurement reference data of the input series output parallel type LLC converter at multiple working points by a sweeping frequency method to form a second data set, and fine-tunes the pre-trained model according to the second data set to obtain an online impedance identification model, which can effectively improve the impedance identification efficiency and accuracy of the submarine LLC converter and facilitate online analysis of system stability.

[0080] like Figure 8 The structure diagram of an impedance online identification device of a resonant converter based on a model fine-tuning technology provided by an embodiment of the present invention is shown, and is applied to a service end. The device includes:

[0081] A network construction unit 801 is used to derive a theoretical impedance model of the input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter, and to construct a BPNN network according to the theoretical impedance model;

[0082] The first model training unit 802 is used to obtain theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and train the BPNN network according to the first data set to obtain a pre-trained model;

[0083] The second model training unit 803 is used to obtain the measured reference data of the input series output parallel type LLC converter at multiple working points by a frequency sweep method to form a second data set, and perform model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model;

[0084] The impedance identification unit 804 is used to input the input parameters of the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series output parallel LLC converter.

[0085] The device of the above embodiment is applied to the corresponding method in the above embodiment, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0086] Fig. 9 An example of a physical structure diagram of an electronic device is shown in FIG. Fig. 9 As shown, the electronic device may include: a processor (processor) 901, a communication interface (Communications Interface) 902, a memory (memory) 903 and a communication bus 904, wherein the processor, the communication interface and the memory communicate with each other through the communication bus. The processor may call the logic instructions in the memory to execute an impedance online identification method of a resonant converter based on a model fine-tuning technology. The method comprises: deriving a theoretical impedance model of an input-series-output parallel type LLC converter according to the structure of the input-series-output parallel type LLC converter, and constructing a BPNN network according to the theoretical impedance model; acquiring theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; acquiring measured reference data of the input-series-output parallel type LLC converter under multiple working points by a frequency sweeping method to form a second data set, and performing model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; inputting input parameters under the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input-series-output parallel type LLC converter.

[0087] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0088] On the other hand, an embodiment of the present invention further provides a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, when the program instructions are executed by a computer, the computer can execute a method for online impedance identification of a resonant converter based on a model fine-tuning technology provided by the above-mentioned method embodiments. The method includes: deriving a theoretical impedance model of an input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter, and constructing a BPNN network according to the theoretical impedance model; obtaining theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; obtaining measured reference data of the input series output parallel type LLC converter under multiple working points by a sweep frequency method to form a second data set, and performing model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; inputting input parameters under the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series output parallel type LLC converter.

[0089] On the other hand, an embodiment of the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it is implemented to execute an impedance online identification method of a resonant converter based on a model fine-tuning technology provided in the above embodiments. The method includes: deriving a theoretical impedance model of an input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter, and constructing a BPNN network according to the theoretical impedance model; obtaining theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; obtaining measured reference data of the input series output parallel type LLC converter at multiple working points by a sweep frequency method to form a second data set, and performing model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; inputting the input parameters under the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series output parallel type LLC converter.

[0090] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0091] The above is only a partial implementation of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An impedance online identification method for a resonant converter based on model fine-tuning technology, characterized in that: The method comprises: Derivation of a theoretical impedance model of the input series and output parallel type LLC converter according to the structure of the input series and output parallel type LLC converter, and construction of a BPNN network according to the theoretical impedance model; Acquiring theoretical parameter data of the theoretical impedance model at multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; Acquire measurement reference data of the input series output parallel type LLC converter at multiple operating points by a frequency sweep method to form a second data set, and perform model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; Inputting input parameters at the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series output parallel LLC converter; The method of deriving the theoretical impedance model of the input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter includes: According to the structure of the input series output parallel type LLC converter, the state space equation of a single LLC converter is constructed by using the extended description function method. Adding a small disturbance signal to each variable in the state space equation to obtain an open-loop small signal state space model of the single LLC converter; Constructing a small signal control structure of the single LLC converter according to an open-loop small signal state space model of the single LLC converter; According to the small signal control structure of a single LLC converter, combined with the circuit cascade method of input series and output parallel, the theoretical equivalent output impedance of the input series and output parallel LLC converter is derived, and the theoretical impedance model of the input series and output parallel LLC converter is obtained.

2. The method according to claim 1, characterized in that The small signal control structure includes: an output voltage outer loop, an output current inner loop and a voltage balancing control loop, wherein the voltage balancing control loop is used to output the control current of each submodule in the input series output parallel type LLC converter, and the output voltage outer loop is used to output the outer loop regulation current of each submodule in the input series output parallel type LLC converter. The control current and the outer loop regulation current are superimposed to obtain the reference current of each submodule, and the output current of each submodule is output through the output current inner loop control according to the reference current. The output current of each submodule is superimposed and converted by load impedance to obtain the output voltage of the input series output parallel type LLC converter.

3. The method according to claim 1, characterized in that The small signal control structure is expressed as: ； ； in, , for The differential of , for The differential of , , They are the sine and cosine of the resonant inductor current with a small disturbance signal added. , are the sine and cosine of the resonant capacitor voltage with a small disturbance signal added, , are the sine and cosine values ​​of the excitation inductor current with a small disturbance signal added, is the output voltage of a single LLC converter with a small disturbance signal added, For a single LLC converter The input voltage of the small disturbance signal is added, is the switching duty cycle, A, B, C are the coefficient matrices of each vector respectively.

4. The method according to claim 1, characterized in that The theoretical impedance model is: ； is the theoretical equivalent output impedance of the input series output parallel LLC converter, represents the output capacitance, s is the complex domain of Laplace transform, N is the number of submodules, represents the gain of the output voltage outer loop, represents the gain of the output current inner loop, represents the output transfer function of the output current inner loop, The output voltage of the ISOP type LLC converter after adding a small disturbance signal. The total output current of the ISOP type LLC converter after adding a small disturbance signal.

5. The method according to claim 1, characterized in that The step of acquiring theoretical parameter data of the theoretical impedance model at multiple working points to form a first data set includes: Obtaining a first output power and a first frequency at multiple operating points; Calculating theoretical equivalent impedance amplitudes and phase angles corresponding to the first output power and the first frequency at multiple operating points according to the theoretical impedance model; The first output power, the first frequency, and the theoretical equivalent impedance amplitude and phase angle at multiple operating points are combined to form a first data set.

6. The method according to claim 1, characterized in that The step of obtaining the measured reference data of the input series output parallel type LLC converter at multiple operating points by a frequency sweep method to form a second data set includes: Injecting a disturbance current signal into the parallel connection of the output ports of each submodule in the input series output parallel LLC converter under different second output power conditions; Changing the frequency of the disturbance current signal at equal logarithmic intervals within a preset frequency range, and obtaining the measured equivalent impedance amplitude and phase angle of the input series output parallel LLC converter corresponding to different second frequencies by a frequency sweep method; The different second output powers and second frequencies, and the corresponding measured equivalent impedance magnitudes and phase angles, are combined to form a second data set of real measurements of a small sample.

7. An impedance online identification device for a resonant converter based on model fine-tuning technology, characterized in that: Executing the method according to claim 1, the device comprises: A network construction unit, used to derive a theoretical impedance model of the input series output parallel type LLC converter according to the structure of the input series output parallel type LLC converter, and construct a BPNN network according to the theoretical impedance model; A first model training unit, used for acquiring theoretical parameter data of the theoretical impedance model under multiple working points to form a first data set, and training the BPNN network according to the first data set to obtain a pre-trained model; A second model training unit is used to obtain measurement reference data of the input series output parallel type LLC converter at multiple operating points by a frequency sweep method to form a second data set, and perform model fine-tuning training on the pre-trained model according to the second data set to obtain an online impedance identification model; The impedance identification unit is used to input the input parameters of the working point to be identified into the online impedance identification model to obtain the equivalent impedance amplitude and phase angle of the input series output parallel LLC converter.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, an impedance online identification method for a resonant converter based on a model fine-tuning technology is implemented as described in any one of claims 1-6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, an impedance online identification method for a resonant converter based on a model fine-tuning technology is implemented as described in any one of claims 1 to 6.

Images