A model-free predictive control method for CETDC

Abstract

The present invention is applicable to the field of power electronic converter control technology and provides a model-free predictive control method for CETDC (Converter-Electronic Transformer) systems. The method comprises the following steps: analyzing the CETDC topology, deriving its mathematical model, and establishing its hyperlocal model based on the CETDC mathematical model; estimating the system state and disturbances by constructing a Neural Observational Order (NESO) algorithm, performing feedforward compensation for the disturbances, designing and selecting controller parameters, and calculating the final reference signal from the observed values; performing stability analysis of the observer based on the Lyapunov function method, and performing delay compensation on the controller. This method utilizes only the system input and output, disregarding any other system parameters. This method reduces tuning workload and computation time, while achieving better performance in reducing tracking error, dynamic overshoot, and transient time experienced by the system, thereby enhancing system robustness.

Description

Technical Field

[0001] The present invention belongs to the technical field of power electronic converter control, and in particular relates to a model-free predictive control method for CETDC. Background Art

[0002] Compared to traditional, stable, controllable power sources like thermal and nuclear power, renewable energy sources, such as solar and wind power, exhibit significant random fluctuations and intermittent operation due to meteorological constraints. As the proportion of installed renewable energy capacity continues to rise, power grids face the severe challenge of insufficient renewable energy absorption capacity. There is an urgent need to develop new transmission technologies to optimize wide-area energy allocation. In the field of long-distance, high-capacity power transmission, traditional AC transmission technology suffers from high line losses due to the skin effect and proximity effect. High-voltage direct current (HVDC) transmission, with its superior economics and lower operating losses, has become a key technology for addressing this issue.

[0003] As a representative of third-generation DC transmission technology, Flexible DC transmission, since its theoretical proposal in 1990, has developed significant advantages over the past three decades. It not only offers flexible active / reactive power control and low harmonic pollution, but also boasts unique capabilities such as power supply to passive networks, traversing AC system faults, and providing dynamic reactive power support. These advantages have driven its widespread application in various fields, including wind power cluster transmission, regional grid interconnection, and island power supply. Currently, Flexible DC transmission voltage levels have surpassed the ±800kV UHV threshold, with transmission capacity comparable to traditional thyristor-based HVDC transmission. This technology is gradually replacing conventional DC transmission systems in the development of the European Energy Interconnection. As a key technology supporting the future integration of a high proportion of renewable energy, Flexible DC transmission is profoundly reshaping the power system landscape, and its technological development will continue to lead innovation in the DC transmission sector.

[0004] However, the traditional capacitive energy transfer DC / DC converter (CETDC) control method uses PI dual closed-loop control. While fixed-parameter PI control has advantages such as simple structure and strong adaptability, it also has disadvantages such as sensitivity to noise, difficulty handling nonlinear systems, slow system response, and poor stability. Therefore, a model-free predictive control strategy for CETDC based on ESO is proposed. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide a model-free predictive control method for CETDC, aiming to solve the problems raised in the above background technology.

[0006] The embodiment of the present invention is implemented as follows: a model-free predictive control method for CETDC includes the following steps:

[0007] Step 1: Analyze the topological structure of CETDC, derive the mathematical model of CETDC, establish its hyperlocal model based on the mathematical model of CETDC, and analyze the traditional model-free predictive control method of CETDC;

[0008] Step 2: Build a nonlinear extended state observer (NESO) to estimate the system state and disturbance, perform feedforward compensation for the disturbance, design and select controller parameters, and calculate the final reference signal from the observed values;

[0009] Step 3: Perform stability analysis on the observer according to the Lyapunov function method and perform delay compensation on the controller.

[0010] Further technical solution, the step 1 has the following specific steps:

[0011] Step 1.1: Derivation of CETDC mathematical model and construction of super-local model;

[0012] The topology of CETDC includes three-phase identical structures, j = a, b, c Each phase energy storage bridge arm includes a converter valve, N half-bridge sub-modules, a buffer inductor L and its parasitic resistance R in series; wherein, the converter valve is composed of several thyristors and anti-parallel diodes in series;

[0013] U H Represents the DC voltage on the high voltage side, U L Represents the DC voltage on the low voltage side; I H represents the DC current on the high voltage side, I L Represents the DC current on the low voltage side; Represents the high voltage side current of each phase, Represents the low-voltage side current of each phase; Represents the current of each phase energy storage bridge arm, Represents the voltage of each phase energy storage bridge arm;

[0014] In the single-phase structure of CETDC, power is transmitted from the high-voltage side to the low-voltage side. a In the photo, N The number of submodules used in the energy storage bridge arm; U C is the capacitor voltage of a single submodule; S 1. S 2 is the commutation switch;

[0015] When the energy storage bridge arm is connected in parallel to the high voltage side, the loop equation of the topology is shown in formula (1):

[0016] (1);

[0017] Similarly, when the energy storage bridge arm is connected in parallel to the low-voltage side, the circuit equation of the topology can be obtained as shown in formula (2):

[0018] (2);

[0019] Combining Equations (1) and (2), the average state equation of CETDC can be obtained as shown in Equation (3):

[0020] (3);

[0021] in, is the switch function, S 1 When it is turned on, S=1. S 2 When opened, S=0; is a symbolic function;

[0022] The first-order hyperlocal model of a single-input single-output system is shown in Equation (4):

[0023] (4);

[0024] in, u is the control variable; y is the output variable; Non-physical scaling factors entered into the system; F Represents the known part and the unknown disturbance part of the system;

[0025] According to formula (4), the super-local model of CETDC is shown in formula (5):

[0026] (5);

[0027] in, For the system a The known and unknown parts of the phase, G is the differential of the disturbance.

[0028] Step 1.2: Analysis of traditional CETDC model-free predictive control method;

[0029] Based on the ultra-local model of CETDC shown in formula (5), a proportional-integral model-free controller is obtained, and the control rate is shown in formula (6):

[0030] (6);

[0031] in, It is the reference value of the system output; is the estimated value of F; is the proportionality coefficient; is the integration coefficient; , is the tracking error of the system;

[0032] In formula (6) The proportional model-free controller can be obtained, and the control rate is shown in formula (7):

[0033] (7);

[0034] The model-free control principle based on the hyperlocal model is applied to the current control of CETDC. The corresponding controller is called conventional model-free predictive current control. The bridge arm voltage reference signal in the three-phase stationary reference frame is calculated as shown in Equations (8)-(10):

[0035] (8);

[0036] (9);

[0037] (10);

[0038] Among them, the subscript " a 、 b 、 c "express abc Three-phase components; superscript " ” indicates reference value.

[0039] Further technical solution, said step 2 includes the following specific steps:

[0040] Step 2.1: Construct the lumped perturbation in the NESO observation hyperlocal model;

[0041] Based on the super-local model of CETDC in formula (5), the i Pa and F a The NESO with is the state variable and the bridge arm current error as feedback is expressed as shown in Equation (11):

[0042] (11);

[0043] in, is the observed value of the bridge arm current; is the observed value of the lumped disturbance; e a is the current observation error; L 1 and L 2 is the error feedback gain of the observer;

[0044] The designed nonlinear function expression is shown in formula (12):

[0045] (12);

[0046] in, and is an adjustable parameter of the nonlinear function; The value range of is 0 ~ 1. The smaller the value, the stronger the nonlinearity of the function. is the filtering parameter of NESO;

[0047] Step 2.2: Design and select controller parameters;

[0048] In NESO, the parameter L 1 and L 2 will affect system performance, so L 1 and L 2. Make design choices; rewrite Equation (11) into a matrix form, as shown in Equation (13):

[0049] (13);

[0050] in, ; ; ; ; ; for y Observed values ​​of

[0051] By Laplace transforming Equation (13), the characteristic equation of NESO is obtained as shown in Equation (14):

[0052] (13);

[0053] in, I represents the unit matrix;

[0054] In order to ensure stability and dynamic characteristics, all roots of the characteristic polynomial of NESO fall within Get L 1 and L 2 As shown in formula (15):

[0055] (15);

[0056] in, For the bandwidth of NESO, in order to The value of is more intuitive. z domain; in actual digital control systems, the sampling frequency is high enough; the first-order forward Euler discretization is used to discretize Equation (11), as shown in Equation (16):

[0057] (16);

[0058] in, T s is the sampling time; L k1 = T s L 1, L k2 = T s L 2 is the gain of the observer, and its value affects the distribution of the closed-loop poles of the system, thereby affecting the stability of the observer;

[0059] From formula (16), we can see that the transfer function of NESO is shown in formula (17):

[0060] (17);

[0061] The characteristic equation of the observer is shown in formula (18):

[0062] (18);

[0063] From formula (18), we can get The extreme point of is shown in formula (19):

[0064] (19);

[0065] but Can be based on z The poles of the domain are calculated as shown in formula (20):

[0066] (20);

[0067] According to the system stability condition, when the poles of the observer are all distributed in z When the domain is within the unit circle, the system is stable; The choice of z1 and z2 makes z1 and z2 inside the unit circle in the z domain, and the observer bandwidth is selected based on the balance between dynamics and robustness.

[0068] Step 2.3: Final reference signal calculation;

[0069] Discretize Equation (5) into the first order, as shown in Equation (21):

[0070] (twenty one);

[0071] The required modulation reference signal can be further obtained as:

[0072] (twenty two);

[0073] in, is the NESO observation value, because F a It cannot be directly derived;

[0074] Assumptions Finally reached the reference value , then the final modulation reference signal obtained by formula (22) is:

[0075] (twenty three);

[0076] in, The reference value of the final output modulation signal is synthesized through the carrier phase shift modulation technology to generate the final gate pulse signal.

[0077] Further technical solution, said step 3 includes the following specific steps:

[0078] definition , ; Here the nonlinear function Simplified equivalent to e a After analysis, combining equations (5) and (11), we get the error equation as shown in equation (24):

[0079] (twenty four);

[0080] The Lyapunov function is defined as follows:

[0081] (25);

[0082] The derivative of formula (25) is:

[0083] (26);

[0084] From formula (15), we can see that ; From formula (26), we can see that when hour, , satisfying the stability condition;

[0085] If the disturbance change rate G If is 0, then:

[0086] (27);

[0087] It can be seen that the proposed NESO observation system is stable;

[0088] Considering that digital controllers usually have a one-step delay, the delay should be compensated. The final reference signal after compensation is:

[0089] (28);

[0090] in, It can be obtained by Lagrange back extrapolation:

[0091] (29).

[0092] An embodiment of the present invention provides a model-free predictive control method for CETDC. This method uses only the system inputs and outputs, without considering any other system parameters. This reduces tuning workload and computation time, while also achieving better performance in terms of reduced tracking error, dynamic overshoot, and transient time experienced by the system, thereby enhancing system robustness. Simulation results demonstrate that this method outperforms traditional PI control in both dynamic and static conditions in terms of response speed and control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 is the topology of CETDC;

[0094] Figure 2 It is a single-phase structure of CETDC;

[0095] Figure 3 This is the overall control block diagram of this method;

[0096] Figure 4 is the power waveform of PI control;

[0097] Figure 5 The power waveform controlled by this method;

[0098] Figure 6 is the bridge arm current waveform controlled by PI;

[0099] Figure 7 The bridge arm current waveform controlled by this method;

[0100] Figure 8 is the bridge arm voltage waveform controlled by PI;

[0101] Figure 9 The bridge arm voltage waveform controlled by this method;

[0102] Figure 10 is the capacitor voltage waveform controlled by PI;

[0103] Figure 11 This is the capacitor voltage waveform controlled by this method. DETAILED DESCRIPTION

[0104] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0105] The specific implementation of the present invention is described in detail below with reference to specific embodiments.

[0106] An embodiment of the present invention provides a model-free predictive control method for CETDC, comprising the following steps:

[0107] Step 1: Analyze the topological structure of CETDC, derive the mathematical model of CETDC, establish its hyperlocal model based on the mathematical model of CETDC, and analyze the traditional model-free predictive control method of CETDC;

[0108] Step 2: Estimate the system state and disturbance by constructing NESO, perform feedforward compensation on the disturbance, design and select controller parameters, and calculate the final reference signal from the observed values;

[0109] Step 3: Perform stability analysis on the observer according to the Lyapunov function method and perform delay compensation on the controller.

[0110] As a preferred embodiment of the present invention, the step 1 comprises the following specific steps:

[0111] Step 1.1: Derivation of CETDC mathematical model and construction of super-local model;

[0112] The topology of CETDC is as follows: Figure 1 As shown. The topology contains three phases of the same structure ( j = a, b, c ), each phase energy storage bridge arm consists of a converter valve, N half-bridge sub-modules, a buffer inductor L and its parasitic resistance R in series. The converter valve is composed of several thyristors and anti-parallel diodes in series. Among them, U H and U L Represent the DC voltage on the high voltage side and the low voltage side respectively, I H and I L Represent the DC current on the high voltage side and the low voltage side respectively. and Represent the high-voltage side and low-voltage side currents of each phase respectively, Represents the current of each phase energy storage bridge arm, Represents the voltage of each phase energy storage bridge arm.

[0113] The single-phase structure of CETDC is as follows: Figure 2 As shown, power is transmitted from the high voltage side to the low voltage side, botha Phase is taken as an example for analysis. N The number of submodules used in the energy storage bridge arm; U C is the capacitor voltage of a single submodule; S 1. S 2 is the commutation switch.

[0114] When the energy storage bridge arm is connected in parallel to the high voltage side, the loop equation of the topology is shown in formula (1):

[0115] (1);

[0116] Similarly, when the energy storage bridge arm is connected in parallel to the low-voltage side, the circuit equation of the topology can be obtained as shown in formula (2):

[0117] (2);

[0118] Combining Equations (1) and (2), the average state equation of CETDC can be obtained as shown in Equation (3):

[0119] (3);

[0120] in, is the switch function, S 1 When it is turned on, S=1. S 2 When opened, S=0; is a symbolic function.

[0121] Compared with other control methods, model-free control has stronger adaptability and fault tolerance to disturbances such as internal disturbances, external disturbances, unknown system dynamics, and measurement noise. The first-order hyperlocal model of a single-input single-output system is shown in Equation (4):

[0122] (4);

[0123] in, u is the control variable; y is the output variable; Non-physical scaling factors entered into the system; F Represents the known part and the unknown disturbance part of the system.

[0124] According to formula (4), the super-local model of CETDC is shown in formula (5):

[0125] (5);

[0126] in, For the system a The known and unknown parts of the phase, G is the differential of the disturbance.

[0127] Step 1.2: Analysis of traditional CETDC model-free predictive control method;

[0128] Based on the ultra-local model of CETDC shown in formula (5), a proportional-integral model-free controller can be derived, and the control rate is shown in formula (6):

[0129] (6);

[0130] in, It is the reference value of the system output; is the estimated value of F; is the proportionality coefficient; is the integration coefficient; , is the tracking error of the system.

[0131] In formula (6) The proportional model-free controller can be obtained, and the control rate is shown in formula (7):

[0132] (7);

[0133] The model-free control principle based on the hyperlocal model is applied to the current control of CETDC, and the corresponding controller is called conventional model-free predictive current control. The calculation of the bridge arm voltage reference signal in the three-phase stationary reference frame is shown in Equations (8)-(10):

[0134] (8);

[0135] (9);

[0136] (10);

[0137] Among them, the subscript " a 、 b 、 c "express abc Three-phase components; superscript " ” indicates reference value.

[0138] As a preferred embodiment of the present invention, step 2 includes the following specific steps:

[0139] Step 2.1: Construct NESO to observe the lumped perturbations in the hyperlocal model;

[0140] Based on the super-local model of CETDC in formula (5), a i Pa and F aThe NESO with is the state variable and the bridge arm current error as feedback is expressed as shown in Equation (11):

[0141] (11);

[0142] in, is the observed value of the bridge arm current; is the observed value of the lumped disturbance. e a is the current observation error. L 1 and L 2 is the error feedback gain of the observer.

[0143] The designed nonlinear function expression is shown in formula (12):

[0144] (12);

[0145] in, and is an adjustable parameter of the nonlinear function; The value range of is generally 0 ~ 1. The smaller the value, the stronger the nonlinearity of the function. is the NESO filtering parameter. By selecting appropriate parameters, the system can estimate the state variables and lumped disturbances, making the system have a better tracking filtering effect.

[0146] Step 2.2: Design and select controller parameters;

[0147] In NESO, the parameter L 1 and L 2 will affect system performance, so L 1 and L 2. Make design choices. Rewrite Equation (11) into a matrix form, as shown in Equation (13):

[0148] (13);

[0149] in, ; ; ; ; ; for y Observed values.

[0150] By Laplace transforming Equation (13), the characteristic equation of NESO can be obtained as shown in Equation (14):

[0151] (14);

[0152] in, IRepresents the unit matrix.

[0153] In order to obtain better stability and dynamic characteristics, all roots of the characteristic polynomial of NESO fall on Get L 1 and L 2 As shown in formula (15):

[0154] (15);

[0155] in, To increase the bandwidth of NESO, a comprehensive design is needed from the perspectives of dynamic performance and stability. Since the inner loop generally requires a fast response speed, the bandwidth of NESO should be appropriately increased while ensuring stability.

[0156] In order to make The value of is more intuitive. z In the actual digital control system, the sampling frequency is high enough. The first-order forward Euler discretization is used to discretize Equation (11), as shown in Equation (16):

[0157] (16);

[0158] in, T s is the sampling time; L k1 = T s L 1, L k2 = T s L 2 is the gain of the observer, and its value affects the distribution of the closed-loop poles of the system, thereby affecting the stability of the observer.

[0159] From formula (16), we can see that the transfer function of NESO is shown in formula (17):

[0160] (17);

[0161] The characteristic equation of the observer is shown in formula (18):

[0162] (18);

[0163] From formula (18), we can get The extreme point of is shown in formula (19):

[0164] (19);

[0165] but Can be based on z The poles of the domain are calculated as shown in formula (20):

[0166] (20);

[0167] According to the system stability condition, when the poles of the observer are all distributed in z When the domain is within the unit circle, the system is stable. The choice is such that z1 and z2 are inside the unit circle of the z domain. If Too small, Close to 1, the dynamic performance of the observer will deteriorate. Too big, When it approaches 0, the robustness of the system will deteriorate, which may lead to system divergence. A trade-off between dynamics and robustness should be made to select an appropriate observer bandwidth.

[0168] Step 2.3: Final reference signal calculation;

[0169] Discretize Equation (5) into the first order, as shown in Equation (21):

[0170] (twenty one);

[0171] The required modulation reference signal can be further obtained as:

[0172] (twenty two);

[0173] in, is the NESO observation value, because F a Cannot be derived directly.

[0174] Assumptions Finally reached the reference value , then the final modulation reference signal can be obtained from formula (22):

[0175] (twenty three);

[0176] in, The reference value of the final output modulation signal is synthesized through the carrier phase shift modulation technology to generate the final gate pulse signal.

[0177] As a preferred embodiment of the present invention, step 3 includes the following specific steps:

[0178] In order to ensure the convergence of the proposed NESO current error and disturbance estimation error, the stability of NESO is analyzed using the Lyapunov function method. , Here the nonlinear function Simplified equivalent to e a By analyzing and combining Equation (5) and Equation (11), the error equation can be obtained as shown in Equation (24):

[0179] (twenty four);

[0180] In order to analyze the stability of the proposed observer, the Lyapunov function is defined as follows:

[0181] (25);

[0182] The derivative of formula (25) is:

[0183] (26);

[0184] From formula (15), we can see that From formula (26), we can see that when hour, , which satisfies the stability condition.

[0185] If the disturbance change rate G If is 0, then:

[0186] (27);

[0187] It can be seen that the proposed NESO observation system is stable.

[0188] Considering that digital controllers usually have a one-step delay problem, the delay should be compensated. The final reference signal after compensation is:

[0189] (28);

[0190] in, It can be obtained by Lagrange back extrapolation:

[0191] (29);

[0192] In order to verify the effectiveness of the proposed method, simulation comparative analysis and experimental tests were conducted on the traditional PI control of CETDC and the proposed MFPC method. The control block diagram of the CETDC drive system based on the proposed control method is shown in the figure below: Figure 3 Table 1 lists the control system parameters used in the MATLAB / Simulink digital simulation.

[0193] Table 1 Simulation model parameters

[0194] ;

[0195] Combine Figure 4 and Figure 5 , the power waveforms of PI control and this method can be compared and analyzed; combined with Figure 6 and Figure 7 The bridge arm current waveforms controlled by PI and this method can be compared and analyzed; combined with Figure 8 and Figure 9 The bridge arm voltage waveforms controlled by PI and this method can be compared and analyzed. Figure 10 and Figure 11 Comparative analysis of the capacitor voltage waveforms for PI control and the proposed control method reveals no overshoot during the bridge arm voltage startup process. The proposed model-free predictive control method exhibits a faster response than PI control, enabling the system to quickly reach a stable operating state, reducing the time the system experiences transient states, resulting in a smoother startup process and enhanced system reliability. Furthermore, the proposed MFPC control method requires only one control parameter to be adjusted, reducing control complexity.

[0196] In summary, the model-free predictive control strategy proposed in this paper uses only the system inputs and outputs without considering any other system parameters. This reduces tuning workload and computation time, while also achieving better performance in terms of reduced tracking error, dynamic overshoot, and transient time experienced by the system, thereby enhancing system robustness. Simulation results demonstrate that the proposed control strategy outperforms traditional PI control in both dynamic and static conditions in terms of response speed and control accuracy.

[0197] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A model-free predictive control method for CETDC, characterized in that: The following steps are involved: Step 1: Analyze the topological structure of CETDC, derive the mathematical model of CETDC, establish its hyperlocal model based on the mathematical model of CETDC, and analyze the traditional model-free predictive control method of CETDC; Step 2: Build a NESO to estimate the system state and disturbance, perform feedforward compensation for the disturbance, design and select controller parameters, and calculate the final reference signal from the observed values; Step 3: Perform stability analysis on the observer according to the Lyapunov function method and perform delay compensation on the controller; The specific steps of step 1 are as follows: Step 1.1: Derivation of CETDC mathematical model and construction of super-local model; Step 1.2: Analysis of CETDC’s traditional model-free predictive control method; The step 1.1 includes the following specific steps: The CETDC topology consists of three identical phases, where j = a, b, and c. Each phase energy storage bridge arm includes a converter valve, N half-bridge submodules, a snubber inductor L, and its parasitic resistance R in series. The converter valve is composed of several thyristors and anti-parallel diodes in series. U H Represents the DC voltage on the high voltage side, U L Represents the DC voltage on the low-voltage side; I H Represents the DC current on the high voltage side, I L Represents the DC current on the low-voltage side; i Hj Represents the high voltage side current of each phase, i Lj Represents the low-voltage side current of each phase; i Pj Represents the current of each phase energy storage bridge arm, u Pj Represents the voltage of each phase energy storage bridge arm; In the single-phase structure of CETDC, power is transmitted from the high-voltage side to the low-voltage side. In phase a, N is the number of submodules put into use in the energy storage bridge arm; U C is the capacitor voltage of a single submodule; S1 and S2 are commutation switches; When the energy storage bridge arm is connected in parallel to the high voltage side, the loop equation of the topology is shown in formula (1): Similarly, when the energy storage bridge arm is connected in parallel to the low-voltage side, the loop equation of the topology is shown in formula (2): Combining equations (1) and (2), the average state equation of CETDC is shown in equation (3): Where S = (1, 0) is the switching function, S = 1 when S1 is turned on, and S = 0 when S2 is turned on; sign(·) is the sign function; The first-order hyperlocal model of a single-input single-output system is shown in formula (4): Where u is the control variable; y is the output variable; α is the non-physical proportional coefficient of the system input; F represents the known part and the unknown disturbance part of the system; According to formula (4), the super-local model of CETDC is shown in formula (5): Among them, F a is the known part and unknown part of phase a of the system, G is the differential of the disturbance; The step 1.2 includes the following specific steps: Based on the ultra-local model of CETDC shown in formula (5), a proportional-integral model-free controller is obtained, and the control rate is shown in formula (6): （6） Among them, y * It is the reference value of the system output; is the estimated value of F; K p is the proportional coefficient; K i is the integral coefficient; e=y * -y, is the tracking error of the system; Let K in formula (6) i = 0, a proportional model-free controller can be obtained, and the control rate is shown in formula (7): The model-free control principle based on the hyperlocal model is applied to the current control of CETDC. The corresponding controller is called conventional model-free predictive current control. The bridge arm voltage reference signal in the three-phase stationary reference frame is calculated as shown in Equations (8)-(10): The subscripts "a, b, c" represent the three-phase components a, b, and c; the superscript "*" represents the reference value.

2. The model-free predictive control method for CETDC according to claim 1, characterized in that: Further technical solution, said step 2 includes the following specific steps: Step 2.1: Construct the lumped perturbation in the NESO observation hyperlocal model; Step 2.2: Design and select controller parameters; Step 2.3: Final reference signal calculation.

3. The model-free predictive control method for CETDC according to claim 2, characterized in that: The step 2.1 includes the following specific steps: Based on the super-local model of CETDC in formula (5), we establish Pa and F a The NESO with is the state variable and the bridge arm current error as feedback is expressed as shown in formula (11): in, is the observed value of the bridge arm current; is the observed value of the lumped disturbance; e a is the current observation error; L1 and L2 are the error feedback gains of the observer; The designed nonlinear function expression is shown in formula (12): Among them, η and λ are adjustable parameters of the nonlinear function; the value of η ranges from 0 to 1, and the smaller the value, the stronger the nonlinearity of the function; λ is the filtering parameter of NESO.

4. The model-free predictive control method for CETDC according to claim 3, characterized in that: The step 2.2 includes the following steps: In NESO, the parameters L1 and L2 affect the system performance, so L1 and L2 should be designed and selected. Rewrite Equation (11) into a matrix form, as shown in Equation (13): where z = [z1 z2] T ; B=[α 0] T ; C = [1 0]; L = [-L1 -L2] T ; is the observed value of y; By Laplace transforming Equation (13), the characteristic equation of NESO is obtained as shown in Equation (14): |sI-(A-LC)|=s 2 +L1s+L2 (14) Where I represents the unit matrix; In order to ensure stability and dynamic characteristics, all roots of the characteristic polynomial of NESO fall at -ω0, and L1 and L2 are obtained as shown in Equation (15): Among them, ω b is the bandwidth of NESO. In order to make ω b The value of is more intuitive, and it is designed in the z domain; in actual digital control systems, the sampling frequency is high enough; the first-order forward Euler discretization is used to discretize Equation (11), as shown in Equation (16): Among them, T s is the sampling time; L k1 =T s L1, L k2 =T s L2 is the gain of the observer, and its value affects the distribution of the closed-loop poles of the system, thereby affecting the stability of the observer; From formula (16), we can see that the transfer function of NESO is shown in formula (17): The characteristic equation of the observer is shown in formula (18): z 2 +(L k1 -2)z-L k1 +L k2 T s +1=0 (18) The extreme points of G(z) can be obtained from equation (18), as shown in equation (19): z 1,2 =1-ω b T s (19) Then ω b It can be calculated based on the poles in the z domain, as shown in formula (20): According to the system stability condition, when all the poles of the observer are distributed within the unit circle of the z domain, the system is stable; ω b The choice of z1 and z2 makes z1 and z2 inside the unit circle in the z domain, and the observer bandwidth is selected based on the balance between dynamics and robustness.

5. The model-free predictive control method for CETDC according to claim 4, characterized in that: The step 2.3 includes the following steps: Discretize Equation (5) into the first order, as shown in Equation (21): i Pa (k+1)=i Pa (k)+T s F a (k)+αT s N(k) (21) The required modulation reference signal can be further obtained as: in, is the NESO observation value, because F a It cannot be directly derived; Assumption i Pa (k+1) eventually reaches the reference value The final modulation reference signal obtained by formula (22) is: in, The reference value of the final output modulation signal is synthesized through the carrier phase shift modulation technology to generate the final gate pulse signal.

6. The model-free predictive control method for CETDC according to claim 5, characterized in that: The step 3 includes the following specific steps: definition Here the nonlinear function fal(e a ,η,λ) is simplified to be equivalent to e a After analysis, combining equations (5) and (11), we get the error equation as shown in equation (24): The Lyapunov function is defined as follows: The derivative of formula (25) is: From formula (15), we know that L1, L2>0; from formula (26), we know that when hour, Satisfy stability conditions; If the disturbance change rate G is 0, then: It can be seen that the proposed NESO observation system is stable; Considering that digital controllers usually have a one-step delay, the delay should be compensated. The final reference signal after compensation is: in, It can be obtained by Lagrange back extrapolation:

Images