A virtual direct-current motor model-free self-learning control method for a port-berthing commercial ship micro-grid

Abstract

The application provides a kind of virtual DC motor model-free self-learning control method of port commercial ship micro-grid, to improve the stability of ship micro-grid DC bus voltage under the access of port commercial ship shore power, comprising the following steps: collecting ship micro-grid DC bus voltage, establishing ship virtual DC motor model and discretizing armature equation;Establish a tight format local linearization data model related to virtual DC motor input moment of inertia and output DC bus voltage;Design moment of inertia model-free self-learning controller and calculate pseudo partial derivative according to input and output data;Design adaptive adjustment of moment of inertia;By controlling ship micro-grid Buck / Boost converter, simulate the external characteristic of DC generator to stabilize DC bus voltage.The application effectively suppresses the fluctuation of ship micro-grid DC bus voltage caused by load switching and new energy output change under the access of port commercial ship shore power, while also optimizes the dynamic response speed of ship micro-grid DC bus voltage regulation.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of port energy control, and particularly relates to a virtual direct-current motor model-free self-learning control method for a port-berthing merchant ship micro-grid. BACKGROUND

[0002] A green port promotes the green and low-carbon circular development of the port by establishing a port ship micro-grid connected with shore power by new energy such as solar photovoltaic, wind power, hydrogen fuel, energy storage power supply, and LNG. Unlike a land micro-grid, the control mechanism is complex, the power supply types are various, the load types are mostly motor loads, and the load switching is frequent under the shore-based power supply of a berthing ship. Due to the lack of inertia and damping when the micro-grid direct-current power electronic converter operates, the intermittent characteristics of new energy generation and the frequent use of high-power equipment such as ship-mounted cranes on ships easily cause the voltage fluctuation and oscillation of the ship micro-grid bus, and even cause the instability of the ship micro-grid. By controlling the technology, the ship micro-grid direct-current power electronic converter can have damping and inertia, which can increase the stability of the ship micro-grid. Therefore, the control problem of the ship micro-grid direct-current power electronic converter has become the focus of research by scholars in the relevant field.

[0003] To solve the fluctuation of the direct-current bus voltage caused by the random fluctuation of new energy generation and load, Huang Di and Fan Shaosheng of Changsha University of Science and Technology proposed a virtual direct-current motor control technology, which applies the mechanical equation and armature equation of a direct-current generator to the control algorithm to simulate the inertia characteristics and damping characteristics of the direct-current generator, so that the micro-grid direct-current bus voltage can remain stable when new energy generation fluctuates and load suddenly changes. However, this method uses a fixed rotational inertia and does not realize flexible control of parameters, resulting in poor voltage regulation dynamic characteristics and making it difficult to apply to a ship micro-grid with frequent load switching. Zhang Qinjin of Dalian Maritime University proposed a virtual generator control method for direct-current micro-sources based on parameter self-adaptation, which introduces a PI loop into the design of the rotational inertia parameter, gives an adaptive adjustment equation of the rotational inertia, and realizes adaptive adjustment of the rotational inertia of the virtual direct-current generator, thereby improving the dynamic response speed of the voltage regulation of the micro-grid system. However, due to the introduction of error integral feedback, the PID control makes the voltage control process of the micro-grid system prone to oscillation, which cannot meet the demand of sensitive load of the ship micro-grid for stable voltage of the direct-current bus. Model-free self-learning control is a data-driven control method that processes the input and output data of the controlled system by using a compact format dynamic linearization method to convert it into a linear affine data model with nonlinear terms. The model-free self-learning control method includes an adaptive estimation algorithm for time-varying linear parameters and a time difference estimation algorithm for uncertain terms, only uses online input and output data of the controlled system for controller design, does not require any model information, has strong robustness, and is very suitable for application in the field of port energy control technology with unknown model structure and nonlinearity.

[0004] Based on the above analysis, the present application proposes a virtual direct current motor model-free self-learning control method for port commercial ship micro-grid, which only uses the input and output data of the virtual direct current generator control system of the ship direct current micro-grid, improves the virtual direct current generator through the model-free self-learning control algorithm to realize the adaptive adjustment of the moment of inertia, and further improves the voltage fluctuation problem of the ship micro-grid caused by the load switching and the change of new energy output under the shore power access of the port commercial ship, and has the advantages of fast dynamic response speed of voltage regulation, and improves the stability of the ship micro-grid. SUMMARY

[0005] Therefore, the present application aims to provide a virtual direct current motor model-free self-learning control method for port commercial ship micro-grid, which uses model-free self-learning control to improve the virtual direct current generator control, designs an adaptive adjustment control algorithm of the moment of inertia, dynamically adjusts the moment of inertia in the virtual direct current generator control, and stabilizes the direct current bus voltage by controlling the Buck / Boost converter of the ship micro-grid to simulate the external characteristic of the direct current generator. The dynamic response speed of the direct current bus voltage regulation of the ship micro-grid is optimized, the voltage fluctuation is limited within a safe range, and the direct current bus voltage fluctuation problem caused by the load switching and the change of new energy output of the ship micro-grid under the shore power access of the port commercial ship is improved.

[0006] In order to achieve the above-mentioned application purposes, the present application adopts the following technical solutions:

[0007] S1: Collecting the direct current bus voltage of the ship micro-grid, establishing a virtual direct current motor model and performing discretization processing on the armature equation;

[0008] S2: Establishing a tight format local linearization data model related to the input moment of inertia and the output direct current bus voltage of the virtual direct current motor;

[0009] S3: Designing a moment of inertia model-free self-learning controller and calculating pseudo partial derivatives according to the input and output data;

[0010] S4: Designing adaptive adjustment of the moment of inertia;

[0011] S5: Stabilizing the direct current bus voltage by controlling the Buck / Boost converter of the ship micro-grid to simulate the external characteristic of the direct current generator;

[0012] Further, in step S1, the collecting of the direct current bus voltage of the ship micro-grid, the establishment of the virtual direct current motor model and the discretization processing of the armature equation specifically include: simulating the output external characteristic of the direct current generator by establishing a virtual direct current motor model.

[0013] (1) The mechanical equation of the virtual direct current generator is:

[0014]

[0015] wherein,

[0016] ω e = ωp, ω represents the actual mechanical angular velocity; ω n represents the rated electrical angular velocity; J is the moment of inertia; D represents the damping coefficient; T m is the mechanical torque; T e is the electromagnetic torque; ω e represents the actual electrical angular velocity; p is the number of generator pole pairs; the electromagnetic power P e = EI a ; E is the armature electromotive force; I a is the armature current;

[0017] The armature equation of the virtual DC generator is:

[0018]

[0019] wherein,

[0020] R a is the armature resistance; C T is the torque coefficient; Φ is the magnetic flux; U represents the micro-grid DC bus voltage;

[0021] (2) Discretize the armature equation in combination with the mechanical equation of the virtual DC generator:

[0022]

[0023] wherein,

[0024] J(t) represents the moment of inertia at time t; U(t) represents the micro-grid DC bus voltage at time t; U(t+1) represents the micro-grid DC bus voltage at time t+1;

[0025] Further, the discretized armature equation of step (2) satisfies:

[0026] The partial derivative of the equation with respect to J(t) exists and is continuous;

[0027] The equation satisfies the generalized Lipschitz condition, that is, given any U(t1)≠U(t2) (t1≠t2 and t1, t2≥0), |J(t1+1)-J(t2+1)|≤m|U(t1)-U(t2)| can be obtained, where m is a normal number;

[0028] Further, in step S2, the establishment of the tight format local linearization data model related to the input moment of inertia and the output voltage of the virtual DC motor specifically includes:

[0029] The discrete-time nonlinear system is established as follows:

[0030] U(t+1) = f(U(t),…, U(t-n U ), J(t),…, J(t-n J ));

[0031] wherein U(t) represents the ship DC micro-grid voltage at t time, J(t) represents the moment of inertia at t time, and and respectively represent the input and output of the system at t time; n U and n J are two unknown positive integers; f(…) is an unknown nonlinear function;

[0032] The traditional model-free adaptive algorithm is used to analyze the system, and the next time ship DC micro-grid voltage change of the system is related to the moment of inertia change at the previous time, and the following is obtained:

[0033] The nonlinear function f(…) has a continuous partial derivative with respect to the system input variable J(t);

[0034] The nonlinear system satisfies the generalized Lipschitz condition, that is, given any U(t1)≠U(t2)(t1≠t2 and t1, t2≥0), |J(t1+1)-J(t2+1)|≤m|U(t1)-U(t2)| can be obtained, wherein m is a normal number;

[0035] The nonlinear term γ(t)

[0036]

[0037] For the ship micro-grid system satisfying the above conditions, there is a time-varying parameter vector ξ(t) that makes the system into the following compact format local linearization data model:

[0038] U(t+1) = U(t) + ξ(t)ΔJ(t) + γ(t);

[0039] wherein,

[0040] U(t+1) represents the armature electromotive force at t+1 time; ΔJ(t) = J(t)-J(t-1); J(t-1) represents the moment of inertia at t-1 time; ξ(t) represents the pseudo partial derivative of the system;

[0041] Further, in step S3, the design of the moment of inertia model-free self-learning controller and the calculation of the pseudo partial derivative according to the input and output data specifically includes:

[0042] (1) considering the input criterion function G(J(t)) = |U r (t+1)-U(t+1)| 2+ p | J(t) - J(t - 1) | 2 ;

[0043] where U r (t + 1) represents the expected DC bus voltage of the ship microgrid; η > 0 is a weight factor;

[0044] Deriving the input criterion function with respect to J(t) and setting it equal to 0, the model-free self-learning control rate is obtained as:

[0045]

[0046] where μ ∈ (0, 1] is a step factor;

[0047] (2) For the pseudo partial derivative of the system, an index function ξ(t) is given, and the index function is given as:

[0048]

[0049] where ρ > 0 is a weight factor;

[0050] Deriving the index function with respect to ξ(t) and setting it equal to 0, the inertia pseudo partial derivative estimation rate under model-free adaptive control is calculated as:

[0051]

[0052] where

[0053] is the estimated value of ξ(t); is the estimated value of ξ(t - 1); λ ∈ (0, 2) is a step factor, which makes the control algorithm more flexible;

[0054] For the nonlinear term γ(t), the input-output data before is used to estimate γ(t) at the current time, which is calculated as follows:

[0055]

[0056] where

[0057] is the estimated value of γ(t); ΔU(t) = U(t) - U(t - 1); ΔJ(t - 1) = J(t - 1) - J(t - 2); J(t - 2) represents the inertia at t - 2; U(t - 1) represents the microgrid DC bus voltage at t - 1;

[0058] Further, in step S4, the adaptive adjustment of the moment of inertia specifically includes:

[0059] According to the maximum power P maxThe design principle of the moment of inertia J is obtained:

[0060]

[0061] Assuming that there is a constant b, when the ship direct current micro-grid accesses a small load, the voltage influence of the ship direct current micro-grid is ΔU < b, a smaller J0 is selected; when the ship direct current micro-grid accesses a large load ΔU > b, dU / dt < 0, the model-free adaptive control algorithm is used to select a suitable moment of inertia J; when the load is removed ΔU > b, dU / dt > 0, in order to prevent the voltage fluctuation caused by too large dU / dt, a larger moment of inertia J = 50 is selected;

[0062] The model-free adaptive control algorithm of the moment of inertia is as follows:

[0063]

[0064] The model-free learning controller utilizes the ship micro-grid direct current bus voltage output from the ship virtual direct current generator and the expected ship micro-grid direct current bus voltage, and the output J(t) of the model-free learning controller tracks the load adjustment demand in real time. The adaptive adjustment of the moment of inertia J is realized through the model-free learning control algorithm design of the virtual direct current generator control, so as to make up for the deficiency of the fixed moment of inertia of the virtual direct current generator, and improve the response speed and stability of the voltage regulation of the ship direct current micro-grid system.

[0065] Compared with the prior art, the control method adopted in the present application improves the ship virtual direct current generator control through the model-free learning control, designs the adaptive adjustment of the moment of inertia parameter, dynamically adjusts the moment of inertia of the ship direct current virtual synchronous generator, reduces the direct current bus voltage fluctuation of the ship micro-grid caused by the ship load switching and the new energy output change under the port calling merchant ship shore power access, optimizes the response speed of the ship direct current micro-grid bus voltage regulation, and meets the requirements of the ship micro-grid operation voltage stability. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 is the principle block diagram of the model-free learning control method of the virtual direct current generator;

[0067] Figure 2 is the flowchart of an embodiment of the model-free learning control method of the virtual direct current generator of the port calling merchant ship micro-grid proposed in the present application;

[0068] Figure 3 is the direct current bus voltage curve of the model-free learning control method of the virtual direct current generator of the port calling merchant ship micro-grid proposed in the present application under the load power mutation;

[0069] Figure 4The application provides a virtual direct-current motor model-free self-learning control method for a port-berthing merchant ship micro-grid. DETAILED DESCRIPTION

[0070] For the purpose of more clearly illustrating the examples of the application, the application will be described below in combination with the drawings.

[0071] The application considers the problems of ship direct-current bus voltage fluctuation of a port-berthing merchant ship micro-grid caused by frequent load switching and new energy output change and slow response speed of a traditional virtual direct-current generator with fixed rotational inertia under shore power access of the port-berthing merchant ship, and proposes a virtual direct-current motor model-free self-learning control method for the port-berthing merchant ship micro-grid by combining a micro-grid direct-current converter control mode of the virtual direct-current generator.

[0072] Please refer to Figures 1-4 For the virtual direct-current motor model-free self-learning control method for the port-berthing merchant ship micro-grid, the method improves the ship virtual direct-current generator control through model-free self-learning control, designs adaptive adjustment of the rotational inertia parameter, dynamically adjusts the rotational inertia of the ship direct-current virtual direct-current generator, reduces the ship micro-grid direct-current bus voltage fluctuation of the port-berthing merchant ship micro-grid caused by frequent load switching and new energy output change under shore power access of the port-berthing merchant ship, and optimizes the response speed of the ship direct-current micro-grid bus voltage regulation.

[0073] Figure 2 The application is a flow chart of a virtual direct-current motor model-free self-learning control method for a port-berthing merchant ship micro-grid, and specifically includes the following steps.

[0074] Step 1: Collecting the ship micro-grid direct-current bus voltage, establishing a ship virtual direct-current motor model and performing discretization processing on the armature equation.

[0075] (1) Virtual direct-current motor control

[0076] Figure 1 The application is a principle block diagram of a virtual direct-current generator model-free self-learning control method. In the figure, U ref represents a bus voltage reference value; I ref represents a converter output current reference value; U1 represents a virtual direct-current generator output voltage; U2 represents a converter output voltage; I1 represents a virtual direct-current generator output current; I2 represents a converter output current; according to the power balance principle, U ref / U1 is used to convert I ref into an input current reference value, and finally the required control signal is obtained through a PI controller of a current loop and PWM modulation.

[0077] The virtual DC generator control provides additional inertia and damping support for the system by simulating the external characteristic of the DC motor, and the DC bus voltage is stabilized by controlling the external characteristic of the DC motor of the micro-grid Buck / Boost converter. Therefore, the mechanical equation of the virtual DC generator can be obtained as shown in the following formula by the model of the virtual DC generator:

[0078]

[0079] wherein ω e represents the actual mechanical angular velocity; ω n represents the rated electrical angular velocity; ω n = ω0p, ω0 is the rated mechanical angular velocity; J is the moment of inertia; D represents the damping coefficient; T m is the mechanical torque; T e is the electromagnetic torque; ω e represents the actual electrical angular velocity; p is the number of generator pole pairs; the electromagnetic power P e = EI a ; E is the armature electromotive force; I a is the armature current;

[0080] The armature equation of the virtual DC generator is as follows:

[0081]

[0082] wherein R a is the armature resistance; C T is the torque coefficient; Φ is the magnetic flux; U is the DC bus voltage of the micro-grid;

[0083] (2) The discretization of the armature equation can be obtained by combining the mechanical equation of the virtual DC generator:

[0084]

[0085] wherein J(t) represents the moment of inertia at time t; U(t) represents the DC bus voltage at time t; U(t+1) represents the DC bus voltage at time t+1;

[0086] In order to improve the accuracy of the discretization equation of the virtual DC generator armature, the discretization equation (3) satisfies the following assumptions:

[0087] Assumption 1: The partial derivative of the equation with respect to J(t) exists and is continuous;

[0088] Assumption 2: The equation satisfies the generalized Lipschitz condition, that is, given any U(t1)≠U(t2) (t1≠t2 and t1, t2≥0), |J(t1+1)-J(t2+1)|≤m|U(t1)-U(t2)| can be obtained, wherein m is a normal number;

[0089] For the virtual rotational inertia of the virtual DC generator control strategy, the variable is continuous and differentiable, so assumption 1 is true. In addition, the limited change of rotational inertia will not cause the sharp fluctuation of the ship DC microgrid bus voltage, so assumption 2 is true.

[0090] Step two: establish a tight format local linearization data model related to the input rotational inertia of the virtual DC motor and the output DC bus voltage.

[0091] The discrete-time nonlinear system is established as follows:

[0092] U(t+1)=f(U(t),…,U(t-n U ),J(t),…,J(t-n J )) (4)

[0093] Where U(t) represents the DC bus voltage of the ship microgrid at time t, J(t) represents the rotational inertia at time t, and U(t) and J(t) represent the input and output of the system at time t, respectively; n U and n J are two unknown positive integers; f(…) is an unknown nonlinear function;

[0094] The traditional model-free self-learning algorithm is used to analyze the system. Considering that the next time ship DC microgrid voltage change is related to the rotational inertia change at the previous time, we get:

[0095] The nonlinear function f(…) has a continuous partial derivative with respect to the system input variable J(t);

[0096] The nonlinear system satisfies the generalized Lipschitz condition, that is, given any U(t1)≠U(t2)(t1≠t2 and t1,t2≥0), we can get |J(t1+1)-J(t2+1)|≤m|U(t1)-U(t2)|, where m is a normal number;

[0097] Define the nonlinear term γ(t)

[0098]

[0099] For the ship microgrid system that meets the above conditions, there must be a time-varying parameter vector ξ(t) that makes the system into the following tight format local linearization data model:

[0100] U(t+1)=U(t)+ξ(t)ΔJ(t)+γ(t) (6)

[0101] Where ΔJ(t)=J(t)-J(t-1); J(t-1) represents the rotational inertia at time t-1; ξ(t) represents the pseudo partial derivative of the system;

[0102] Step three: design the inertia model-free self-learning controller and calculate the pseudo partial derivative according to the input and output data.

[0103] (1) Consider the input criterion function:

[0104] G(J(t)) = |U r (t+1)-U(t+1)| 2 +ρ|J(t)-J(t-1)| 2 (7)

[0105] Where, U r (t+1) represents the expected DC bus voltage of the ship micro-grid; η>0 is the weight factor;

[0106] Derive the model-free self-learning control rate by taking the derivative of the input criterion function on both sides with respect to J(t) and setting it to zero:

[0107]

[0108] Where, μ∈(0,1] is the step factor;

[0109] (2) For the pseudo partial derivative of the system, give the index function ξ(t), and give the index function:

[0110]

[0111] Where, ρ>0 is the weight factor;

[0112] Take the derivative of the index function on both sides with respect to ξ(t) and set it to zero to calculate the inertia pseudo partial derivative estimation rate under model-free adaptive control:

[0113]

[0114] Where, is the estimated value of ξ(t); is the estimated value of ξ(t-1); λ∈(0,2) is the step factor, which makes the control algorithm more flexible;

[0115] For the nonlinear term γ(t), use the previous input and output data to estimate γ(t) at the current time, which is calculated as follows:

[0116]

[0117] Where, γ(t) is the estimated value of γ(t); ΔU(t) = U(t)-U(t-1); ΔJ(t-1) = J(t-1)-J(t-2); J(t-2) represents the inertia at t-2; U(t-1) represents the micro-grid DC bus voltage at t-1;

[0118] Step four: design of the adaptive adjustment of the moment of inertia:

[0119] According to the maximum power P max Get the design principle of the moment of inertia J:

[0120]

[0121] Assuming there is a constant b, when the ship DC micro-grid access to small load, the impact of the ship DC micro-grid voltage ΔU < b, select a smaller J0; when the ship DC micro-grid access to large load ΔU > b, dU / dt < 0, using model-free adaptive control algorithm to select the appropriate moment of inertia J; when the load is removed ΔU > b, dU / dt > 0, in order to prevent dU / dt too large caused by voltage fluctuation, select a larger moment of inertia J = 50;

[0122] The design of the moment of inertia model-free adaptive control algorithm is as follows:

[0123]

[0124] Step five: through the control of the ship micro-grid Buck / Boost converter to simulate the DC generator external characteristic to stabilize the DC bus voltage.

[0125] The model-free learning controller uses the output DC micro-grid voltage from the ship virtual DC generator and the expected ship micro-grid DC bus voltage, and its output J(t) real-time tracking load adjustment requirements. Through the design of the virtual DC generator control by model-free learning control algorithm, the adaptive adjustment of the moment of inertia J is realized, so as to make up for the deficiency of the fixed moment of inertia of the virtual DC generator, and improve the response speed and stability of the voltage regulation of the ship DC micro-grid system.

[0126] Therefore, by controlling the ship micro-grid Buck / Boost converter to simulate the DC generator external characteristic, the adjustment of the moment of inertia is realized to stabilize the ship micro-grid DC bus voltage under complex sea conditions.

[0127] Therefore, the port calling merchant ship micro-grid virtual direct current motor model-free self-learning control method considers the problems of ship direct current bus voltage fluctuation caused by frequent load switching and new energy output change of the ship micro-grid under the shore power access of the port calling merchant ship and slow response speed of the traditional virtual direct current generator adopting fixed rotational inertia voltage regulation, improves the ship virtual direct current generator control through the model-free self-learning control, designs adaptive adjustment of the rotational inertia parameter, dynamically adjusts the rotational inertia of the ship direct current virtual direct current generator, reduces the ship direct current micro-grid voltage fluctuation caused by frequent load switching and new energy output change of the ship micro-grid under the shore power access of the port calling merchant ship, and optimizes the response speed of the ship direct current micro-grid bus voltage regulation.

[0128] The control system of the embodiment is simulated and analyzed below.

[0129] The reference of the ship micro-grid direct current bus is set as 1250V of the Zhongfengfei roll-on / roll-off wheel micro-grid voltage, the port calling merchant ship micro-grid virtual direct current motor model-free self-learning control system is built by using MATLAB / Simulink and is analyzed, the distributed energy direct current converter parameter of the virtual direct current generator algorithm is set as follows: the output side capacitor of the photovoltaic converter is 705 mu F; the input side inductance of the photovoltaic converter is 2.4 Mh; the switching frequency is 50 kHz; the damping coefficient D is 0.01. p The total simulation duration is set as 10 s, and the ship micro-grid is input with 12 kW of direct current load at 5 s. Figure 3 The direct current bus voltage curve under the load power mutation of the port calling merchant ship micro-grid virtual direct current motor model-free self-learning control method proposed in the application is shown in the following figure, Figure 3 It can be seen that when the load switching occurs, the output voltage fluctuation range of the distributed energy direct current converter adopting the virtual direct current generator algorithm is only 30 V, the direct current bus voltage fluctuation is effectively inhibited, and the voltage regulation is realized in 0.3 s.

[0130] Figure 4 The rotational inertia dynamic change curve under the load power mutation of the port calling merchant ship micro-grid virtual direct current motor model-free self-learning control method proposed in the application is shown in the following figure, according to the rotational inertia dynamic change curve under the load power mutation of the port calling merchant ship micro-grid virtual direct current motor model-free self-learning control method proposed in the application, Figure 4 It can be seen that the ship micro-grid virtual direct current generator model-free adaptive control method proposed in the application can realize adaptive adjustment of the rotational inertia according to the voltage fluctuation, and the voltage regulation speed of the virtual direct current motor control method is optimized.

[0131] The port calling merchant ship micro-grid virtual direct current motor model-free self-learning control method is proposed in the embodiment. Through the simulation experiment, the control strategy effectively inhibits the bus voltage fluctuation of the ship micro-grid system, and obviously improves the voltage regulation dynamic response speed of the ship micro-grid system.

[0132] For those skilled in the art, corresponding changes and modifications, beautification, combination can be made according to the technical solutions and concepts described above, and all the changes, modifications, beautification and combination should belong to the protection scope of the claims of the present application.

Claims

1. A model-free self-learning control method for a virtual DC motor in a microgrid on a merchant ship docking at port, characterized in that: The method comprises the following steps: S1: Collecting the ship micro-grid DC bus voltage, establishing a virtual DC motor model and discretizing the armature equation, specifically including: S11: Simulating the output external characteristic of the DC generator by establishing a virtual DC motor model: The mechanical equation of the virtual DC generator is: ； wherein, represents the actual electrical angular velocity; represents the rated electrical angular velocity; is the moment of inertia; represents the damping torque; is the mechanical torque; is the electromagnetic torque; is the number of generator pole pairs; electromagnetic power ; is the armature electromotive force; is the armature current; The armature equation of the virtual DC generator is: ； wherein, is an armature resistance; is a torque coefficient; is a magnetic flux; is a ship microgrid DC bus voltage; S12: Discretizing the armature equation in combination with the mechanical equation of the virtual DC generator: ； wherein, denotes t the moment of inertia at the time instant; denotes t the microgrid DC bus voltage at the time instant; denotes t+ the microgrid DC bus voltage at the time instant; S2: Establishing a compact local linearization data model related to the input moment of inertia and the output DC bus voltage of the virtual DC motor, specifically including: S21: Establishing a discrete-time nonlinear system as follows: ； wherein, and are two unknown positive integers; is an unknown nonlinear function; S22: Analyzing the system using a model-free self-learning algorithm, considering that the next time ship micro-grid DC bus voltage change is related to the previous time moment of inertia change, obtaining: Non-linear function There exists a continuous partial derivative with respect to the system input variable x; The nonlinear system satisfies the generalized Lipschitz condition, that is, given any , and , the following can be obtained wherein is a positive constant; Defining the non-linear term : ； For the ship micro-grid system satisfying the above conditions, there is a time-varying parameter vector Discrete-time nonlinear systems are converted into the following compact format local linearization data model: ； wherein ; denotes the moment of inertia at the time instant; denotes the pseudo partial derivative of the system; S3: Designing a moment of inertia model-free self-learning controller and calculating pseudo partial derivatives according to input and output data, specifically including: S31: Consider the input criterion function ; wherein, Vdc_desired represents the ship microgrid desired DC bus voltage; is a weight factor; The input criterion function is derived on both sides with respect to The derivation is taken and set equal to 0, and the model-free self-learning control rate is obtained as: ； wherein is a step factor; S32: Give an indicator function for the pseudo-derivative of the system , give an indicator function: ； wherein is a weight factor for the pseudo-derivative estimate rate; Taking the derivative of the index function on both sides with respect to and setting it equal to zero, the inertia pseudo-derivative estimation rate under model-free learning control is calculated: ； wherein is an estimate of is an estimate of is a step factor, making the control algorithm more flexible; For the nonlinear term , the previous input-output data is used to estimate the current time , which is calculated as follows: ； wherein is an estimate of ; ; denotes t the moment of inertia at time -2; denotes t the microgrid DC bus voltage at time -1; S4: Designing adaptive adjustment of the moment of inertia, specifically including: According to the maximum power output of the ship micro-grid DC converter Obtaining moment of inertia J Design principles: ； Assuming the existence of a constant b, what is the impact on the voltage of the ship's DC microgrid when a small load is connected to it? When choosing, select the smaller value. J 0; When a ship's DC microgrid is connected to a large load , At that time, a model-free self-learning control algorithm is used to select a suitable moment of inertia. J When the load is removed , At that time, in order to prevent Excessive voltage fluctuations necessitate the selection of a larger moment of inertia. J =50; The moment of inertia model-free self-learning control algorithm is designed as follows: ； S5: Stabilizing the DC bus voltage by controlling the ship micro-grid Buck / Boost converter to simulate the external characteristic of the DC generator.

Images