Intelligent predictive control method based on self-adaptive three-phase staggered hydrogen production converter

Abstract

The invention discloses an intelligent predictive control method based on a self-adaptive three-phase interleaved hydrogen production converter, and relates to the technical field of power electronics. A hyperlocal model of the converter is constructed, so that the dependence of a control system on an accurate mathematical model is reduced, and the reliability of the converter is improved. In the hydrogen production system, a nonlinear expansion state observer is used for monitoring total disturbance in a voltage and current control loop and performing real-time compensation; meanwhile, by utilizing a dead-beat prediction control technology, the system can predict a future reference vector and calculate a direct-current voltage value according to the current and target current magnitudes, so that parameters in the hydrogen production process are learned, continuously adjusted and optimized from actual operation data in real time, the stability and reliability of hydrogen production are improved, and the method is suitable for popularization and application. Rapid dynamic response is achieved, the prediction control precision is improved, and the electrolytic cell can generate a certain amount of hydrogen at a stable rate and flow.

Description

Technical Field

[0001] The present invention relates to the field of power electronics technology, and particularly to an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter. Background Art

[0002] In switching power supplies and energy conversion systems, especially at the present stage of the rapid development of hydrogen energy, three-phase interleaved parallel DC-DC converters are widely used in high-power hydrogen production scenarios. During the hydrogen production process, due to the involvement of complex chemical reactions and physical processes, it may be difficult to establish an accurate environmental model or the calculation cost may be too high. At the same time, traditional control methods such as PID control have limitations in dealing with system dynamic response, robustness, and load change adaptation. For example, traditional control algorithms mostly adopt PI control, etc., but the switching frequency is not fixed, resulting in a large current ripple, and its control performance depends to a large extent on the accuracy of model parameters. Under actual working conditions, when subjected to internal and external disturbances such as temperature changes, vibrations, and sudden load changes, the parameters of inductive sensitive devices are prone to non-linear changes, causing a mismatch between the actual values and the system mathematical model. Summary of the Invention

[0003] The present invention proposes an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter for one or more of the above existing problems.

[0004] According to the first aspect of the present application, an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter is proposed, including the following steps:

[0005] S100. Establish a circuit mathematical model of a three-phase interleaved parallel Buck converter, and establish a corresponding state space model according to the operating conditions of the Buck converter in CCM (continuous inductor mode);

[0006] S200. Construct a hyperlocal model of the first-order system of the Buck converter, and establish hyperlocal models of the inner and outer loops of the three-phase interleaved parallel Buck converter in combination with the state space model; the hyperlocal model includes a current model of the inner loop and a voltage model of the outer loop. The state of the current model of the inner loop is as shown in formula (1), and the state of the voltage model of the outer loop is as shown in formula (2) below:

[0007]

[0008] Wherein,

[0009]

[0010] Wherein, In formulas (1) and (2): b 0m (m = 1, 2, 3) and b1 are respectively the weights of the input variables in the current and voltage loops; fLm (m = 1, 2, 3) is the total current disturbance and f uo is the total voltage disturbance; f o1m (m = 1, 2, 3) represents other uncertain disturbances, f o2 represents the unmodeled disturbance; L m (m = 1, 2, 3) are the inductances corresponding to the three-phase connections, i.e., L m represents the inductance value of the m-th branch; R m (m = 1, 2, 3) are the equivalent resistances corresponding to the three-phase connections, i.e., R m represents the resistance value of the m-th branch; i Lm (m = 1, 2, 3) are the inductor currents corresponding to the three-phase connections, i.e., i Lm represents the inductor current of the m-th branch;

[0011] C o represents the output capacitance value; R o represents the load resistance value; U o represents the output voltage; u m (t)

[0012] (m = 1, 2, 3) represents the switch state; u in represents the input voltage in the circuit; i L is the current flowing to the load after the superposition of the three-phase inductor currents;

[0013] S300. Construct a model-free predictive control system to obtain the outer-loop model-free predictive output signal and the inner-loop model-free predictive output signal;

[0014] S400. Design a nonlinear extended state observer based on the outer-loop model-free predictive output signal and the inner-loop model-free predictive output signal. Use the nonlinear extended state observer to observe and estimate the observed value of the total disturbance within a sampling period, and compensate the observed value of the total current disturbance and the observed value of the total voltage disturbance to the prediction model, thereby updating the reference values of the inductor current and duty ratio of the prediction model:

[0015] where, i Lm refers to the inductor current of each phase, m = 1, 2, 3, represents the output voltage reference value, u o (k + 1) represents the predicted output voltage value at the next moment, represents the observed value of the voltage disturbance, T s represents the sampling period, represents the reference value of the total current obtained by the superposition of the three currents, D m(k) Duty ratio of the model - free prediction system, i Lm (k + 1) represents the predicted inductor current value at the next moment, i Lm (k) represents the current inductor current value represents the observed value of current disturbance

[0016] As an achievable and preferred method, step S100 specifically includes:

[0017] S101. Establish a three - phase interleaved parallel Buck converter with 6 power switches. Considering that the power switches work in an ideal state and ignoring the influence of parasitic parameters of the power switches, for the power switch Sm (m = 1, 2, 3, 4, 5, 6), a continuous - time model is obtained:

[0018]

[0019] Where: L m , R m and i Lm (m = 1, 2, 3) are the inductors, equivalent resistors, and inductor currents corresponding to the three phases respectively; C o , R o and U o are the output capacitor, load resistor, and output voltage respectively; u m (t) (m = 1, 2, 3) represents the switch state. S1, S3, S5 are the main control power switches, and S2, S4, S6 are the switches whose trigger signals are complementary to the corresponding main control power switches. The switches S2, S4, S6 play a free - wheeling role and are used to control the three - phase main control switches respectively. The switch state of S1 is u1(t), the switch state of S3 is u2(t), and the switch state of S5 is u3(t). The switch state "1" indicates that the switch is on, and the switch state "0" indicates that the switch is off;

[0020] S102. Considering that if the inductor parameters do not match the model, the predictive control will be inaccurate, the conditions for the interleaved Buck converter to operate in CCM (continuous conduction mode of the inductor) are adopted to obtain the corresponding state - space model:

[0021]

[0022] In the formula: x is the state - variable vector, x = [i L1 i L2 i L3 u O T ; u is the control - input vector, u = [u1(t) u2(t) u3(t)] T ; i Lm (m = 1, 2, 3) represents the inductor current in the m - th branch​ Represents the derivative of the state vector x with respect to time;

[0023]

[0024] In step S200, a super-local model of the first-order system of the Buck converter is constructed, and a super-local model of the inner and outer loops of the three-phase interleaved parallel Buck converter is established in combination with the state-space model, specifically including:

[0025] S201. Construct a super-local model of the first-order system of the Buck converter:

[0026]

[0027] Where: y and y * Are the system output variable and its corresponding reference variable respectively; Represents the derivative of the state variable y with respect to time, Represents the derivative of the reference value y of the state variable * With respect to time, u represents the control input variable, that is, the state of the switch, b represents the weight coefficient corresponding to the input variable; F represents the total disturbance value of the system, Represents the estimated value of the total disturbance of the system; ζ(e) is the error feedback, e = y - y * , implemented through a PI controller;

[0028] S202. Substitute the formula ζ(e) = K p ·e(t) + K i ·∫e(t)dt into formula (6) to optimize the super-local model, and u is transformed into the following formula:

[0029] In the formula: k p And k i Represent the proportional gain and integral gain respectively, and u is used to represent the inner and outer loop control output signals. Represents the derivative of the state variable with respect to time, such as And

[0030] As a preferable implementable way, S300. Construct a model-free predictive control system to obtain the outer-loop model-free predictive output signal and the inner-loop model-free predictive output signal; specifically including:

[0031] S301. Based on the forward Euler difference method, rewrite the voltage model formula (2) of the outer loop into the outer-loop model-free predictive output signal: u o (k + 1) = b1T s ·i L (k) + T s ·f uo + uo (k)(8);

[0032] S302. Based on the forward Euler difference method, rewrite the current model formula (1) of the inner loop as: rewrite it as the model-free prediction output signal of the inner loop: i Lm (k + 1) = b0T s ·D m (k) + T s ·f Lm + i Lm (k)(9); where, i Lm (k + 1) represents the predicted value of the inductor current at the next moment, and i Lm (k) represents the current value of the inductor current at present.

[0033] As an implementable preferred manner, S401. Design a voltage control cost function, which is formula (10): According to formula (9) and formula (10), it can be obtained that:

[0034] Wherein: represents the output voltage reference value, u o (k + 1) represents the predicted value of the output voltage at the next moment, g o represents the voltage control cost function of the designed voltage loop, g im represents the current cost function of the designed current loop;

[0035] S402. Design an inner loop inductor current cost function, which is formula (12): (11), and obtain the duty ratio of the three-phase interleaved parallel Buck converter according to formula (1) and formula (12):

[0036] Wherein: i * L is the current controlled by the voltage outer loop output; i Lm is set to to achieve current sharing; D m (k) is the inner loop output signal, that is, the duty ratio, D m (k + 1) represents D of the next cycle m (k);

[0037] S403. After compensating the observed value of the total current disturbance and the observed value of the total voltage disturbance into the prediction model, it can be obtained that:

[0038] As an implementable preferred manner, it further includes step S500, and step S500 specifically includes:

[0039] S501. To estimate the total disturbance accuracy, an additional disturbance term γ1e is added to the state variables in the super-local models of the inner and outer loops of the three-phase interleaved parallel Buck converter 1m , and the change rate of the observed value of the disturbance term is estimated, and we can get:

[0040]

[0041]

[0042] In the formula: is the observed value of the inductor current; is the observed value of the output voltage; is the observed value of the total disturbance of the inductor current; is the observed value of the total disturbance of the output voltage; e 1m is the error between the observed value and the actual value of the inductor current, e2 is the error between the observed value and the actual value of the output voltage, α 1m and α2 are non-linear coefficients; δ1 and δ2 are filtering factors; γ1, γ2, γ3 and γ4 are the observer coefficients to be adjusted;

[0043] The non-linear feedback fal function is as follows:

[0044]

[0045] Where: if α is set to 1, then fal(e, α, δ) = e, and the non-linear extended state observer is linearized;

[0046] S502. Discretize formulas (14) and (15) to obtain the following formulas:

[0047]

[0048] As an implementable preferred method, to make the observer stable, according to the Lyapunov function for stability analysis, the observer coefficients need to meet the following requirements: the observer coefficients

[0050] Because e 1m is considered a bounded dynamic value, γ 1m greater than or equal to 0 is a limit value, and non-negative numbers are required, represents the derivative of e 1m ,

[0051] And f LmIts change is affected by the observed value and will only change within the difference between the observed value and the actual value. Therefore, The following formula can be deduced:

[0052]

[0053] Similarly, it can be obtained that:

[0054] The beneficial effects of the present invention are:

[0055] By constructing a super-local model of the converter in this application, the control system reduces its dependence on an accurate mathematical model. In the hydrogen production system, a non-linear extended state observer is used to monitor the total disturbance in the voltage and current control loops and perform real-time compensation. At the same time, using the deadbeat predictive control technology, the system can predict future reference vectors and calculate the DC voltage value based on the current and target current values. Thus, it can learn from the actual operation data in real time, continuously adjust and optimize the parameters of the hydrogen production process, which helps to improve the stability and reliability of hydrogen production, achieves fast dynamic response and improves the accuracy of predictive control, enabling the electrolyzer to produce a certain amount of hydrogen at a stable rate and flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 FIG. is a system block diagram for the application of an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter;

[0057] Figure 2 FIG. is a schematic diagram of output voltage and current of an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter;

[0058] Figure 3 FIG. is a schematic diagram of output voltage and current of another embodiment of an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter;

[0059] Figure 4 FIG. is a schematic diagram of output voltage and current of another embodiment of an intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter;

[0060] Through the above-mentioned drawings, specific embodiments of the present disclosure have been shown, and there will be more detailed descriptions hereinafter. These drawings and textual descriptions are not intended to limit the scope of the concept of the present disclosure in any way, but to illustrate the concept of the present disclosure to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0061] The technical solution of the application will be further described in detail below with reference to the drawings.

[0062] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0063] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs. The terms used in the description of the present invention herein are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0064] The following will Figures 1-4 be described in detail with reference to the drawings for some embodiments of the present invention. Without conflict, the following embodiments and the features in the embodiments may be combined with each other. For the same or similar concepts or processes, they may not be repeated in some embodiments.

[0065] As shown in the figure, it is a block diagram of an intelligent prediction control system based on an adaptive three-phase interleaved hydrogen production converter. It should be noted that BUCK (full English name: buck circuit, Chinese name: step-down circuit), and the BUCK converter is a step-down circuit converter. The principle of the three-phase interleaved parallel Buck converter is based on multi-phase operation to reduce the ripple of the output current and improve the overall efficiency. The core of the three-phase interleaved parallel Buck converter is to operate multiple Buck converter units in parallel, but the working phases of each unit are staggered from each other, that is, "interleaved", so as to achieve the effects of ripple cancellation and current sharing. S1, S3, and S5 are the main control power tubes, that is, the switching tubes that actually play the role of step-down; S2, S4, and S6 play the role of freewheeling, and their trigger signals are the complementary signals of the main control tubes on the corresponding branches. It can be seen from the figure that the working state of each converter unit can be adjusted by a model-free control strategy to ensure current sharing and phase control between them, thereby optimizing the overall performance, without relying on the accurate mathematical model of the system, but adjusting the control parameters according to the real-time feedback of the system. The three-phase interleaved parallel Buck converter of the present application can meet the requirements of low ripple and small size while providing a large output current through multi-phase operation and model-free control strategy, and is very suitable for modern electronic devices requiring high-performance power management. The intelligent prediction control method based on the adaptive three-phase hydrogen production converter proposed in the present application includes the following steps:

[0066] S100. Establish the circuit mathematical model of a three-phase interleaved parallel Buck converter, and establish the corresponding state-space model according to the operating conditions of the Buck converter in CCM (continuous inductor mode); specifically including:

[0067] S101. The three-phase interleaved parallel Buck converter includes six power switches. Considering that the power switches operate in an ideal state and ignoring the influence of the parasitic parameters of the power switches, the power switch Sm (m = 1, 2, 3, 4, 5, 6) obtains a continuous-time model:

[0068]

[0069] Where: L m , R m and i Lm (m = 1, 2, 3) are the inductors, equivalent resistors, and inductor currents corresponding to the three phases respectively; C o , R o and U o are the output capacitor, load resistor, and output voltage respectively; u m (t) (m = 1, 2, 3) represents the switch state. S1, S3, and S5 are the main control power switches, and S2, S4, and S6 are the switches whose trigger signals are complementary to the corresponding main control power switches. The switches S2, S4, and S6 play a freewheeling role and are used to control the three-phase main control switches respectively. The switch state of S1 is u1(t), the switch state of S3 is u2(t), and the switch state of S5 is u3(t). The switch state "1" indicates that the switch is on, and the switch state "0" indicates that the switch is off;

[0070] S102. Considering that if the inductor parameters do not match the model, the predictive control will be inaccurate, the operating conditions of the interleaved Buck converter in CCM (continuous inductor mode) are adopted to obtain the corresponding state-space model:

[0071]

[0072] In the formula: x is the state variable vector, x = [i L1 i L2 i L3 u O T ; u is the control input vector, u = [u1(t) u2(t) u3(t)] T ; i Lm (m = 1, 2, 3) represents the inductor current on the m-th branch; represents the derivative of the state vector x with respect to time;

[0073] ​

[0074] S200. Construct the super-local model of the first-order system of the Buck converter, and establish the super-local models of the inner and outer loops of the three-phase interleaved parallel Buck converter in combination with the state-space model; the super-local model includes the current model of the inner loop and the voltage model of the outer loop. The state of the current model of the inner loop is as shown in formula (1), and the state of the voltage model of the outer loop is as shown in formula (2) below:

[0075]

[0076] Among them,

[0077]

[0078] Among them, In formulas (1) and (2): b 0m (m = 1, 2, 3) and b1 are respectively the weights of the input variables in the current and voltage loops; f Lm (m = 1, 2, 3) is the total current disturbance and f uo is the total voltage disturbance; f o1m (m = 1, 2, 3) represents other uncertain disturbances, and f o2 represents the unmodeled disturbance; L m (m = 1, 2, 3) are the inductors connected to the three phases correspondingly, that is, L m represents the inductance value of the m-th branch; R m (m = 1, 2, 3) are the equivalent resistors connected to the three phases correspondingly, that is, R m represents the resistance value of the m-th branch; i Lm (m = 1, 2, 3) are the inductor currents connected to the three phases correspondingly, that is, i Lm represents the inductor current of the m-th branch;

[0079] C o represents the output capacitance value; R o represents the load resistance value; U o represents the output voltage; u m (t)(m = 1, 2, 3) represents the switch state; u in represents the input voltage in the circuit; i L is the current flowing from the superposition of the three-phase inductor currents to the load;

[0080] Step S200 specifically includes: S201. Construct the super-local model of the first-order system of the Buck converter:

[0081]

[0082] Among them: y and y * are respectively the system output variable and its corresponding reference variable; Denotes the derivative of the state variable y with respect to time, Denotes the reference value y of the state variable * The derivative with respect to time, u represents the control input variable, i.e., the state of the switch, b represents the weight coefficient corresponding to the input variable; F represents the total disturbance value of the system, Denotes the estimated value of the total disturbance of the system; ζ(e) is the error feedback, e = y - y * , which is implemented through a PI controller;

[0083] S201. Substitute the formula ζ(e) = K p ·e(t) + K i ·∫e(t)dt into Formula 6 to optimize the super-local model, and u is transformed into the following formula:

[0084] In the formula: k p and k i respectively represent the proportional gain and the integral gain, and u is used to represent the inner and outer loop control output signals.

[0085] S300. Construct a model-free predictive control system to obtain the outer loop model-free predictive output signal and the inner loop model-free predictive output signal; specifically including:

[0086] S301. Based on the forward Euler difference method, rewrite the voltage model formula (2) of the outer loop as the outer loop model-free predictive output signal: u o (k + 1) = b1T s ·i L (k) + T s ·f uo + u o (k)(8);

[0087] S302. Based on the forward Euler difference method, rewrite the current model formula (1) of the inner loop as:

[0088] Rewrite it as the inner loop model-free predictive output signal: i Lm (k + 1) = b0T s ·D m (k) + T s ·f Lm + i Lm (k)(9); Combine Formula 8 and Formula 9, and i L in this formula is because the output value of the voltage loop is used as the reference value of the current loop.

[0089] S400. Design a nonlinear extended state observer based on the outer-loop model-free prediction output signal and the inner-loop model-free prediction output signal. Use the nonlinear extended state observer to observe and estimate the observed value of the total disturbance within one sampling period, and compensate the observed value of the total current disturbance and the observed value of the total voltage disturbance to the prediction model, so as to update and obtain the reference values of the inductor current and duty cycle of the prediction model:

[0090] where, i Lm refers to the inductor current of each phase, m = 1, 2, 3, represents the reference value of the output voltage,

[0091] u o (k + 1) represents the predicted output voltage value at the next moment, represents the observed value of the voltage disturbance, T s represents the sampling period, represents the reference value of the total current which is the superposition of three currents, D m (k) is the duty cycle of the model-free prediction system, i Lm (k + 1) represents the predicted inductor current value at the next moment, i Lm (k) represents the current inductor current value at present, represents the observed value of the current disturbance. It should be noted that

[0092] Step S400 specifically includes: S401. Design a voltage control cost function, which is Formula (9):

[0093] According to Formula (9) and Formula (10), it can be obtained that:

[0094] represents the reference value of the output voltage, u o (k + 1) represents the predicted output voltage value at the next moment, g o represents the designed voltage control cost function of the voltage loop, g im represents the current cost function of the designed current loop;

[0095] Thus, calculate u at the next moment o through u at the previous moment o (k + 1), and make by calculating the voltage control cost function Formula (9), so as to obtain Formula (11).

[0096] S402. Design an inner-loop inductor current cost function, which is Formula (12): (12) Obtain the duty cycle of the three-phase interleaved parallel Buck converter according to Formula (1) and Formula (12):

[0097] Where: i * L is the current controlled by the outer voltage loop; i Lm is set to to achieve current sharing; D m (k) is the output signal of the inner loop, that is, the duty cycle, D m (k + 1) represents D of the next cycle m (k);

[0098] In this step, according to the inductor current cost function Formula (11), make the current cost function Combine Formula 1 to obtain Formula (13);

[0099] S403. After compensating the observed value of the total current disturbance and the observed value of the total voltage disturbance into the prediction model, we can get:

[0100] Preferably, this application also has step S500, which specifically includes:

[0101] S501. In order to estimate the total disturbance accuracy, an additional disturbance term γ1e 1m is added to the state variables in the super-local models of the inner and outer loops of the three-phase interleaved parallel Buck converter, and the change rate of the observed value of the disturbance term is estimated, and we can get:

[0102]

[0103] In the formula: is the observed value of the inductor current; is the observed value of the output voltage; is the observed value of the total inductor current disturbance; is the observed value of the total output voltage disturbance; e 1m is the error between the observed value and the actual value of the inductor current, e2 is the error between the observed value and the actual value of the output voltage, α 1m and α2 are non-linear coefficients; δ1 and δ2 are filtering factors; γ1, γ2, γ3 and γ4 are the observer coefficients to be adjusted;

[0104] The non-linear feedback fal function is as follows:

[0105]

[0106] Wherein: if α is set to 1, then fal(e, α, δ) = e, and the non-linear extended state observer is linearized;

[0107] S502. Discretize Formulas (14) and (15) to obtain the following formulas:

[0108]

[0109] To make the observer stable, according to the Lyapunov function perform stability analysis, and the observer coefficients need to meet the following requirements: the observer coefficients

[0110]

[0111] Substitute into the calculation in Formula 16. Since e 1m is considered a bounded dynamic value, γ 1m greater than or equal to 0 is a limit value, and non-negative numbers are required. represents the derivative of e 1m while the change of f is affected by the observed value and will only change within the difference between the observed value and the actual value. Therefore Lm the following formula can be deduced: Thus, the following formula is deduced:

[0112]

[0113] Similarly, it can be obtained that:

[0114] As Figure 2 shown, the waveform diagram experimental parameters are: input voltage 700V, inductance 10mH, output side capacitance 2000μF, load resistance 12.25Ω. Figure 2 It means that when the switch completes a full operation in 0.01 seconds, under the disturbance, the reference voltage drops from 350V to 150V, and the output voltage and current can remain stable; Figure 3 It means that when the switch completes a full operation in 0.01 seconds, under the disturbance, the reference voltage rises from 150V to 350V, and the output voltage and current can remain stable; Figure 4 It means that when the switch completes a full operation in 0.0001 seconds, under the disturbance, the reference voltage drops from 35V0 to 150V, and the output voltage and current can remain stable.

[0115] Therefore, Figure 2 under the dynamic condition that the reference voltage decreases within one switching period in 0.01 seconds, a stable output current can be maintained to supply the load; Figure 3 when the voltage recovers within one switching period in 0.01 seconds, it can quickly return to the normal working state under short-term fluctuations; Figure 4Under the dynamic condition of reducing the reference voltage within 0.0001 seconds to complete a switching cycle, it is possible to maintain a stable output current to supply the load. This proves that in the presence of environmental factors in the natural hydrogen production condition, the working condition changes, and through this control under interference conditions, the stability of the system can be maintained, the influence of external factors such as temperature change on the converter during actual operation is reduced, fast dynamic response is achieved, and the accuracy of predictive control is improved, enabling the electrolytic cell to produce a certain amount of hydrogen at a stable rate and flow rate.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

[0117] The above are only some embodiments of the present invention. For those of ordinary skill in the art, without departing from the inventive concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. An intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter, characterized in that It includes the following steps: S100. Establish the circuit mathematical model of the three-phase interleaved parallel Buck converter, and establish the corresponding state-space model according to the operating conditions of the Buck converter in CCM (continuous inductor mode); S200. Construct the super-local model of the first-order system of the Buck converter, and establish the super-local models of the inner and outer loops of the three-phase interleaved parallel Buck converter in combination with the state-space model; the super-local model includes the current model of the inner loop and the voltage model of the outer loop. The state of the current model of the inner loop is as shown in formula (1), and the state of the voltage model of the outer loop is as shown in formula (2) below: Among them, Among them, In Formulas (1) and (2): b 0m (m = 1, 2, 3) and b1 are the weights of the input variables in the current and voltage loops respectively; f Lm (m = 1, 2, 3) is the total current disturbance and f uo is the total voltage disturbance; f o1m (m = 1, 2, 3) represents other uncertain interferences, f o2 represents the unmodeled interference; L m (m = 1, 2, 3) are the inductances corresponding to the three-phase connections, i.e., L m represents the inductance value of the m-th branch; R m (m = 1, 2, 3) are the equivalent resistances corresponding to the three-phase connections, i.e., R m represents the resistance value of the m-th branch; i Lm (m = 1, 2, 3) are the inductance currents corresponding to the three phases respectively, that is, i Lm represents the inductance current on the m-th branch; C o represents the output capacitance value; R o represents the load resistance value; U o represents the output voltage; u m (t) (m = 1, 2, 3) represents the switch state; u in represents the input voltage in the circuit; i L is the current flowing from the superposition of the three-phase inductor currents to the load; S300. Construct a model-free predictive control system to obtain the model-free predictive output signal of the outer loop and the model-free predictive output signal of the inner loop; S400. Design a nonlinear extended state observer based on the outer-loop model-free prediction output signal and the inner-loop model-free prediction output signal. Use the nonlinear extended state observer to observe and estimate the observed value of the total disturbance within one sampling period, and use the observed value of the total current disturbance (m = 1, 2, 3) and the observed value of the total voltage disturbance Compensate them into the prediction model, so as to update the reference values of the inductor current and the duty cycle of the prediction model: where i Lm refers to the inductor current of each phase, m = 1, 2, 3, represents the output voltage reference value, u o (k + 1) represents the predicted output voltage value at the next moment, represents the observed value of voltage disturbance, T s represents the sampling period represents the reference value of the total current which is the superposition of three currents D m (k) duty cycle of the model-free prediction system, i Lm (k + 1) represents the predicted inductor current value at the next moment, i Lm (k) represents the current inductor current value represents the current disturbance observation value 2. The intelligent prediction control method based on an adaptive three-phase interleaved hydrogen production converter according to claim 1, wherein Step S100 specifically includes: S101. The three-phase interleaved parallel Buck converter includes 6 power switches. Considering that the power switches work under ideal conditions and ignoring the influence of the parasitic parameters of the power switches, the power switch Sm (m = 1, 2, 3, 4, 5, 6,) obtains a continuous-time model: Where: L m , R m and i Lm (m = 1, 2, 3) are the inductance, equivalent resistance, and inductor current corresponding to the three phases respectively; C o , R o and U o are the output capacitor, load resistance, and output voltage respectively; u m (t) (m = 1, 2, 3) represents the switch state. S1, S3, S5 are the main control power transistors, and S2, S4, S6 are the switching transistors complementary to the trigger signals of the corresponding main control power transistors respectively. The switching transistors S2, S4, S6 play a freewheeling role and are used to control the three-phase main control transistors respectively. The switch state of S1 is u1(t), the switch state of S3 is u2(t), and the switch state of S5 is u3(t). The switch state "1" indicates that the switch is on, and the switch state "0" indicates that the switch is off; S102. Considering that if the inductor parameters do not match the model, the predictive control will be inaccurate, then use the operating conditions of the interleaved Buck converter in CCM (continuous inductor mode) to obtain the corresponding state-space model: where: x is the state variable vector, x = [i L1 i L2 i L3 u O T ; u is the control input vector,​ u = [u1(t) u2(t) u3(t)] T ; i Lm (m = 1, 2, 3) represents the inductor current in the m-th branch; represents the derivative of the state vector x with respect to time; 3. The intelligent prediction control method based on an adaptive three-phase interleaved hydrogen production converter according to claim 2, wherein In step S200, construct the super-local model of the first-order system of the Buck converter, and establish the super-local models of the inner and outer loops of the three-phase interleaved parallel Buck converter in combination with the state-space model, specifically including: S201. Construct the super-local model of the first-order system of the Buck converter: where: y and y * are the system output variable and its corresponding reference variable respectively; represents the derivative of the state variable y with respect to time, represents the derivative of the reference value y of the state variable * with respect to time, u represents the control input variable, i.e., the state of the switch, b represents the weight coefficient corresponding to the input variable; F represents the total disturbance value of the system, represents the estimated value of the total disturbance of the system; ζ(e) is the error feedback, e = y - y * , and is achieved through a PI controller; S202. Substitute the formula ζ(e) = K p ·e(t) + K i ·∫e(t)dt into Formula 6 to optimize the hyperlocal model, and u is transformed into the following formula: Where: k p and k i respectively represent the proportional gain and the integral gain, and u is used to represent the inner and outer loop control output signal.

4. The intelligent prediction control method based on an adaptive three-phase interleaved hydrogen production converter according to claim 3, wherein S300. Construct a model-free predictive control system to obtain the model-free predictive output signal of the outer loop and the model-free predictive output signal of the inner loop; specifically including: S301. Based on the forward Euler difference method, rewrite the voltage model formula (2) of the outer loop into the outer loop model-free prediction output signal: u o (k + 1) = b1T s ·i L (k) + T s ·f uo + u o (k)(8); S302. Based on the forward Euler difference method, rewrite the current model formula (1) of the inner loop as: rewrite it as the model-free prediction output signal of the inner loop: i Lm (k + 1) = b0T s ·D m (k) + T s ·f Lm + i Lm (k)(9); where, i Lm (k + 1) represents the predicted inductor current value at the next moment, and i Lm (k) represents the current inductor current value.

5. The intelligent prediction control method based on an adaptive three-phase interleaved hydrogen production converter according to claim 4, wherein Step S400 specifically includes: S401. Design a voltage control cost function, which is formula (10): According to formula (9) and formula (10), it can be obtained that: Wherein: represents the output voltage reference value, u o (k + 1) represents the predicted output voltage value at the next moment, g o represents the voltage control cost function of the designed voltage loop, g im represents the current cost function of the designed current loop; S402. Design the inductor current cost function of the current loop, which is Formula (12): Obtain the duty cycle of the three-phase interleaved parallel Buck converter according to Formula (1) and Formula (12): where: i * L is the current output by the voltage outer loop control; i Lm is set to to achieve current sharing; D m (k) is the inner loop output signal, i.e., the duty cycle, D m (k + 1) represents the D of the next cycle m (k); S403. The observed value of the total current disturbance (m = 1, 2, 3) and the observed value of the total voltage disturbance After being compensated into the prediction model, the following can be obtained:

6. The intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter according to claim 5, characterized in that, The method further includes: S501. To estimate the total disturbance accuracy, an additional disturbance term γ1e is added to the state variables in the super-local models of the inner and outer loops of the three-phase interleaved parallel Buck converter 1m , and the change rate of the observed value of the disturbance term is estimated, and we can get: Where: is the observed value of the inductor current; is the observed value of the output voltage; is the observed value of the total disturbance of the inductor current; is the observed value of the total disturbance of the output voltage; e1m is the error between the observed value and the actual value of the inductor current, e2 is the error between the observed value and the actual value of the output voltage, α 1m and α2 are non-linear coefficients; δ1 and δ2 are filtering factors; γ1, γ2, γ3 and γ4 are the observer coefficients to be adjusted; The nonlinear feedback fal function is as follows: Where: if α is set to 1, then fal(e, α, δ) = e, and the nonlinear extended state observer is linearized; S502. Discretize formulas (14) and (15) to obtain the following formulas:

7. The intelligent predictive control method based on an adaptive three-phase interleaved hydrogen production converter according to claim 6, wherein To stabilize the observer, based on the Lyapunov function for stability analysis, the observer coefficients need to meet the following requirements: the observer coefficients Since e 1m is considered a bounded dynamic value, γ 1m greater than or equal to 0 is a limit value and is required to be non - negative, denotes the derivative of e 1m and while the change of f Lm is affected by the observed value and will only change within the difference between the observed value and the actual value, so Thus, the following formula is deduced: Similarly, it can be obtained that:

Images