Control method and system of improved consistency algorithm based on multiple agents

Abstract

The invention discloses a multi-agent-based control method and system for an improved consistency algorithm. The method comprises the following steps: building a coordinated voltage control model of a light-hydrogen-storage system consisting of a distributed photovoltaic system and a hydrogen energy storage system; in the photovoltaic power generation process, an inverter is responsible for stably transmitting reactive power to a power network, a certain adjusting capacity is reserved, and the voltage of a photovoltaic grid-connected point is adjusted and controlled by absorbing or releasing reactive electric energy; the hydrogen energy storage system monitors the voltage change condition by using the peak clipping and valley filling characteristics of the hydrogen energy storage system, plays a role in smoothing the influence brought by photovoltaic power generation fluctuation in a certain interval range, and optimizes an electrolytic cell; the principle of a consistency algorithm is analyzed, a typical finite time consistency algorithm is improved, and control gains c1 and c2 are introduced, so that the multi-agent system achieves average consistency within limited time; according to the invention, the control time of the photovoltaic inverter and the hydrogen energy storage system is reduced, and the reliability of a multi-agent coordinated voltage control strategy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed photovoltaic grid-connected voltage regulation, and in particular to a control method and system based on an improved consensus algorithm for multiple intelligent agents. Background Art

[0002] Compared with traditional energy, the biggest feature of hydrogen energy is that it is environmentally friendly and low-carbon. It is one of the most ideal raw materials and an energy storage material that can be used in related applications in the future. Energy storage is mainly used to smooth system power fluctuations, frequency and voltage regulation. It is used as a power source to supply power to loads when the distributed power generation is insufficient, and as an electrical appliance to absorb stored energy when the distributed power generation is excessive. Using HESS to participate in coordinated voltage regulation is an important technical approach. Ensuring that photovoltaic inverters and HESS coordinate and jointly maintain voltage stability is crucial for the safe and stable operation of the distribution network. Because the convergence speed of typical consistency algorithms is not enough to meet the needs of inverters and HESS to regulate voltage in the distribution network, voltage over-limit situations occur from time to time, threatening the safe operation of the distribution network. Therefore, the optimization problem of the control strategy of the light-hydrogen-storage system needs to be solved urgently. Summary of the invention

[0003] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a control method and system based on an improved consensus algorithm of multiple agents to solve the voltage over-limit problem caused by power flow reverse transmission caused by the current distributed photovoltaic access to the distribution network.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is: a control method based on an improved consensus algorithm of multiple agents, comprising the following steps:

[0005] Step 1: Determine the structure of the distributed photovoltaic and hydrogen energy storage system, and build a coordinated voltage control model of the light-hydrogen-storage system consisting of distributed photovoltaic and hydrogen energy storage systems;

[0006] Step 2: During the photovoltaic power generation process, the inverter is responsible for stably transmitting reactive power to the power grid and reserving a certain amount of regulation capacity to regulate the voltage at the photovoltaic grid connection point by absorbing or releasing reactive power;

[0007] Step 3: The hydrogen energy storage system uses its "peak shaving and valley filling" characteristics to monitor voltage changes, smooth out the impact of photovoltaic power generation fluctuations within a certain range, and optimize the electrolyzer to address the problem of frequent start and stop of existing PEM electrolyzers and proton exchange membrane fuel cell control units;

[0008] Step 4: Analyze the principle of the consensus algorithm. Aiming at the problem that the typical finite-time consensus algorithm does not converge fast enough, improve the typical finite-time consensus algorithm and introduce control gains c1 and c2 to make the multi-agent system reach average consensus within a limited time, where c1>0 and c2>0.

[0009] Step 5: Simulate the coordinated voltage control model of the constructed photovoltaic-hydrogen-storage system to reduce the control time of the photovoltaic inverter and the hydrogen energy storage system and improve the reliability of the multi-agent coordinated voltage control strategy.

[0010] Furthermore, the step 1 is specifically as follows:

[0011] (1) Establish a photovoltaic system model, the formula is expressed as:

[0012]

[0013] Among them, I is the output current of the photovoltaic cell, U is the output voltage of the photovoltaic cell, and I ph is the photocurrent, I0 is the reverse saturation current of the diode, q is the electron charge, R s is the series resistance, R sh is the parallel resistance, K is the Boltzmann constant, A is the PN junction ideality factor, and T is the absolute temperature;

[0014] Parameter I in the expression ph , R s , A, R sh and I0 are very sensitive to environmental changes and are difficult to set to an accurate value; considering R sh The value of R is relatively large and s The value of is small, and omitting it will not affect the final result, so ignore it. sc is the short-circuit current, let I ph =I sc , the simplified formula is:

[0015]

[0016] In the open circuit state, U oc is the open circuit voltage, satisfying I=0, U=U oc , Ump is the voltage at the maximum power point, and at the maximum power point, U=U mp , I mp is the current at the maximum power point, and at the maximum power point, I=I mp , B and C are intermediate coefficients, and their expressions are:

[0017]

[0018] Maximum power point tracking is one of the functions that photovoltaic systems must have; the core of the perturbation observation method is to regularly fine-tune the output voltage of the photovoltaic array. This adjustment action is called "perturbation"; by monitoring the change trend of the output power after the disturbance, it is judged: if the increase in output power ΔP>0, it means that the current voltage adjustment direction is correct, so this "disturbance" direction can be maintained; on the contrary, if ΔP<0, it means that the voltage adjustment direction is incorrect, and the "disturbance" direction should be changed at this time;

[0019] (2) Establish a hydrogen energy storage system model;

[0020] The voltage model consists of open circuit voltage, activation overvoltage and ohmic overvoltage, expressed as:

[0021] V=V ocv +V act +V ohm

[0022] Among them, V ocv is the open circuit voltage, which is also the minimum theoretical voltage for PEM water electrolysis; V act is the activation overvoltage; V ohm It is an ohmic overvoltage.

[0023] Furthermore, in step 2, the relationship between the adjustable reactive capacity of the inverter and the inverter capacity is:

[0024]

[0025] Among them, Q PV，max is the maximum adjustable reactive capacity of the photovoltaic inverter; S INV is the rated capacity of the photovoltaic inverter; P PV is the active power of photovoltaic power generation;

[0026] The inverter control process is expressed as:

[0027]

[0028] Among them, Q PV is the reactive power of photovoltaic power generation; V is the voltage at the photovoltaic grid connection point; V max 、V min are the upper and lower limits of the node voltage respectively; V tp is the starting voltage value of active power control; V tQ 、V aQ are the starting voltage values ​​for reactive power absorption and compensation respectively; P MPPT is the peak active power output of photovoltaic; Q max It is the maximum reactive power compensation value of the PV inverter.

[0029] Furthermore, the step 3 is specifically as follows:

[0030] In order to solve the problem of frequent start and stop of existing PEM electrolyzer and proton exchange membrane fuel cell control unit, the electrolyzer is optimized and the control strategy is improved;

[0031] In low-voltage distribution networks, HESS is installed to absorb or release active power to achieve the purpose of voltage regulation. Unlike inverter reactive power regulation, HESS active power regulation not only considers the capacity of the equipment, but also the charge state of the HESS. The relationship between the adjustable active power of the HESS device and the SOC is:

[0032] S(f)=S(f-Δf)+ΔS(f)

[0033]

[0034] Where S(f) is the SOC of HESS at time f; S(f-Δf) is the SOC of HESS at time f-Δf; ΔS(f) is the SOC change of HESS at time f; P HESS (f) is the active power of HESS at time f; C HESS is the capacity of HESS; Δf is the time interval.

[0035] Furthermore, the specific process of optimizing the electrolyzer and improving the control strategy is as follows:

[0036] (1) Input photovoltaic power generation and electrolyzer related data;

[0037] (2) GWO-LSTM predicts photovoltaic power generation;

[0038] (3) Taking the first-order difference of the prediction results and accumulating the results;

[0039] (4) Obtaining the minimum number of electrolytic cells in the future time period;

[0040] (5) Determine whether the rolling optimization cycle has been reached; if so, proceed to step (6); if not, proceed to step (7);

[0041] (6) Arranging the electrolytic cell positions according to the fitness function values;

[0042] (7) Allocate electrolytic cell power in real time according to operating conditions;

[0043] (8) Determine whether the maximum iteration cycle has been reached; if so, proceed to step (9); if not, proceed to step (1);

[0044] (9) Calculate the number of times the electrolytic cell is started and stopped.

[0045] Furthermore, the step 4 is specifically as follows:

[0046] (1) Improve the consistency algorithm:

[0047] Distributed control involves multiple agents collaborating through local information exchange, adjusting their states according to task requirements, and jointly performing complex tasks; agents are entities or software programs that can make autonomous decisions and interact with the environment; multi-agent systems are a collection of these agents;

[0048] The adjacency matrix A is an n-order matrix, A={a ij}, a ij represents the connection weight between node i and node j, expressed as:

[0049]

[0050] Among them, n i 、n j Respectively represent the number of adjacent nodes of node i and node j;

[0051] For MAS to achieve average consistency within a limited time, it must meet the following requirements:

[0052]

[0053] Among them, x e is the average value of the initial state of the agent;

[0054] A typical finite-time consensus algorithm is expressed as:

[0055]

[0056] Among them, ε is the control factor, 0<ε<1, sign is the sign function, which can be expressed as:

[0057]

[0058] In order to improve the convergence speed of the consistency algorithm, the typical finite time consistency algorithm is improved, and control gains c1 and c2 (c1>0, c2>0) are introduced to improve the consistency algorithm, which can be expressed as:

[0059]

[0060] sig ε (x)=|x ε sign(x)

[0061] Assuming L(A) is semi-positive definite and σ is a characteristic root of L(A), under the improved consensus algorithm, the agent states reach a consensus within time T; T can be expressed as:

[0062]

[0063] x i (f) = x i (f)-x′(f)

[0064]

[0065] Among them, x i (f) represents the information state of agent i at time f; is the state error; x′(f) is the average state trajectory of the system; for

[0066] (2) Multi-agent coordinated control strategy:

[0067] The state equation of the agent is expressed as:

[0068]

[0069] u ii (f) = K ii x i (f)

[0070] Among them, u ii (f) is the self-feedback input of the agent, A ii , B ii ; A ij is the system matrix, K ii is the state feedback gain matrix;

[0071] The agent and the coordinated controller can be represented as:

[0072] u ij (f) = K ij x j (f), i, j = 1, 2, ..., n and i ≠ j

[0073]

[0074] Among them, B ij is a known constant matrix; K ij is the gain matrix reflecting the state feedback; u ij (f) is the coordination controller, which represents the feedback input from agent j to agent i.

[0075] In addition, the present invention also discloses the control system based on the improved consensus algorithm of multi-agent, including:

[0076] Photovoltaic cell module, used to convert solar energy into electrical energy; MPPT control module, used to track the photovoltaic output power point, and achieve the maximum power output of the photovoltaic system by matching the photovoltaic system module; inverter module, used to convert DC into AC grid-connected, and maintain voltage stability through reactive power control; HESS, which maintains voltage stability through active power control, is used to convert excess electrical energy into hydrogen for storage, and provide support during peak load periods, improve the consistency algorithm, increase the consistency convergence speed, reduce the control time for photovoltaic inverters and HESS, and improve the reliability of multi-agent coordinated voltage control strategy.

[0077] Beneficial effects of the invention: The invention aims to reduce the control time for the photovoltaic inverter and HESS to reach a consensus and solve the voltage over-limit problem caused by the access of distributed photovoltaic to the distribution network, and proposes a control method and system based on a multi-agent improved consensus algorithm; a photovoltaic system model is built, and a hydrogen energy storage model is added to realize the observation of the effect of converting electrical energy into hydrogen energy; by studying the problem of frequent start and stop of PEM electrolyzer and proton exchange membrane fuel cell control unit, an optimization scheme for the electrolyzer is designed, and the control strategy is improved to balance the working performance of the electrolyzer, reduce the start and stop frequency and response time of the unit, enhance the tracking ability of HESS, smooth power fluctuations, and alleviate The burden of the grid-connected inverter is reduced, thereby improving the voltage quality; the method uses an improved consistency algorithm to control the photovoltaic inverter and HESS to reach consistency in a faster time, and coordinates to solve the problem of voltage over-limit in the distribution network; the present invention controls the light-hydrogen-storage system, and coordinates the inverter reactive power control and the HESS active power control to stabilize the distribution network voltage within a certain range; the present invention improves the consistency algorithm, the photovoltaic inverter uses the reactive power utilization rate as the consistency variable, the HESS uses the SOC change as the consistency variable, and uses the improved consistency algorithm for coordinated control, which solves the problem of insufficient coordinated control of the photovoltaic inverter and the HESS, and better suppresses the voltage over-limit. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 This is a flow chart of the perturbation observation method in the present invention;

[0079] Figure 2 is the photovoltaic inverter capacity curve in the present invention;

[0080] Figure 3 This is the inverter control principle diagram of the present invention;

[0081] Figure 4 It is the optimization control flow chart of the electrolytic cell in the present invention;

[0082] Figure 5 The voltage condition when no control means is adopted in the present invention;

[0083] Figure 6 Convergence comparison of the algorithm in the present invention;

[0084] Figure 7 This is to coordinate the voltage control situation in the present invention. DETAILED DESCRIPTION

[0085] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0086] Example 1: Figure 1-4 As shown, a coordinated voltage control method and system based on an improved consistency algorithm mainly includes the following main steps:

[0087] Step 1: Determine the structure of the distributed photovoltaic and hydrogen energy storage system, and build a coordinated voltage control model of the light-hydrogen-storage system consisting of distributed photovoltaic and hydrogen energy storage systems;

[0088] Step 2: During the photovoltaic power generation process, the inverter is responsible for the stable transmission of reactive power to the power grid and reserves a certain amount of regulation capacity to regulate the voltage of the photovoltaic grid-connected point by absorbing or releasing reactive power;

[0089] Step 3: The hydrogen energy storage system uses its "peak shaving and valley filling" characteristics to monitor voltage changes, smoothing the impact of photovoltaic power generation fluctuations within a certain range, and optimizing the electrolyzer to address the problem of frequent start and stop of existing PEM electrolyzers and proton exchange membrane fuel cell control units;

[0090] Step 4: Analyze the principle of the consensus algorithm. Aiming at the problem that the typical finite-time consensus algorithm does not converge fast enough, an improved typical finite-time consensus algorithm is proposed. The control gains c1 and c2 (c1>0, c2>0) are introduced to make the multi-agent system reach average consensus within a limited time.

[0091] Step 5: Simulate the coordinated voltage control model of the constructed solar-hydrogen-storage system, which effectively reduces the control time of the photovoltaic inverter and the hydrogen energy storage system and improves the reliability of the multi-agent coordinated voltage control strategy.

[0092] Specifically, in step 1, the structure of the distributed photovoltaic and hydrogen energy storage system is determined, and a coordinated voltage control model of the light-hydrogen-storage system consisting of the distributed photovoltaic and hydrogen energy storage systems is built.

[0093] (1) Establish a photovoltaic system model, the formula can be expressed as:

[0094]

[0095] Among them, I is the output current of the photovoltaic cell, U is the output voltage of the photovoltaic cell, and I phis the photocurrent, I0 is the reverse saturation current of the diode, q is the electron charge, R s is the series resistance, R sh is the parallel resistance, K is the Boltzmann constant, A is the PN junction ideality factor, and T is the absolute temperature;

[0096] Parameter I in the expression ph , R s , A, R sh and I0 are very sensitive to environmental changes and are difficult to set to an accurate value; considering R sh The value of R is relatively large and s The value of is small, and omitting it will not affect the final result, so ignore I sc is the short-circuit current, let I ph =I sc , the simplified formula is:

[0097]

[0098] In the open circuit state, U oc is the open circuit voltage, satisfying I=0, U=U oc , U mp is the voltage at the maximum power point, and at the maximum power point, U=U mp , I mp is the current at the maximum power point, and at the maximum power point, I=I mp , B and C are intermediate coefficients, and their expressions are:

[0099]

[0100] Maximum power point tracking (MPPT) is one of the essential functions of photovoltaic systems. As a common MPPT technology, the core of the perturbation observation method is to regularly fine-tune the output voltage of the photovoltaic array. This adjustment action is called "perturbation". It is judged by monitoring the change trend of the output power after the disturbance: if the increase in output power ΔP>0, it means that the current voltage adjustment direction is correct, so this "disturbance" direction can be maintained; on the contrary, if ΔP<0, it means that the voltage adjustment direction is incorrect, and the "disturbance" direction should be changed at this time. Figure 1 This is the flow chart of the perturbation and observation method.

[0101] (2) Establish a hydrogen energy storage system model.

[0102] The voltage model consists of open circuit voltage, activation overvoltage and ohmic overvoltage, which can be expressed as:

[0103] V=V ocv +V act +V ohm

[0104] Among them, V ocv is the open circuit voltage, which is also the minimum theoretical voltage for PEM water electrolysis; V act is the activation overvoltage; V ohm It is the ohmic overvoltage;

[0105] The open circuit voltage of the electrolytic cell is expressed by the Nernst equation:

[0106]

[0107] Where V0 is the standard electromotive force, z is the number of moles of electrons involved in the electrolysis reaction, and α i is the activity of substance i;

[0108] The activation overvoltage can be expressed as:

[0109]

[0110] Where α is the transfer coefficient, i is the current density, and i0 is the exchange current density;

[0111] The ohmic overvoltage can be expressed as:

[0112] V ohm =iR ohm

[0113] Among them, R ohm is the membrane resistance, which can be expressed as:

[0114]

[0115] Among them, σ m is the conductivity of the membrane;

[0116] The hydrogen storage model can be expressed as:

[0117]

[0118] Among them, P b is the pressure of the hydrogen tank, P bi is the initial pressure of the hydrogen tank, Z is the compression constant of hydrogen, M H2 is the molar mass of hydrogen, V b is the volume of the hydrogen tank, N H2 is the number of moles of hydrogen entering the hydrogen tank per second, T b is the temperature of the hydrogen tank.

[0119] Specifically, in step 2, during the photovoltaic power generation process, the inverter is responsible for stably transmitting reactive power to the power network, and reserves a certain regulation capacity to regulate the voltage of the photovoltaic grid-connected point by absorbing or releasing reactive power.

[0120] Figure 2 is the photovoltaic inverter capacity curve, and the relationship between the inverter adjustable reactive capacity and the inverter capacity is:

[0121]

[0122] Among them, Q PV，max is the maximum adjustable reactive capacity of the photovoltaic inverter; S INV is the rated capacity of the photovoltaic inverter; P PV is the active power of photovoltaic power generation.

[0123] Figure 3 This is the inverter control schematic diagram, which can be expressed as:

[0124]

[0125] Among them, Q PV is the reactive power of photovoltaic power generation; V is the voltage at the photovoltaic grid connection point; V max 、V min are the upper and lower limits of the node voltage respectively; V tp is the starting voltage value of active power control; V tQ 、V aQ are the starting voltage values ​​for reactive power absorption and compensation respectively; P MPPT is the peak active power output of photovoltaic; Q max It is the maximum reactive power compensation value of the PV inverter.

[0126] Specifically, in step 3, the hydrogen energy storage system utilizes its "peak shaving and valley filling" characteristics to monitor voltage changes, thereby smoothing the impact of photovoltaic power generation fluctuations within a certain range, and optimizing the electrolyzer in view of the problem of frequent start and stop of existing PEM electrolyzers and proton exchange membrane fuel cell control units.

[0127] In order to solve the problem of frequent start and stop of existing PEM electrolyzer and proton exchange membrane fuel cell control unit, the electrolyzer is optimized and the control strategy is improved. Figure 4For the optimization control flow chart of the electrolyzer, the long short-term memory (LSTM) network is used to predict the photovoltaic power generation, and the parameters of the LSTM are optimized and adjusted by the grey wolf optimizer (GWO). Then, the number of units required to operate at each time point is determined according to the prediction results of GWO-LSTM, and the unit position is optimized according to the fitness function. Subsequently, the power is distributed in real time according to the time series.

[0128] In low-voltage distribution networks, HESS is installed to absorb or release active power to achieve the purpose of voltage regulation. Different from inverter reactive power regulation, HESS active power regulation not only considers the capacity of the equipment but also the state of charge (SOC) of the HESS. The relationship between the adjustable active power of the HESS device and the SOC is:

[0129] S(f)=S(f-Δf)+ΔS(f)

[0130]

[0131] Where S(f) is the SOC of HESS at time f; S(f-Δf) is the SOC of HESS at time f-Δf; ΔS(f) is the SOC change of HESS at time f; P HESS (f) is the active power of HESS at time f; C HESS is the capacity of HESS; Δf is the time interval.

[0132] Specifically, in step 4, the principle of the consistency algorithm is analyzed, and to address the problem that the typical finite-time consistency algorithm does not converge fast enough, an improved typical finite-time consistency algorithm is proposed, and control gains c1 and c2 (c1>0, c2>0) are introduced to enable the multi-agent system to reach average consistency within a limited time.

[0133] (1) Improve the consistency algorithm

[0134] Distributed control involves multiple agents collaborating through local information exchange, adjusting their states according to task requirements, and jointly executing complex tasks. Agents are entities or software programs that can make autonomous decisions and interact with the environment. A multi-agent system (MAS) is a collection of these agents.

[0135] The adjacency matrix A is an n-order matrix, A={a ij}, a ij represents the connection weight between node i and node j, which can be expressed as:

[0136]

[0137] Among them, n i 、n j Represent the number of adjacent nodes of node i and node j respectively.

[0138] Laplace matrix L = {l ij} can be expressed as:

[0139]

[0140] In a distributed multi-agent control system with n agents, the state update of each agent depends on its own instantaneous state and the instantaneous state of its neighboring agents, which can be expressed as:

[0141] ∫x i (f)df=u i (f)

[0142] Among them, x i is the state variable; u i is the control variable.

[0143] The specific control can be expressed as:

[0144]

[0145] For MAS to achieve average consistency within a limited time, it must meet the following requirements:

[0146]

[0147] Among them, x e is the average value of the initial state of the agent.

[0148] A typical finite-time consensus algorithm can be expressed as:

[0149]

[0150] Among them, ε is the control factor, 0<ε<1, sign is the sign function, which can be expressed as:

[0151]

[0152] In order to improve the convergence speed of the consistency algorithm, this paper proposes an improved typical finite time consistency algorithm, introducing control gains c1 and c2 (c1>0, c2>0). The improved consistency algorithm can be expressed as:

[0153]

[0154] sig ε (x)=|x ε sign(x)

[0155] Assuming L(A) is semi-positive definite and σ is a characteristic root of L(A), under the improved consensus algorithm, the agent states reach a consensus within time T. T can be expressed as:

[0156]

[0157] x i (f) = x i (f)-x′(f)

[0158]

[0159] Among them, x i (f) represents the information state of agent i at time f; is the state error; x′(f) is the average state trajectory of the system; for for

[0160] (2) Multi-agent coordinated control strategy

[0161] The state equation of the agent can be expressed as:

[0162]

[0163] u ii (f) = K ii x i (f)

[0164] Among them, u ii (f) is the self-feedback input of the agent, A ii , B ii ; A ij is the system matrix, K ii is the state feedback gain matrix.

[0165] The agent and the coordinated controller can be represented as:

[0166] u ij (f) = K ij x j (f), i, j = 1, 2, ..., n and i ≠ j

[0167]

[0168] Among them, B ij is a known constant matrix; K ij is the gain matrix reflecting the state feedback; u ij (f) is the coordination controller, which represents the feedback input from agent j to agent i.

[0169] Specifically, in step 5, the constructed light-hydrogen-storage system model is simulated in the IEEE33 node system to reduce the control time of the photovoltaic inverter and HESS, and achieve the purpose of faster coordinated voltage control. This strategy is based on the reactive power control of the photovoltaic inverter and the active power control characteristics of the HESS. The photovoltaic inverter uses the reactive power utilization rate as the consistency variable, and the HESS uses the SOC change as the consistency variable. The coordinated control is carried out using an improved consistency algorithm, which solves the problem of insufficient coordinated control between the photovoltaic inverter and the HESS, and better suppresses the voltage over-limit. Hydrogen energy storage optimizes the electrolyzer, balances the working performance of the electrolyzer, reduces the start-stop frequency and response time of the unit, enhances the tracking capability of the HESS, smoothes power fluctuations, and reduces the burden on the grid-connected inverter, thereby improving the voltage quality.

[0170] Embodiment 2: A control method and system based on an improved consensus algorithm of multiple agents are disclosed, comprising:

[0171] Photovoltaic cell module, used to convert solar energy into electrical energy; MPPT control module, used to track the photovoltaic output power point, and by matching the photovoltaic system module, the photovoltaic system can reach the maximum power output; inverter module, converts DC into AC for grid connection; maintains voltage stability through reactive power control; HESS, maintains voltage stability through active power control, is used to convert excess electrical energy into hydrogen for storage, and provides support during peak load periods; improves the consistency algorithm, increases the consistency convergence speed, reduces the control time for photovoltaic inverters and HESS, and improves the reliability of multi-agent coordinated voltage control strategy.

[0172] Embodiment 3:

[0173] like Figure 5-7 As shown, the present invention takes a photovoltaic power station in a certain city throughout 2023 as the test object, and the inspection period is set from September to October 2023. The time period from 0:00 in the morning to 24:00 in the evening every day is selected. The sensor continuously monitors and records the light intensity and power load. These data are averaged every 30 minutes and stored in the database. The present invention takes the data of a certain day in October as the verification sample, selects the time point 14:00 when the photovoltaic output is the largest and the load is the smallest in the verification sample, and the time point 20:00 when the photovoltaic output is 0 and the load is the largest as the research point, and connects the distributed photovoltaic to the distribution network. Figure 5 This is the voltage condition when no control measures are adopted.

[0174] Comparing the convergence speed of the classic consistency algorithm and the improved consistency algorithm proposed in this invention, Figure 6 This is a comparison of the convergence speed of the algorithms, which shows that the improved consistency algorithm used in the present invention has a faster convergence property. Figure 5In data application and improved consistency algorithm, the coordinated voltage control situation is as follows Figure 7 As shown, after adopting the improved consistency algorithm, the voltage over-limit situation is well controlled, which meets the requirements of the present invention.

[0175] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A control method based on improved consensus algorithm of multi-agent, Features: The following steps are involved: Step 1: Determine the structure of the distributed photovoltaic and hydrogen energy storage system, and build a coordinated voltage control model of the light-hydrogen-storage system consisting of distributed photovoltaic and hydrogen energy storage systems; Step 2: During the photovoltaic power generation process, the inverter is responsible for stably transmitting reactive power to the power grid and reserving a certain amount of regulation capacity to regulate the voltage at the photovoltaic grid connection point by absorbing or releasing reactive power; Step 3: The hydrogen energy storage system uses its "peak shaving and valley filling" characteristics to monitor voltage changes, smooth out the impact of photovoltaic power generation fluctuations within a certain range, and optimize the electrolyzer to address the problem of frequent start and stop of existing PEM electrolyzers and proton exchange membrane fuel cell control units; Step 4: Analyze the principle of the consensus algorithm. Aiming at the problem that the typical finite-time consensus algorithm does not converge fast enough, improve the typical finite-time consensus algorithm and introduce control gains c1 and c2 to make the multi-agent system reach average consensus within a limited time, where c1>0 and c2>0. Step 5: Simulate the coordinated voltage control model of the constructed photovoltaic-hydrogen-storage system to reduce the control time of the photovoltaic inverter and the hydrogen energy storage system and improve the reliability of the multi-agent coordinated voltage control strategy.

2. According to claim 1, a control method based on an improved consensus algorithm of multiple agents, Features: The step 1 is specifically as follows: (1) Establish a photovoltaic system model, the formula is expressed as: Among them, I is the output current of the photovoltaic cell, U is the output voltage of the photovoltaic cell, and I ph is the photocurrent, I 0 is the reverse saturation current of the diode, q is the electron charge, R s is the series resistance, R sh is the parallel resistance, K is the Boltzmann constant, A is the PN junction ideality factor, and T is the absolute temperature; Parameter I in the expression ph , R s , A, R sh and I 0 It is very sensitive to environmental changes and difficult to set to an accurate value; considering R sh The value of R is relatively large and s The value of is small, and omitting it will not affect the final result, so ignore it. sc is the short-circuit current, let I ph =I sc , the simplified formula is: In the open circuit state, U oc is the open circuit voltage, satisfying I=0, U=U oc , Ump is the voltage at the maximum power point, and at the maximum power point, U=U mp , I mp is the current at the maximum power point, and at the maximum power point, I=I mp , B and C are intermediate coefficients, and their expressions are: Maximum power point tracking is one of the functions that photovoltaic systems must have; the core of the perturbation observation method is to regularly fine-tune the output voltage of the photovoltaic array. This adjustment action is called "perturbation". It is judged by monitoring the change trend of the output power after the disturbance: if the increase in output power ΔP>0, it means that the current voltage adjustment direction is correct, so this "disturbance" direction can be maintained; on the contrary, if ΔP<0, it means that the voltage adjustment direction is incorrect, and the "disturbance" direction should be changed at this time; (2) Establish a hydrogen energy storage system model; The voltage model consists of open circuit voltage, activation overvoltage and ohmic overvoltage, expressed as: V=V ocv +V act +V ohm Among them, V ocv is the open circuit voltage, which is also the minimum theoretical voltage for PEM water electrolysis; V act is the activation overvoltage; V ohm It is an ohmic overvoltage.

3. A control method based on an improved consensus algorithm of multiple agents according to claim 1, Features: In step 2, the relationship between the adjustable reactive capacity of the inverter and the inverter capacity is: Among them, Q PV，max is the maximum adjustable reactive capacity of the photovoltaic inverter; S INV is the rated capacity of the photovoltaic inverter; P PV is the active power of photovoltaic power generation; The inverter control process is expressed as: Among them, Q PV is the reactive power of photovoltaic power generation; V is the voltage at the photovoltaic grid connection point; V max 、V min are the upper and lower limits of the node voltage respectively; V tp is the starting voltage value of active power control; V tQ 、V aQ are the starting voltage values ​​for reactive power absorption and compensation respectively; P MPPT is the peak active power output of photovoltaic; Q max It is the maximum reactive power compensation value of the PV inverter.

4. A control method based on an improved consensus algorithm of multiple agents according to claim 1, Features: The step 3 is as follows: In order to solve the problem of frequent start and stop of existing PEM electrolyzer and proton exchange membrane fuel cell control unit, the electrolyzer is optimized and the control strategy is improved; In low-voltage distribution networks, HESS is installed to absorb or release active power to achieve the purpose of voltage regulation. Unlike inverter reactive power regulation, HESS active power regulation not only considers the capacity of the equipment, but also the charge state of the HESS. The relationship between the adjustable active power of the HESS device and the SOC is: S(f)=S(f-Δf)+ΔS(f) Where S(f) is the SOC of HESS at time f; S(f-Δf) is the SOC of HESS at time f-Δf; ΔS(f) is the SOC change of HESS at time f; P HESS (f) is the active power of HESS at time f; C HESS is the capacity of HESS; Δf is the time interval.

5. A control method based on an improved consensus algorithm of multiple agents according to claim 4, Features: The specific process of optimizing the electrolyzer and improving the control strategy is as follows: (1) Input photovoltaic power generation and electrolyzer related data; (2) GWO-LSTM predicts photovoltaic power generation; (3) Taking the first-order difference of the prediction results and accumulating the results; (4) Obtaining the minimum number of electrolytic cells in the future time period; (5) Determine whether the rolling optimization cycle has been reached; if so, proceed to step (6); if not, proceed to step (7); (6) Arranging the electrolytic cell positions according to the fitness function values; (7) Allocate electrolytic cell power in real time according to operating conditions; (8) Determine whether the maximum iteration cycle has been reached; if so, proceed to step (9); if not, proceed to step (1); (9) Calculate the number of times the electrolytic cell is started and stopped.

6. A control method based on an improved consensus algorithm of multiple agents according to claim 1, Features: The step 4 is specifically as follows: (1) Improve the consistency algorithm: Distributed control involves multiple agents collaborating through local information exchange, adjusting their states according to task requirements, and jointly performing complex tasks; agents are entities or software programs that can make autonomous decisions and interact with the environment; multi-agent systems are a collection of these agents; The adjacency matrix A is an n-order matrix, A={a ij }, a ij represents the connection weight between node i and node j, expressed as: Among them, n i 、n j Respectively represent the number of adjacent nodes of node i and node j; For MAS to achieve average consistency within a limited time, it must meet the following requirements: Among them, x e is the average value of the initial state of the agent; A typical finite-time consensus algorithm is expressed as: Among them, ε is the control factor, 0<ε<1, sign is the sign function, which can be expressed as: In order to improve the convergence speed of the consensus algorithm, the typical finite time consensus algorithm is improved and the control gain c is introduced. 1 、c 2 (c 1 >0,c 2 >0), the improved consistency algorithm is expressed as: sig ε (x)=|x| ε sign(x) Assuming L(A) is semi-positive definite and σ is a characteristic root of L(A), under the improved consensus algorithm, the agent states reach a consensus within time T; T can be expressed as: Among them, x i (f) represents the information state of agent i at time f; is the state error; x′(f) is the average state trajectory of the system; for for (2) Multi-agent coordinated control strategy: The state equation of the agent is expressed as: u ii (f)=K ii x i (f) Among them, u ii (f) is the self-feedback input of the agent, A ii , B ii ; A ij is the system matrix, K ii is the state feedback gain matrix; The agent and the coordinated controller can be represented as: u ij (f) = K ij x j (f), i, j = 1, 2, ..., n and i ≠ j Among them, B ij is a known constant matrix; K ij is the gain matrix reflecting the state feedback; u ij (f) is the coordination controller, which represents the feedback input from agent j to agent i.

7. A control system based on improved consensus algorithm of multi-agent, It is characterized in that include: Photovoltaic cell modules, which convert solar energy into electricity; The MPPT control module is used to track the photovoltaic output power point and achieve the maximum power output of the photovoltaic system by matching the photovoltaic system module. The inverter module is used to convert DC into AC for grid connection and maintain voltage stability through reactive power control. The HESS maintains voltage stability through active power control and is used to convert excess electrical energy into hydrogen for storage and provide support during peak load periods. It improves the consistency algorithm, increases the consistency convergence speed, reduces the control time for photovoltaic inverters and HESS, and improves the reliability of the multi-agent coordinated voltage control strategy.

Images