Economic efficiency and electric energy quality collaborative optimization intelligent micro-grid control method

Abstract

The invention relates to the technical field of micro-grid electric energy quality control and economical efficiency regulation and control scheduling optimization. The invention further discloses an economical efficiency and electric energy quality collaborative optimization intelligent micro-grid control method. According to the invention, a micro-grid overall control framework is established, based on the designed and obtained dynamic and static analysis and optimization control strategy and the optimal scheduling strategy of each unit, the micro-grid multi-source load storage overall environment is oriented, and high electric energy quality stability, system safety and overall economy are taken as targets. Low-cost power generation and quick response capability of the main control unit are used as driving, and the problems of micro-grid electric energy quality control and overall cooperative control design and implementation are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid power quality control and economic regulation and dispatch optimization, and specifically to a smart microgrid control method for collaboratively optimizing economy and power quality. Background Art

[0002] In response to the key issues exposed in the continuous advancement of the industry, technical analysis and breakthroughs are needed from the following perspectives:

[0003] Impact of high uncertainty source load on microgrid power quality stability

[0004] The power system requires that power generation and grid load demand reach a transient balance. However, with the increase in the proportion of grid-connected renewable energy power generation and demand-side resources such as wind and solar, their strong volatility and high uncertainty significantly increase the probability of power imbalance in the grid, seriously restricting the grid's ability to absorb renewable energy, resulting in frequent "wind and solar abandonment" phenomena. [9-12] At the same time, while pursuing improvements in microgrid absorption capacity, without sufficient feasible technical support, connecting a high proportion of renewable energy to the grid could lead to risks such as line overload, load loss, and static voltage instability, posing serious safety risks to the power system. Essentially, this raises requirements and challenges for the overall control architecture of microgrids and the renewable energy absorption model.

[0005] The economic efficiency of microgrid operation needs to be guaranteed in principle

[0006] Key components of microgrids include energy storage units and distributed generation units. The construction and power generation costs of these key components are crucial indicators that constrain the overall economic viability of microgrids. To meet the demands of industrial development, core bottleneck technologies such as efficient and reliable clean power generation technologies, energy scheduling strategies for optimizing operating costs, and coordinated optimization and management of microgrids must be researched in a multifaceted and integrated manner.

[0007] In view of the high uncertainty of renewable energy units, first of all, microgrids must introduce multiple types of efficient and low-cost non-intermittent clean energy as high-quality controllable source units, and meet the needs of different load characteristics to form a multi-source load storage microgrid that can operate efficiently and collaboratively.

[0008] After the scale of the microgrid and the main source-storage-load network topology are determined, the microgrid optimization and control process is the process of designing and implementing energy management optimization strategies, which mainly includes the following two steps:

[0009] 1) Based on the established integrated microgrid model of multiple units including source, load, and storage, the collaborative optimization goal is set for the overall real-time load demand in steady state and transient state, and the optimal energy scheduling configuration is optimized to obtain the reasonable load of each unit (source is regarded as negative load).

[0010] 2) For the design of the optimal scheduling strategy, it is necessary to perform power flow calculations and implement efficient, fast, and low-cost control while ensuring system stability.

[0011] Based on this, a consensus has been reached in the field: under the premise of controllable costs, using energy storage units as the main control units of microgrids and system balancing nodes can compensate for the redundant, uncontrollable and insufficient shortcomings of energy storage units in real time, and is an effective way to fundamentally improve the system's power quality and comprehensive regulation capabilities.

[0012] Optional types of energy storage units include: lithium battery energy storage, pumped water energy storage, flywheel energy storage, etc.; the present invention selects lithium batteries as the energy storage unit, which are the most widely used, have high energy density and stable performance. Summary of the Invention

[0013] The present invention provides a smart microgrid control method for collaboratively optimizing economy and power quality. A model is established to fully utilize the surplus capacity of wind and solar power generation and establish a hydrogen production process model. Forecasts are made for various types of loads to provide data support for the design of energy optimization scheduling strategies. A reasonable scheduling strategy is designed, and the scheduling strategy is optimized based on energy optimization scheduling and rapid power flow calculation based on the evaluation of dispatchable capacity. Driven by the low-cost power generation and rapid response capabilities of the main control unit, microgrid power quality control and overall collaborative control are designed and implemented, which solves the shortcomings of the background technology.

[0014] The present invention provides the following technical solution: a smart microgrid control method for collaboratively optimizing economy and power quality, specifically comprising:

[0015] In a preferred embodiment, the inventive method is specifically described as follows:

[0016] 1. Source-load unit modeling and energy scheduling strategy

[0017] The main control unit lithium battery model is as follows Figure 1 .

[0018] 1.1 Non-master source load unit modeling, load modeling and prediction

[0019] (1) Modeling of non-master source units

[0020] Modeling methods for non-master control units such as wind and solar power generation, hydrogen production, electric vehicles, and micro gas turbines can be found in Figure 2The random wind speed sequence in wind turbine power generation follows a Weibull distribution, and the light intensity in photovoltaic power generation follows a beta distribution. In the microgrid model built in this project, electric vehicles (PEVs, all assumed to be lithium batteries) are similar to other relatively independent and somewhat random energy storage units. The difference is that PEVs can have both on-grid and off-grid states, while lithium batteries (before being scrapped or repaired) can only have an on-grid state. Each electric vehicle must meet strict time and energy constraints.

[0021] 1.2 Unified load model

[0022] For dispatch capacity optimization, referring to the unified load modeling method (see Table 2), electricity demand can be divided into:

[0023] Non-shiftable loads (NSLs): loads whose distribution is predictable but uncontrollable, such as lighting and emergency demand;

[0024] Transferable Loads (PLs): These are divided into time-shiftable loads (TLs) (such as washing machines and dishwashers) and power-shiftable loads (PLs). TLs are loads with adjustable time and fixed power (such as water heaters).

[0025] Load Forecasting

[0026] By analyzing and studying the historical load data of the microgrid, a short-term load forecasting model for the microgrid based on the Elman neural network is established, and the particle swarm algorithm is used to optimize the model parameters. Specifically, it includes:

[0027] 1) Microgrid short-term load forecasting modeling based on Elman neural network, the process is as follows Figure 1 .

[0028] 2) Optimization of weight parameters of load forecasting model based on particle swarm algorithm.

[0029] 1.3 Energy Optimization Scheduling Strategy Design

[0030] In microgrids, since a large amount of heat is generated during the power generation process, in addition to a portion used to maintain the heat demand of the power generation itself, the excess heat will be used for the heating load of each household, and a thermal energy storage unit will be introduced in the heating part. This part mainly studies the optimization scheduling strategy of electric energy based on the balance of supply and demand. Therefore, the specific design principles of the thermal scheduling strategy are first determined as follows:

[0031] The microgrid's heat supply and demand are balanced as much as possible by the SOFC, micro gas turbine, thermal energy storage, and user heat demand. As a constraint on the overall energy supply and demand balance, excess heat within the microgrid goes to the thermal energy storage unit and is not traded with the main grid.

[0032] The user's heat load is preferentially supplied by the heat source of the microgrid system. When the heat supply is insufficient, electric energy is used for heating;

[0033] The excess and insufficient thermal energy storage are converted into the value of corresponding electricity for dispatching schemes and operational economic evaluation.

[0034] (1) Design of microgrid power optimization dispatch strategy:

[0035] The overall power balance of the microgrid and the constraints of related units (output range constraints, micro gas turbine ramp rate constraints, network power flow constraints, etc.) are expressed as follows:

[0036] 1) Power balance

[0037] To ensure the safe and stable operation of the microgrid, the microgrid dispatch must meet the power balance, that is, the supply (including wind and solar power, energy storage (including lithium batteries and electric vehicles), and micro gas turbines) is equal to the demand:

[0038] P LOAD (t) = P WT (t)+P PV (t)+P BAT (t)+P MT (t)

[0039] 2) Micro power output

[0040] Define the maximum and minimum output constraints of the micro power source. The specific expressions are as follows:

[0041] P i,min ≤P i (t)≤P i,max

[0042] Where, P i,max 、P i,min are the maximum and minimum output of each micro power source respectively.

[0043] 3) Lithium battery state of charge

[0044] The lithium battery charge constraint is established, and the expression is as follows:

[0045] SOC min ≤SOC(t)≤SOC max

[0046] Where, SOC(t), SOC max , SOC min They are the state of charge, maximum and minimum values ​​of the lithium battery in time period t. The state of charge calculation method is as follows:

[0047]

[0048] Where SOC(t-1) is the state of charge of the lithium battery at time t-1, C bm The maximum storage capacity of the lithium battery.

[0049] 4) Multiple optimization objectives: The comprehensive optimization objectives are to minimize the microgrid operating cost, maximize the renewable energy self-use rate, minimize the network loss, and minimize the voltage deviation value.

[0050] Traditional mathematical methods are difficult to solve accurately, so this project plans to use a multi-objective differential evolution algorithm for rapid solution.

[0051] In addition, because the weight factors of the influence of the dispatchable evaluation indicators on the optimization objectives are subjective and empirical to a certain extent, a certain number of reasonable weight factor combinations can be selected to obtain a series of optimization scheduling method sets, which is conducive to the rapid iterative acquisition of feasible solutions for power flow calculations during the optimization design of scheduling strategies.

[0052] 2. Power flow calculation

[0053] In the process of optimizing the dispatching strategy design, flow calculations are required based on the power of each power source and load point in the system, the voltage at the hub point, the voltage at the balance point, and the phase angle to determine the steady-state operating parameters of each part of the power system: including the voltage amplitude and phase angle of each bus node in the power grid, as well as the power distribution of each branch, the power loss of the network, etc., in order to determine whether the designed strategy is physically feasible; and based on the flow calculation results, the dispatching strategy design can be further iterated and optimized.

[0054] This project takes the microgrid with SOFC+lithium battery as the main control unit balancing node as the research object. The microgrid power flow model based on node power is described in matrix form as follows:

[0055] F(x)=0,x∈R n

[0056] Where F(x) is a set of node power nonlinear function vectors, x is the system unknown variable vector, and n is the number of unknown variables. Suppose there are N nodes, of which there is 1 balance node, numbered 1; there are M PQ nodes, numbered 2, 3, ..., M+1; there are (NM-1) PV nodes, numbered M+2, M+3, ..., N. F(x) = [F PPQ2 ,…,F PPQ(M+1) ,F PPV(M+2) ,…,F PPVN ,F QPQ2 ,…,F QPQ(M+1) ] T

[0057] 3 Microgrid Overall Coordinated Control

[0058] (1) Determination of the overall control architecture of the microgrid for this project

[0059] 1) Microgrid Grid-Connected Operation: A grid-connected microgrid uses master-slave control. In this mode, the interconnection line between the microgrid and the distribution system operates at a predetermined power value. The microgrid's master control unit (PQ constant power control) regulates the randomness of the microgrid system, maintaining frequency and voltage stability and absorbing internal fluctuations.

[0060] 2) Microgrid Island Operation: Island operation employs master-slave control. In this mode, the microgrid is disconnected from the distribution system. The microgrid's master control unit, which operates under constant voltage and frequency control, provides voltage and frequency references to other distributed power units, maintaining frequency and voltage stability. Other distributed power units utilize constant power control.

[0061] (2) Power quality control with lithium battery as the main control unit

[0062] The main control unit formed by the lithium battery makes up for the lack of redundancy in the power quality control of conventional energy storage units during the power quality control process.

[0063] The PQ control method is used to couple and separate current and power to analyze the factors that lead to microgrid voltage instability.

[0064] To offset the disturbance of harmonic distortion and voltage imbalance in the microgrid, a software phase-locked loop method based on a biquad generalized integrator is used to separate the positive and negative sequence components of the voltage fundamental in the microgrid, calculate the reference current, and detect the amount of current required for compensation.

[0065] The lithium battery main control unit uses adjustable residual capacity to compensate for power quality through a current converter to suppress unstable vibrations of the power waveform in the microgrid, and fully compensates for unbalanced voltage, reactive power waveforms, and harmonic distortion in all frequency bands by using all-band compensation control methods.

[0066] When selectively compensating for a specific frequency, a proportional vector integral control method is used to adjust the specific frequency of the unbalanced voltage, reactive power waveform and harmonic distortion.

[0067] (3) Establishment of a multi-source load storage microgrid control architecture based on Bayesian learning

[0068] Model Predictive Control (MPC) has a powerful ability to handle input / state constraints, cross-coupling, and system uncertainty. This project selected it as the control algorithm for implementing a series of microgrid optimization strategies. Based on the designed steady-state optimal strategy and dynamic switching optimization strategy for the master control unit, the established performance prediction, fault detection, and location methodologies were further expanded to the entire microgrid environment. By combining models with data-driven approaches and introducing Bayesian algorithms, the system parameters of the entire microgrid were identified and predicted.

[0069] The Bayesian algorithm first treats the unknown parameter vector to be estimated as a random vector conforming to a certain prior distribution. Based on previous knowledge of the desired parameter, the prior distribution is determined. Then, based on sample information and relevant rules, the posterior probability distribution is calculated. Finally, the prior information and posterior probability are combined to infer the unknown parameter. By utilizing a sparse Bayesian learning framework, feature quantities can be selected to derive an accurate prediction model, typically using fewer basis functions. Furthermore, compared to other learning algorithms, Bayesian learning algorithms have the advantages of rapid convergence and reduced number of subsequent learning steps, which facilitates model-based and data-driven research.

[0070] By acquiring a large amount of experimental data and based on the Bayesian online learning method, the system parameters are estimated. Then, using the model predictive control algorithm as the framework, the input variables, constraint intervals and optimization indicators for power reference trajectory matching are designed to achieve coordinated control for the optimization of indicators such as system efficiency, high self-use rate of renewable energy, and economy under the conditions of smooth and safe and rapid power tracking throughout the entire microgrid operation process.

[0071] The present invention has the following beneficial effects:

[0072] 1. Main control unit lithium battery modeling, non-main control unit modeling: Model various loads including uncertain wind and solar units, hydrogen storage devices, micro gas turbines, electric vehicles, and households in microgrids, and analyze their output characteristics, dynamic responses, and constraint ranges.

[0073] 2. Forecast various loads to provide data support for the design of energy optimization scheduling strategies. Through comprehensive analysis and research of microgrid historical load data, a short-term load forecasting model is established and the model parameters are optimized.

[0074] 3. Energy optimization scheduling and fast power flow calculation based on dispatchability evaluation

[0075] Based on the characteristics of each source-load unit, and aiming at the needs of economy, high self-use rate of renewable energy, and safe and fast microgrid scheduling, the energy scheduling strategy is optimized by taking into account the dispatchability of each source unit and the transferability of each load, with the strong dispatchability of the main control unit and the real-time compensation capability of energy storage redundancy as the core support for achieving "response speed and cost optimization".

[0076] 4. Overall coordinated control of microgrids: dynamic and static analysis and optimization control strategies and optimal dispatch strategies for each unit are obtained. Aiming at the overall environment of multi-source load storage in microgrids, with high power quality stability, system safety and overall economy as the goals, and driven by the low-cost power generation and rapid response capabilities of the main control unit, the power quality control and overall coordinated control of microgrids are designed and implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 This is a flow chart of the microgrid short-term load forecasting modeling method based on Elman neural network of the present invention;

[0078] Figure 2 This is a predictive control diagram of a multi-source load storage microgrid model based on Bayesian online learning in the present invention;

[0079] Figure 3 This is a test result diagram of the energy storage unit of the present invention;

[0080] Figure 4 This is a curve diagram of the power change of the battery and the interaction with the distribution network under the improved MPC of the present invention. DETAILED DESCRIPTION

[0081] The technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention. In addition, the forms of the various structures described in the following embodiments are merely examples. The method for controlling a smart microgrid for collaborative optimization of economic efficiency and power quality involved in the present invention is not limited to the various structures described in the following embodiments. All other implementations obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0082] Here’s how it works:

[0083] (1) Main control unit lithium battery modeling and non-main control unit modeling:

[0084] The main control unit lithium battery model is shown in the following table.

[0085]

[0086] Modeling of non-master source units

[0087] The modeling methods for non-master control units, such as wind and solar power generation, hydrogen production, electric vehicles, and micro-gas turbines, are shown in the table below. The random wind speed sequence for wind turbine generation follows a Weibull distribution, and the light intensity for photovoltaic generation follows a beta distribution. In the microgrid model established in this project, electric vehicles (PEVs, all assumed to be lithium batteries) are similar to other relatively independent and somewhat random energy storage units. The difference is that PEVs can have both on-grid and off-grid states, while lithium batteries (before being scrapped or repaired) can only have an on-grid state. Each electric vehicle must meet strict time and energy constraints.

[0088]

[0089]

[0090]

[0091] Unified load model

[0092] For dispatch capacity optimization, referring to the unified load modeling method, electricity demand can be divided into:

[0093] Non-shiftable loads (NSLs): loads whose distribution is predictable but uncontrollable, such as lighting and emergency demand;

[0094] Transferable Loads (PLs): These are divided into time-shiftable loads (TLs) (such as washing machines and dishwashers) and power-shiftable loads (PLs). TLs are loads with adjustable time and fixed power (such as water heaters).

[0095] Load Forecasting

[0096] By analyzing and studying the historical load data of the microgrid, a short-term load forecasting model for the microgrid based on the Elman neural network is established, and the particle swarm algorithm is used to optimize the model parameters. Specifically, it includes:

[0097] 1) Microgrid short-term load forecasting modeling based on Elman neural network, the process is as follows Figure 1 .

[0098] 2) Optimization of weight parameters of load forecasting model based on particle swarm algorithm.

[0099] 2. Energy optimization scheduling strategy design

[0100] In microgrids, since a large amount of heat is generated during the power generation process, in addition to a portion used to maintain the heat demand of the power generation itself, the excess heat will be used for the heating load of each household, and a thermal energy storage unit will be introduced in the heating part. This part mainly studies the optimization scheduling strategy of electric energy based on the balance of supply and demand. Therefore, the specific design principles of the thermal scheduling strategy are first determined as follows:

[0101] The microgrid's heat supply and demand are balanced as much as possible by the SOFC, micro gas turbine, thermal energy storage, and user heat demand. As a constraint on the overall energy supply and demand balance, excess heat within the microgrid goes to the thermal energy storage unit and is not traded with the main grid.

[0102] The user's heat load is preferentially supplied by the heat source of the microgrid system. When the heat supply is insufficient, electric energy is used for heating;

[0103] The excess and insufficient thermal energy storage are converted into the value of corresponding electricity for dispatching schemes and operational economic evaluation.

[0104] (1) Design of microgrid power optimization dispatch strategy:

[0105] The overall power balance of the microgrid and the constraints of related units (output range constraints, micro gas turbine ramp rate constraints, network power flow constraints, etc.) are expressed as follows:

[0106] 1) Power balance

[0107] To ensure the safe and stable operation of the microgrid, the microgrid dispatch must meet the power balance, that is, the supply (including wind and solar power, energy storage (including lithium batteries and electric vehicles), and micro gas turbines) is equal to the demand:

[0108] P LOAD (t) = P WT (t)+P PV (t)+P BAT (t)+P MT (t)

[0109] 2) Micro power output

[0110] Define the maximum and minimum output constraints of the micro power source. The specific expressions are as follows:

[0111] P i,min ≤P i (t)≤P i,max

[0112] Where, P i,max 、P i,min are the maximum and minimum output of each micro power source respectively.

[0113] 3) Lithium battery state of charge

[0114] The lithium battery charge constraint is established, and the expression is as follows:

[0115] SOC min ≤SOC(t)≤SOC max

[0116] Where, SOC(t), SOC max, SOC min They are the state of charge, maximum and minimum values ​​of the lithium battery in time period t. The state of charge calculation method is as follows:

[0117]

[0118] Where SOC(t-1) is the state of charge of the lithium battery at time t-1, C bm The maximum storage capacity of the lithium battery.

[0119] 4) Multiple optimization objectives: The comprehensive optimization objectives are to minimize the microgrid operating cost, maximize the renewable energy self-use rate, minimize the network loss, and minimize the voltage deviation value.

[0120] (2) Establish a comprehensive evaluation system for the dispatchability of each unit, and use a comprehensive evaluation method based on the TOPSIS method and the Delphi method to design a multi-objective optimization method for the dispatch method:

[0121] Based on the core dominance of the main control unit and the strong adjustability of the energy storage unit in the design of the dispatching strategy, combined with the online status and historical data of various energy storage units (including the lithium batteries in the main control unit and the lithium batteries in the electric vehicle), it is proposed to first establish a comprehensive evaluation system for the dispatchability of each energy storage unit. Specifically, the following four evaluation indicators are considered: battery loss degree, charging urgency, reverse power supply capability, and confidence level (taking into account the essential attributes of electric vehicles as a means of transportation, their automatic disconnection from the grid due to emergencies will to a certain extent affect the regulation effect; the concept of credit rating is introduced to characterize the possibility of the energy storage response subject participating in the regulation strategy within a certain time period; at the same time, its failure rate is taken into account). Based on this, the Delphi method is used to determine the weight factors of each indicator to form an evaluation matrix for the dispatchability of the energy storage unit.

[0122] Similarly, the dispatchability evaluation matrix of other related units can be obtained.

[0123] Then, for the multi-objective optimization of scheduling strategies, the TOPSIS method is used based on vector normalization to obtain the canonical decision matrix of the scheduling strategy alternative set, which is combined with the obtained unit dispatchability evaluation matrix to form an optimized weighted canonical matrix. Then, the Euclidean distance between the ideal solution and the negative ideal solution corresponding to different schemes is calculated and compared to achieve quantitative evaluation of each alternative scheduling strategy.

[0124] 3. Power flow calculation

[0125] In the process of optimizing the dispatching strategy design, flow calculations are required based on the power of each power source and load point in the system, the voltage at the hub point, the voltage at the balance point, and the phase angle to determine the steady-state operating parameters of each part of the power system: including the voltage amplitude and phase angle of each bus node in the power grid, as well as the power distribution of each branch, the power loss of the network, etc., in order to determine whether the designed strategy is physically feasible; and based on the flow calculation results, the dispatching strategy design can be further iterated and optimized.

[0126] This project takes the microgrid with SOFC+lithium battery as the main control unit balancing node as the research object. The microgrid power flow model based on node power is described in matrix form as follows:

[0127] F(x)=0,x∈R n

[0128] Where F(x) is a set of node power nonlinear function vectors, x is the system unknown variable vector, and n is the number of unknown variables.

[0129] 4 Microgrid overall coordinated control

[0130] (1) Determination of the overall control architecture of the microgrid for this project

[0131] 1) Microgrid Grid-Connected Operation: A grid-connected microgrid uses master-slave control. In this mode, the interconnection line between the microgrid and the distribution system operates at a predetermined power value. The microgrid's master control unit (PQ constant power control) regulates the randomness of the microgrid system, maintaining frequency and voltage stability and absorbing internal fluctuations.

[0132] 2) Microgrid Island Operation: Island operation employs master-slave control. In this mode, the microgrid is disconnected from the distribution system. The microgrid's master control unit, which operates under constant voltage and frequency control, provides voltage and frequency references to other distributed power units, maintaining frequency and voltage stability. Other distributed power units utilize constant power control.

[0133] (2) Power quality control with lithium battery as the main control unit

[0134] The main control unit formed by the lithium battery makes up for the lack of redundancy in the power quality control of conventional energy storage units during the power quality control process.

[0135] The PQ control method is used to couple and separate current and power to analyze the factors that lead to microgrid voltage instability.

[0136] To offset the disturbance of harmonic distortion and voltage imbalance in the microgrid, a software phase-locked loop method based on a biquad generalized integrator is used to separate the positive and negative sequence components of the voltage fundamental in the microgrid, calculate the reference current, and detect the amount of current required for compensation.

[0137] The lithium battery main control unit uses adjustable residual capacity to compensate for power quality through a current converter to suppress unstable vibrations of the power waveform in the microgrid, and fully compensates for unbalanced voltage, reactive power waveforms, and harmonic distortion in all frequency bands by using all-band compensation control methods.

[0138] When selectively compensating for a specific frequency, a proportional vector integral control method is used to adjust the specific frequency of the unbalanced voltage, reactive power waveform and harmonic distortion.

[0139] (3) Establishment of a multi-source load storage microgrid control architecture based on Bayesian learning

[0140] Model Predictive Control (MPC) has a powerful ability to handle input / state constraints, cross-coupling, and system uncertainties. This project selected it as the control algorithm for the implementation of a series of microgrid optimization strategies. Based on the designed steady-state optimal strategy and dynamic switching optimization strategy for the main control unit, combined with the existing mature unit-level analysis methods and results for other types of units (including wind, solar, micro gas turbines, etc.) as described in the previous "Overview of Field Research", the established performance prediction, fault detection and location method system is further expanded to the entire microgrid environment. By combining models with data-driven, the Bayesian algorithm is introduced to identify and predict the overall microgrid system parameters.

[0141] The Bayesian algorithm first treats the unknown parameter vector to be estimated as a random vector conforming to a certain prior distribution. Based on previous knowledge of the desired parameter, the prior distribution is determined. Then, based on sample information and relevant rules, the posterior probability distribution is calculated. Finally, the prior information and posterior probability are combined to infer the unknown parameter. By utilizing a sparse Bayesian learning framework, feature quantities can be selected to derive an accurate prediction model, typically using fewer basis functions. Furthermore, compared to other learning algorithms, Bayesian learning algorithms have the advantages of rapid convergence and reduced number of subsequent learning steps, which facilitates model-based and data-driven research.

[0142] By acquiring a large amount of experimental data and based on the Bayesian online learning method, the system parameters are estimated. Then, using the model predictive control algorithm as the framework, the input variables, constraint intervals and optimization indicators for power reference trajectory matching are designed to achieve coordinated control for the optimization of indicators such as system efficiency, high self-use rate of renewable energy, and economy under the conditions of smooth and safe and rapid power tracking throughout the entire microgrid operation process.

[0143] Implementation method such as Figure 2 shown.

[0144] 5. Case analysis

[0145] 5.1 Control Strategy Verification

[0146] Figure 1 This is a flow chart for the short-term load forecasting modeling method for microgrids based on an Elman neural network. To verify the effectiveness of the energy storage unit control strategy in grid-connected mode, this section conducts experimental tests using a DC-side energy storage unit as the test object. In this experiment, an AC / DC hybrid grid is connected to the main grid for grid-connected operation. In grid-connected mode, the DC subgrid bus voltage is controlled by the BPC, while the charge and discharge power of the energy storage battery is regulated by the controller in grid-connected mode based on the battery's SOC. Since the main grid can provide power support to the hybrid grid during grid-connected operation, the energy storage battery only needs to charge and discharge according to the reference power provided by the controller. In this experiment, the DC-side photovoltaic output power is constant at 40kW, and the DC load power demand is equal to the photovoltaic output power, also 40kW.

[0147] Taking the charging process of the DC side energy storage unit as an example, the relevant experimental test was carried out, and the SOC starting value was 50%. The relevant experimental results are as follows Figure 3 shown. Figure 3 (a) is the charging power curve of the DC side energy storage unit. Figure 3 (b) is the SOC curve of the DC side energy storage unit.

[0148] In stage I, since the SOC of the energy storage unit is in a lower range, the energy storage unit is charged at maximum power, and the SOC gradually increases. In stage II, since the SOC of the energy storage unit rises to a higher value, the energy storage unit charging power gradually decreases under the action of the energy storage unit grid-connected mode controller. The SOC still gradually increases, but the rate of increase slows down. By stage III, the energy storage unit SOC is close to the maximum value of 95%. Under the action of the controller, the energy storage unit charging power is reduced to 0, the SOC no longer increases, and the system continues to operate stably. The relevant results of this experiment show that in the grid-connected mode, the designed energy storage unit control strategy can ensure that the SOC of the energy storage unit is within the allowable range while adjusting its charging power in real time according to the SOC of the energy storage unit, making the use of the energy storage unit safer, smarter and more reasonable. The relevant experimental results verify the effectiveness of the proposed control strategy.

[0149] 5.2 Dispatch Economic Efficiency Calculation Example

[0150] This paper analyzes a microgrid topology based on a previous study. The renewable energy generation system includes an 800kW photovoltaic system and an 800kW wind turbine. The predicted and actual power generation and load demand for the photovoltaic and wind turbines over a 24-hour period are presented. The battery has a total capacity of 2500kWh, a maximum charge and discharge power of 800kW, SOC upper and lower limits of 0.9 and 0.5, respectively, and a charge and discharge efficiency of 0.95. The battery maintenance cost is 0.025 yuan / kW. The maximum interactive power between the microgrid and the distribution network is 800kW, and the electricity price is a uniform 0.58 yuan / kW.

[0151] The total cost of compensating for the prediction error with reference to the traditional distribution network regulation method is 65.55 yuan, and the total operating cost plus the compensation cost is 896.1 yuan.

[0152] The improved MPC strategy in this paper is as follows Figure 2 The block diagram of the predictive control of the multi-source load storage microgrid model based on Bayesian online learning is as follows: Figure 4 As shown in the figure, the trend of the improved power interaction graph is similar to the MPC strategy power graph, and the interactive power with the distribution network at 20-22 is reduced to 790kW, which is lower than the power upper limit of 800kW. There is no behavior of exceeding the upper limit during the entire period. Figure 3 It can be calculated that the total cost of the improved MPC is 1,465.9 yuan, and the remaining battery power is still 2,087.1 kWh (because the prediction error compensation power is only added to the interaction power with the distribution network, and the battery output power does not need to be compensated). It can be seen that the total cost of the improved MPC of 1,465.9 yuan is lower than the total cost of the traditional MPC strategy of 1,896.1 yuan, and the total cost is reduced by 22.69%.

[0153] It can be seen from this that the scheduling and control strategies proposed in this patent have achieved good results in terms of economy and controllability.

[0154] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0155] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A smart microgrid control method for collaboratively optimizing economic efficiency and power quality, characterized in that: The method comprises the following steps: Step 1: Under the premise of controllable costs, using the energy storage unit as the microgrid master control unit and system balancing node can compensate for the shortcomings of uncontrollable and insufficient redundancy of the energy storage unit in real time, which is an effective way to fundamentally improve the system's power quality and comprehensive control capabilities. Step 2: Modeling the main control unit lithium battery and non-main control unit: Models were created for various load types within the microgrid, including uncertain wind and solar power units, hydrogen storage devices, micro gas turbines, electric vehicles, and households, analyzing their output characteristics, dynamic responses, and constraint ranges. A hydrogen production process model was established to fully utilize the excess capacity of wind and solar power generation. Step 3: It is necessary to predict various loads to provide data support for the design of energy optimization scheduling strategies In actual operation, the output characteristics of the relevant uncertain source units (as negative loads) and the uncertain thermal power demand need to be accurately predicted in order to design a reasonable scheduling strategy. Through comprehensive analysis and research on the microgrid's historical load data, a short-term load forecasting model is established and the model parameters are optimized. Step 4: Energy optimization scheduling and fast power flow calculation based on dispatchability evaluation Based on the characteristics of each source-load unit, and targeting the needs of economic efficiency, high renewable energy self-use rate, and safe and fast microgrid scheduling, the energy optimization scheduling strategy is designed to obtain a series of optimized scheduling strategy sets, taking into account the dispatchability of each source unit and the transferability of each load. The strong dispatchability of the main control unit and the real-time compensation capability of energy storage redundancy are used as the core support for achieving "response speed and cost optimization". Step 5: Microgrid overall coordinated control Determine the overall control framework of the microgrid. Based on the dynamic and static analysis and optimization control strategies and optimal scheduling strategies of each unit obtained above, facing the overall environment of multi-source load storage in the microgrid, with high power quality stability, system safety and overall economy as the goals, and driven by the low-cost power generation and rapid response capabilities of the main control unit, design and implement microgrid power quality control and overall coordinated control.

2. The method for controlling a smart microgrid by collaboratively optimizing economic efficiency and power quality according to claim 1, characterized in that: It is necessary to model the main control unit lithium battery and non-main control unit. The specific steps are as follows: Models were created for various load types within the microgrid, including uncertain wind and solar power units, hydrogen storage devices, micro gas turbines, electric vehicles, and households, analyzing their output characteristics, dynamic responses, and constraint ranges. A hydrogen production process model was established to fully utilize the excess capacity of wind and solar power generation. Main control unit lithium battery modeling: Battery state space expression: The decay state space expression is: Q k+1 =Q k +W Q , When the battery SOC is lower than SOC low Start charging when the SOC is higher than high Stop charging when So, It is estimated by the following method: Among them, W k and V k represent the process noise and measurement noise of the battery respectively; T s represents the model sampling time, τ b Indicates the voltage response time constant of the lithium battery activation polarization link, its value is the polarization capacitance C b and polarization resistance R b The product of τ p is the time constant of the concentration polarization link, and its value is C p and R p The product of U b 、U p They represent the polarization voltages generated by the two polarization links; η represents the Coulomb effect; Q represents the battery capacity; W Q and V Q denote the attenuation process noise and measurement noise, respectively.

3. The method for controlling a smart microgrid by collaboratively optimizing economic efficiency and power quality according to claim 1, characterized in that: It is necessary to forecast various loads to provide data support for the design of energy optimization scheduling strategies. The specific steps are as follows: In actual operation, the output characteristics of the relevant uncertain source units (as negative loads) and the uncertain thermal power demand need to be accurately predicted in order to design a reasonable scheduling strategy. Through comprehensive analysis and research on the microgrid's historical load data, a short-term load forecasting model is established and the model parameters are optimized. Load Forecasting: By analyzing and studying the historical load data of the microgrid, a short-term load forecasting model for the microgrid based on the Elman neural network is established, and the particle swarm algorithm is used to optimize the model parameters. Specifically, it includes: 1) Microgrid short-term load forecasting modeling based on Elman neural network. 2) Optimization of weight parameters of load forecasting model based on particle swarm algorithm.

4. The method for controlling a smart microgrid by collaboratively optimizing economic efficiency and power quality according to claim 1, characterized in that: Energy optimization scheduling and fast power flow calculation based on dispatchability evaluation are required. The specific steps are as follows: Based on the characteristics of each source-load unit, and targeting the needs of economic efficiency, high renewable energy self-use rate, and safe and fast microgrid scheduling, the energy optimization scheduling strategy is designed to obtain a series of optimized scheduling strategy sets, taking into account the dispatchability of each source unit and the transferability of each load. The strong dispatchability of the main control unit and the real-time compensation capability of energy storage redundancy are used as the core support for achieving "response speed and cost optimization". (1) Design of microgrid power optimization dispatch strategy: The overall power balance of the microgrid and the constraints of related units (output range constraints, micro gas turbine ramp rate constraints, network power flow constraints, etc.) are expressed as follows: 1) Power balance To ensure the safe and stable operation of the microgrid, the microgrid dispatch must meet the power balance, that is, the supply (including wind and solar power, energy storage (including lithium batteries and electric vehicles), and micro gas turbines) is equal to the demand: P LOAD (t)=P WT (t)+P PV (t)+P BAT (t)+P MT (t) 2) Micro power output Define the maximum and minimum output constraints of the micro power source. The specific expressions are as follows: P i,min ≤P i (t)≤P i,max Where, P i,max 、P i,min are the maximum and minimum output of each micro power source respectively. 3) Lithium battery state of charge The lithium battery charge constraint is established, and the expression is as follows: SOC min ≤SOC(t)≤SOC max Where, SOC(t), SOC max , SOC min They are the state of charge, maximum and minimum values ​​of the lithium battery in time period t. The state of charge calculation method is as follows: In the formula, SOC(t-1) is the state of charge of the lithium battery at time t-1, C bm The maximum storage capacity of the lithium battery. 4) Multiple optimization objectives: The comprehensive optimization objectives are to minimize the microgrid operating cost, maximize the renewable energy self-use rate, minimize the network loss, and minimize the voltage deviation value. The energy scheduling optimization design of multi-source load storage microgrid is a complex optimization problem with multiple objectives, multiple variables and nonlinearity. Traditional mathematical methods are difficult to solve accurately. This project plans to use a multi-objective differential evolution algorithm for a fast solution. In addition, because the weight factors of the influence of the dispatchable evaluation indicators on the optimization objectives are subjective and empirical to a certain extent, a certain number of reasonable weight factor combinations can be selected to obtain a series of optimization scheduling method sets, which is conducive to the rapid iterative acquisition of feasible solutions for power flow calculations during the optimization design of scheduling strategies. (2) Power flow calculation In the process of optimizing the dispatching strategy design, the power flow calculation is required to determine the steady-state operating parameters of each part of the power system, including the voltage amplitude and phase angle of each bus node, the power distribution of each branch, and the power loss of the network, based on the power of each source and load point, the voltage of the hub point, and the voltage and phase angle of the balance point in the system. This is to determine whether the designed strategy is physically feasible. The scheduling strategy design can be further iterated and optimized based on the power flow calculation results. This project takes the microgrid with SOFC+lithium battery as the main control unit balancing node as the research object. The microgrid power flow model based on node power is described in matrix form as follows: F(x)=0,x∈R n Where F(x) is a set of node power nonlinear function vectors, x is the system unknown variable vector, and n is the number of unknown variables. Suppose there are N nodes, of which there is 1 balance node, numbered 1; there are M PQ nodes, numbered 2, 3, ..., M+1; there are (NM-1) PV nodes, numbered M+2, M+3, ..., N. F(x) = [F PPQ2 ,…,F PPQ(M+1) ,F PPV(M+2) ,…,F PPVN ,F QPQ2 ,…,F QPQ(M+1) ] T Where x=[U2,…,U M+1 ,δ2,…,δ M+1 ,δ M+2 ,…δ N ], then the nonlinear node power equations of the microgrid power flow calculation system are: The entire power flow calculation can be mathematically modeled as a high-dimensional non-convex optimization problem with equality and inequality constraints.

5. The method for controlling a smart microgrid by collaboratively optimizing economic efficiency and power quality according to claim 1, characterized in that: It is necessary to carry out overall coordinated control of the microgrid. The specific steps are as follows: Determine the overall control framework of the microgrid. Based on the dynamic and static analysis and optimization control strategies and optimal scheduling strategies of each unit obtained above, facing the overall environment of multi-source load storage in the microgrid, with high power quality stability, system safety and overall economy as the goals, and driven by the low-cost power generation and rapid response capabilities of the main control unit, design and implement microgrid power quality control and overall coordinated control. (1) Determination of the overall control architecture of the microgrid 1) Microgrid Grid-Connected Operation: A grid-connected microgrid uses master-slave control. In this mode, the interconnection line between the microgrid and the distribution system operates at a predetermined power value. The microgrid's master control unit (PQ constant power control) regulates the randomness of the microgrid system, maintaining frequency and voltage stability and absorbing internal fluctuations. 2) Microgrid Island Operation: Island operation employs master-slave control. In this mode, the microgrid is disconnected from the distribution system. The microgrid's master control unit, which operates under constant voltage and frequency control, provides voltage and frequency references to other distributed power units, maintaining frequency and voltage stability. Other distributed power units utilize constant power control. (2) Power quality control with lithium battery as the main control unit The PQ control method is used to couple and separate current and power to analyze the factors that lead to microgrid voltage instability. To offset the disturbance of harmonic distortion and voltage imbalance in the microgrid, a software phase-locked loop method based on a biquad generalized integrator is used to separate the positive and negative sequence components of the voltage fundamental in the microgrid, calculate the reference current, and detect the amount of current required for compensation. The lithium battery main control unit uses adjustable residual capacity to compensate for power quality through a current converter to suppress unstable vibrations of the power waveform in the microgrid, and fully compensates for unbalanced voltage, reactive power waveforms, and harmonic distortion in all frequency bands by using all-band compensation control methods. (3) Establishment of a multi-source load storage microgrid control architecture based on Bayesian learning Model Predictive Control (MPC) has a powerful ability to handle input / state constraints, cross-coupling, and system uncertainty. This project selected it as the control algorithm for implementing a series of microgrid optimization strategies. Based on the designed steady-state optimal strategy and dynamic switching optimization strategy for the main control unit, combined with the existing mature unit-level analysis methods and results for other types of units (including wind, solar, micro gas turbines, etc.) as described in the "Overview of Field Research", the established performance prediction, fault detection, and location methodologies are further expanded to the entire microgrid environment. Utilizing a combination of model and data-driven approaches, and introducing a Bayesian algorithm, the overall microgrid system parameters are identified and predicted.

Images