A Microgrid Scheduling Method and Terminal for a Hybrid Energy Storage System

Abstract

The present invention proposes a microgrid scheduling method and terminal for a hybrid energy storage system, and conducts research on energy optimization scheduling for the slow dynamic component in the power generation and consumption energy deviation of an independent microgrid system, including proposing a prediction model based on historical data for source-load data, adding source-load uncertainty to the prediction results and calculating the relevant three-phase unbalance, researching the establishment and solution method of a day-ahead optimization scheduling mathematical model with multiple energy storages, which can improve the scheduling accuracy of the microgrid with a hybrid energy storage system. During the optimization scheduling process, it avoids the particle swarm from falling into local or marginal areas based on the particle swarm; by guiding the moving direction of the particle swarm, the population can quickly and efficiently meet this constraint and find the optimal solution, thereby improving the scheduling calculation efficiency.

Description

Technical Field

[0001] The present invention relates to the field of microgrid dispatching, and in particular to a microgrid dispatching method and terminal for a hybrid energy storage system. Background Art

[0002] At present, independent microgrid systems operate independently of the main power grid and cannot obtain support from the external power grid. The risk of system collapse is relatively high after a fault occurs. The fluctuations in wind and solar energy output and various types of loads can easily cause problems such as three-phase imbalance in the microgrid.

[0003] The optimal dispatch of independent microgrid systems is a typical nonlinear programming problem with multiple constraints. Conventional optimal dispatch constraints include: wind / solar output constraints, diesel generator ramp constraints, minimum start / stop time constraints, energy storage SOC constraints, etc. Since the optimal dispatch model contains many nonlinear constraints or network constraints, the current linear programming method is difficult to adapt to various optimal dispatch models. How to reasonably select or improve intelligent algorithms with high computational efficiency, matching constraints, and small errors has also become a research hotspot for microgrid dispatch optimization. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a microgrid scheduling method and terminal for a hybrid energy storage system, which are used to solve the stability control problem of an independent microgrid system, and can improve the scheduling accuracy of the microgrid of the hybrid energy storage system while improving the computing efficiency.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A microgrid dispatching method for a hybrid energy storage system, comprising the steps of:

[0007] S1. Predict the renewable energy power and load power of the next day based on the historical data of the hybrid energy storage system, and perform uncertainty analysis on the renewable energy power according to the predicted data, wherein the renewable energy power includes photovoltaic power and wind power;

[0008] S2. Calculate the power three-phase imbalance degree of the predicted data at each moment without the participation of energy storage;

[0009] S3, taking renewable energy power, energy storage power, load switching power and diesel generator power as particles, initializing the particles according to constraint conditions, and obtaining a particle population;

[0010] S4. Iterate the particle population, calculate the local optimal solution of the particle population for each iteration. If the local optimal solution of the particle population in the current iteration is better than that in the previous iteration, then use the local optimal solution of the particle population in the current iteration as the global optimal solution; otherwise, use the local optimal solution of the particle population in the previous iteration as the global optimal solution.

[0011] S5. Determine whether the global optimal solution or the number of iterations reaches the termination condition. If so, output the output plans of the energy storage, diesel generator, wind power, and photovoltaic system, as well as the switching plan of the load, for the next day predicted in the global optimal solution; otherwise, update the particle population based on the power three-phase unbalance degree.

[0012] To solve the above technical problems, another technical solution adopted by the present invention is:

[0013] A microgrid dispatching terminal for a hybrid energy storage system includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements each step of the above microgrid dispatching method for a hybrid energy storage system.

[0014] The beneficial effects of the present invention are as follows: Research on energy optimization dispatching for the slow dynamic component in the power generation and consumption energy deviation of an independent microgrid system, including proposing a prediction model based on historical data for source and load data, adding source and load uncertainties to the prediction results and calculating the relevant three-phase unbalance, and research on the establishment and solution method of a day-ahead optimization dispatching mathematical model with multiple energy storages, which can improve the dispatching accuracy of the microgrid of the hybrid energy storage system. During the optimization dispatching process, on the basis of the particle swarm, avoid the particle swarm from falling into local or marginal areas; by guiding the moving direction of the particle swarm, enable the population to quickly and efficiently meet this constraint and find the optimal solution, thereby improving the dispatching calculation efficiency. Description of the Drawings

[0015] Figure 1 It is a flowchart of a microgrid dispatching method for a hybrid energy storage system according to an embodiment of the present invention;

[0016] Figure 2 It is a structural diagram of a microgrid dispatching terminal for a hybrid energy storage system according to an embodiment of the present invention;

[0017] Figure 3 It is a schematic diagram of the specific steps of a microgrid dispatching method for a hybrid energy storage system according to an embodiment of the present invention;

[0018] Figure 4 It is a fuzzy rule membership function graph of a probability fuzzy rule particle swarm optimization algorithm according to an embodiment of the present invention.

[0019] Label description:

[0020] 1. A microgrid dispatching terminal for a hybrid energy storage system; 2. A memory; 3. A processor. Specific implementation mode

[0021] To describe the technical content, achieved objectives and effects of the present invention in detail, the following is described in conjunction with the implementation modes and with reference to the accompanying drawings.

[0022] Please refer to Figure 1 , this embodiment provides a microgrid dispatching method for a hybrid energy storage system, including the steps:

[0023] S1. Predict the renewable energy power and load power for the next day based on the historical data of the hybrid energy storage system, and perform uncertainty analysis on the renewable energy power according to the prediction data. The renewable energy power includes photovoltaic power and wind power;

[0024] S2. Calculate the power three-phase unbalance degree at each moment without energy storage participation for the prediction data;

[0025] S3. Use the renewable energy power, energy storage power, load switching power and diesel generator power as particles, and initialize the particles according to the constraint conditions to obtain a particle population;

[0026] S4. Iterate the particle population, calculate the local optimal solution of the particle population for each iteration. If the local optimal solution of the particle population for the current iteration is better than the local optimal solution of the particle population for the previous iteration, then use the local optimal solution of the particle population for the current iteration as the global optimal solution; otherwise, use the local optimal solution of the particle population for the previous iteration as the global optimal solution;

[0027] S5. Judge whether the global optimal solution or the number of iterations reaches the termination condition. If so, output the output plans of the energy storage, diesel generator, wind power, photovoltaic system and the load switching plan for the next day predicted in the global optimal solution; otherwise, update the particle population based on the power three-phase unbalance degree.

[0028] As can be seen from the above description, the beneficial effects of the present invention are as follows: Research on energy optimization scheduling for the slow dynamic component in the power generation and consumption energy deviation of an independent microgrid system, including proposing a prediction model based on historical data for source and load data, adding source and load uncertainties to the prediction results and calculating the relevant three-phase unbalance, and researching the establishment and solution method of a day-ahead optimization scheduling mathematical model with multiple energy storages, which can improve the scheduling accuracy of the microgrid with a hybrid energy storage system. During the optimization scheduling process, on the basis of the particle swarm, it is avoided that the particle swarm falls into local or marginal areas; by guiding the moving direction of the particle swarm, the population can quickly and efficiently meet this constraint and find the optimal solution, thereby improving the scheduling calculation efficiency.

[0029] Further, the prediction of the renewable energy power and the load power for the next day based on the historical data of the hybrid energy storage system includes:

[0030] S11. Simulate the wind speed prediction error distribution using the normal probability distribution and predict the total wind power for the next day;

[0031] S12. Calculate the prediction error of the photovoltaic output power through the light intensity and predict the total photovoltaic power for the next day;

[0032] S13. Simulate the load prediction error using the normal distribution and predict the load power for the next day.

[0033] As can be seen from the above description, the corresponding power is obtained by calculating the prediction error, thereby improving the prediction efficiency.

[0034] Further, taking the renewable energy power, the energy storage power, the load switching power, and the diesel generator power as particles includes:

[0035] Establish a set P of planned curves and a set z of operating state values for the energy storage power, the load switching power, and the diesel generator power:

[0036] P = {P o,i , P bess,j , P pv,k , P w,l}, z = {z o,i , z bess,j , z sload,q , z cload,r};

[0037]

[0038] In the formula, P o , P bess , P pv , P w respectively represent the output plan matrices of the diesel generator set, the energy-type energy storage unit, the photovoltaic power generation system, and the wind power generation system; zo , z bess respectively represent the operation status matrices of the diesel generator set and the energy storage unit; z sload , z cload respectively represent the switching plan matrices of the secondary load and the interruptible load participating in the demand-side response; P o,i represents the power generation plan curve of the i-th diesel generator set in the next 24 hours; P bess,j represents the charge and discharge plan curve of the j-th energy storage in the next 24 hours; P pv,k represents the output plan of the k-th photovoltaic power generation system in the next 24 hours; P w,l represents the output plan of the l-th wind power generation system in the next 24 hours; z o,i represents the start-stop plan of the i-th diesel generator set in the next 24 hours; z bess,j represents the start-stop plan of the j-th energy storage in the next 24 hours; z sload,q represents the switching plan of the q-th secondary load in the next 24 hours; z cload,r represents the switching plan of the r-th interruptible load participating in the demand-side response; n o , n bess , n pv , n w , n sload , n cload respectively represent the quantities of diesel generator sets, energy storage units, photovoltaic power generation systems, wind power generation systems, secondary loads, and interruptible loads participating in the demand-side response in the independent microgrid.

[0039] Furthermore, the initialization of the particles according to the constraint conditions includes:

[0040] Generating an initial population of the particle swarm under the constraints of system power balance, energy storage system operation, diesel power generation system operation, renewable energy power generation system operation, reserve capacity, and three-phase unbalance.

[0041] Furthermore, after iterating the particle population, it includes:

[0042] Calculating the fitness of each particle, where the fitness includes an operating cost item and a decision penalty item;

[0043] The operating cost item includes the renewable energy power generation system cost, the diesel generator system operating cost, the energy storage unit operating cost, and the load revenue;

[0044] The penalty item includes a penalty item for wind and light curtailment and a penalty item for load loss.

[0045] As described above, in the optimized scheduling model, the objective function includes an operating cost term and a penalty term. To minimize the operating cost of the microgrid system, two penalty terms for curtailed wind and curtailed light and the load loss are introduced to ensure that when solving the scheduling optimization model for a given date, decisions to curtail wind and light or shed load are only made when the relevant system security and stability constraints are not met.

[0046] Further, step S5 specifically includes the following steps:

[0047] S51. Update the particle velocity and position, with the formulas as follows:

[0048] v i,p (t + 1) = ε(t)v i,p (t) + c1(t)m1[b i,p (t) - x i,p (t)] + c2(t)m2[b ap (t) - x i,p (t)];

[0049] x i,p (t + 1) = x i,p (t) + v i,p (t + 1);

[0050] In the formula, v i,p (t) represents the velocity of the i-th particle in the t-th generation; ε represents the inertia factor; c1 and c2 represent the acceleration factors; m1 and m2 represent random numbers obeying the uniform normal distribution; b i,p (t) represents the individual optimal position found by particle i during the search process; b ap (t) represents the globally optimal position found by the entire particle swarm during the entire search process; x i,p (t) represents the position of the i-th particle in the t-th generation;

[0051] S52. Optimize the optimal position of the particle swarm using the chaos rule and calculate the feasible solution vector corresponding to the optimal solution;

[0052] S53. Randomly select a particle in the current subgroup and replace the position vector of the selected particle with the position vector of the feasible solution vector;

[0053] S54. When the calculated fitness variance is less than the given preset variance value, re-initialize the current population. The calculation method of the population fitness variance is:

[0054]

[0055]

[0056] In the formula, fg , f i , respectively represent the normalization factor, fitness value, and average fitness value of the current particle population of particle i;

[0057] S55. Determine whether the current operation meets the stop condition. If it does, terminate the operation and output the optimal result.

[0058] Further, the step S52 specifically includes the following steps:

[0059] S521. Map the optimal position:

[0060]

[0061] In the formula, b ap represents the optimal position that the entire particle swarm can currently search to; represents the upper bound of the value range of the particle position variable; represents the lower bound of the value range of the particle position variable;

[0062] S522. Use the operating equation of a typical chaotic system to perform K iterations on y t to generate a chaotic sequence

[0063] S523. Inverse-map the chaotic sequence back to the original solution space to obtain a feasible solution sequence b ap =(b ap,1 , b ap,2 ,..., b ap,K );

[0064] S524. By calculating the fitness value of each feasible solution vector in the feasible solution sequence and retaining the feasible solution vector corresponding to the optimal solution, denoted as

[0065] As can be seen from the above description, the chaos theory is introduced on the basis of the particle swarm algorithm to prevent the particle swarm from falling into local or marginal areas. Aiming at the aftereffect of the SOC constraint, the present invention proposes a modified parameter based on the probability of the SOC value to guide the moving direction of the particle swarm, so that the population can quickly and efficiently meet this constraint and find the optimal solution.

[0066] Please refer to Figure 2 , a microgrid dispatching terminal of a hybrid energy storage system, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements each step of the above-mentioned microgrid dispatching method of a hybrid energy storage system.

[0067] The above-mentioned microgrid scheduling method and terminal of a hybrid energy storage system of the present invention are applicable to the optimal scheduling of a microgrid with a hybrid energy storage system based on a chaotic particle swarm algorithm. The specific implementation manner is as follows:

[0068] Embodiment 1

[0069] Please refer to Figure 1 and Figure 3 , a microgrid scheduling method for a hybrid energy storage system, characterized by comprising the steps of:

[0070] S1. Predict the renewable energy power and load power of the next day based on the historical data of the hybrid energy storage system, and perform uncertainty analysis on the renewable energy power according to the prediction data. The renewable energy power includes photovoltaic power and wind power.

[0071] Specifically, based on the historical data of the system, predict the total photovoltaic power P pvt , total wind power P wt , and load power P lt of the next day for 24 hours, where t = 1, 2,..., 24, and perform uncertainty analysis on the renewable energy power according to the source-load prediction data, and reflect the analysis results in the output power of renewable energy such as photovoltaic and wind power.

[0072] S11. Use the normal probability distribution to simulate the wind speed prediction error distribution, and predict the total wind power of the next day.

[0073] Specifically, for the output of a wind power plant, use the normal probability distribution model to simulate the wind speed prediction error distribution. The historical wind speed error data obtained through observation can be fitted to obtain that the wind speed error basically follows a normal distribution of N(0, ) at a certain time t, and its continuous probability density function is:

[0074]

[0075] In the formula, σ wt represents the standard deviation of the wind speed error distribution, that is, it reflects the situation of different levels of prediction error. The larger the value of σ wt in the formula, the larger the peak value of the fitted curve. Considering the different interval times t, the prediction error deviation will be greater, so fitting processing is considered at adjacent times t to reduce the error.

[0076] In this embodiment, the autoregressive moving method is used to predict the wind speed error value at the next moment, and the expression is as follows:

[0077] f(v)0 = 0, N0 = 0;

[0078] f(v) t = mf(v) t-1 + nNt-1 +N t ;

[0079] where f(v) t represents the wind speed prediction error at time t, and N t represents a random variable subject to a normal distribution, which can be obtained by fitting historical error data. m and n are the weight parameters of the autoregressive moving model and can be set according to experience.

[0080] S12. Calculate the prediction error of the photovoltaic output power based on the light intensity, and predict the total photovoltaic power for the next day.

[0081] Specifically, similar to the prediction error of the wind turbine output power, the photovoltaic output power is related to the light intensity at the current location and time. Therefore, the prediction error of the photovoltaic output power can be calculated from the prediction error of the light intensity. The light intensity has obvious stages. That is, at night, the light intensity is almost zero, so the photovoltaic output power is also almost zero. During the day, the light intensity is related to the cosine value of the solar zenith angle and the clearness index. Through a large number of studies, it has been proven that the photovoltaic error at a certain time t can also be represented by a normal distribution, and its associated probability density function is:

[0082]

[0083] S13. Simulate the load prediction error using a normal distribution and predict the load power for the next day.

[0084] Specifically, generally, the distribution characteristics of the load prediction error within a certain scheduling period are generally simulated using a normal distribution. The standard deviation parameter required for the normal distribution can be estimated from empirical prediction data. Although the output powers of the wind turbine units and the photovoltaic array have a certain impact on the load, this impact is relatively small. For the simplicity of the calculation process, it can be considered that these three power predictions are independent of each other. Therefore, their error functions can all be considered to follow a normal distribution, where:

[0085]

[0086] The continuous probability density function formula of the total prediction error function is as follows:

[0087]

[0088] S14. The reliability of the micro-power generating unit can be represented by a two-state Markov model. When operating under normal conditions, the normal working time of the generating unit and the unit repair time are related to the correction rate ω i of unit i and the outage rate μ iIt is related and can be expressed by its exponential function as follows:

[0089]

[0090] A i (t) = 1 - U i (t);

[0091] Considering that the repair of photovoltaic, wind power, and micro gas turbine generator sets is relatively difficult, when the units cannot operate normally, they generally cannot be repaired or replaced within a one-week scheduling cycle. Therefore, the repair rate is ignored when calculating or optimizing the power generation strategy in a scheduling cycle or in day-ahead scheduling. Thus, in the scheduling cycle t, the failure probability and reuse rate of unit i can be simplified as:

[0092]

[0093] A i (t) = 1 - U i (t);

[0094] S2. Calculate the three-phase power unbalance degree at each moment of the predicted data without energy storage participation.

[0095] S3. Take the renewable energy power, energy storage power, load switching power, and diesel generator power as particles, and initialize the particles according to the constraint conditions to obtain a particle population.

[0096] Among them, regarding the optimization variables, on the source side, the independent microgrid is considered to include multiple types of distributed power sources such as wind power generation systems, photovoltaic power generation systems, energy-type energy storage units, power-type energy storage units, and diesel power generation systems. On the load side, the independent microgrid is considered to include three types of loads: important loads, secondary loads, and interruptible loads participating in demand-side response.

[0097] S31. Establish a set P of planned curves and a set z of operating state values for energy storage power, load switching power, and diesel generator power:

[0098] P = {P o,i , P bess,j , P pv,k , P w,l}, z = {z o,i , z bess,j , z sload,q , z cload,r};

[0099]

[0100] In the formula, P o , P bess , Ppv , P w respectively represent the output plan matrices of diesel generator sets, energy storage units, photovoltaic power generation systems, and wind power generation systems; z o , z bess respectively represent the operation status matrices of diesel generator sets and energy storage units; z sload , z cload respectively represent the switching plan matrices of secondary loads and interruptible loads participating in demand-side response; P o,i represents the power generation plan curve of the i-th diesel generator set in the next 24 hours; P bess,j represents the charge and discharge plan curve of the j-th energy storage in the next 24 hours; P pv,k represents the output plan of the k-th photovoltaic power generation system in the next 24 hours; P w,l represents the output plan of the l-th wind power generation system in the next 24 hours; z o,i represents the start-stop plan of the i-th diesel generator set in the next 24 hours; z bess,j represents the start-stop plan of the j-th energy storage in the next 24 hours; z sload,q represents the switching plan of the q-th secondary load in the next 24 hours; z cload,r represents the switching plan of the r-th interruptible load participating in demand-side response; n o , n bess , n pv , n w , n sload , n cload respectively represent the numbers of diesel generator sets, energy storage units, photovoltaic power generation systems, wind power generation systems, secondary loads, and interruptible loads participating in demand-side response in the independent microgrid.

[0101] S32. Generate the initial population of the particle swarm under the system power balance constraint, the operation constraint of the energy storage system, the operation constraint of the diesel power generation system, the operation constraint of the renewable energy power generation system, the reserve capacity constraint, and the three-phase unbalance constraint.

[0102] The constraint conditions are as follows:

[0103] 1. System balance constraint. When the microgrid is in the off-grid operation state, the output of each distributed power source in the system at each moment must be balanced with the load power in the system, as described by the following formula:

[0104]

[0105] In the formula, n iload represents the number of important loads in the system, and P iload,u (t) represents the power of the u-th important load at time t.

[0106] 2. Operational constraints of energy-type energy storage. The energy storage constraints include operating power constraints, state of charge (SOC) constraints, and power balance constraints over the scheduling period. The specific formulas are as follows:

[0107]

[0108] SOC bess,j,min ≤SOC bess,j (t)≤SOC bess,j,max ;

[0109] SOC bess,j (0)-ΔSOC ba ≤SOC bess,j (T)≤SOC bess,j (0)+ΔSOC ba ;

[0110] In the above formulas, respectively represent the rated charging and discharging powers of the j-th energy storage; SOC bess,j,max , SOC bess,j,min respectively represent the upper and lower limits of the state of charge of the j-th energy storage; SOC bess,j (0), SOC bess,j (T) respectively represent the state of charge of the j-th energy storage unit at the beginning and end of the scheduling period, and ΔSOC ba represents the maximum allowable error value of the average power of the energy storage cycle.

[0111] 3. Operational constraints of the diesel power generation system. The diesel power generation system needs to meet the operating power constraints and power ramp rate constraints:

[0112] β o,i,min P o,i,m ≤P o,i (t)≤P o,i,m ;

[0113] -ΔP o,i,down ≤P o,i (t)-P o,i (t - 1)≤ΔP o,i,up ;

[0114] In the formula, P o,i,m represents the rated power of the i-th diesel generator set; β o,i,min represents the minimum operating power coefficient of the i-th diesel generator set, and ΔP o,i,down , ΔP o,i,up respectively represent the maximum downward ramp rate and the maximum upward ramp rate of the i-th diesel generator set.

[0115] 4. Operating constraints of renewable energy power generation systems. In day-ahead optimal dispatch decisions, the output plans of wind power generation systems and photovoltaic power generation systems need to be less than the corresponding day-ahead predicted power, as described by the following formula:

[0116]

[0117] 5. Reserve capacity constraints. The reserve capacity is provided by diesel power generation systems and energy storage systems, as described by the following formula:

[0118]

[0119] In the formula, E bess,j,m represents the rated capacity of the j-th energy storage; τ bess,d represents the discharge efficiency of the energy storage; R s (t) represents the reserve capacity demand of the stand-alone microgrid.

[0120] 6. Three-phase unbalance constraints. In a microgrid system, especially in a stand-alone microgrid system, due to the access of a large number of single-phase loads and a large number of photovoltaic power sources, to a certain extent, it exacerbates the problem of three-phase current unbalance. Three-phase current unbalance will affect the normal operation of electrical equipment in the power system and increase the power loss of the power grid. Therefore, the three-phase current unbalance degree is one of the main indicators to measure the power quality of the power grid. In order to minimize the adverse effects of three-phase unbalance on the operation of the microgrid system, when performing day-ahead optimal dispatch of the microgrid system, it is necessary to meet the index requirements of the three-phase unbalance degree of the system for actual applications.

[0121]

[0122]

[0123] In the formula, P A , P B , P C respectively represent the active power of phases A, B, and C in the stand-alone microgrid system; Q A , Q B , Q C respectively represent the reactive power of phases A, B, and C in the stand-alone microgrid system; I1 and I2 respectively represent the root mean square values of the positive sequence and negative sequence components of the three-phase current.

[0124] S4. Iterate the particle population, calculate the local optimal solution of the particle population for each iteration. If the local optimal solution of the particle population in the current iteration is better than the local optimal solution of the particle population in the previous iteration, then take the local optimal solution of the particle population in the current iteration as the global optimal solution; otherwise, take the local optimal solution of the particle population in the previous iteration as the global optimal solution;

[0125] Specifically, iterate over the particle population and calculate the fitness of each particle in each iteration. The fitness includes an operating cost term G and a decision penalty term H.

[0126] The operating cost term G includes the cost of the renewable energy power generation system, the operating cost of the diesel generator system, the operating cost of the energy storage unit, and the load revenue (electricity sales revenue + interruptible load subsidy cost); the penalty term H includes the penalty term for wind and light curtailment and the penalty term for load shedding.

[0127] Since the above two types of penalty terms are introduced into the objective function, when solving the day-ahead optimal scheduling model, the system will only make decisions to curtail wind and light or shed loads at times when the relevant system security and stability constraints are not met.

[0128] In this embodiment, the optimization objective function of the stand-alone microgrid day-ahead optimal scheduling model is shown as follows:

[0129] f(P,z) = G o (P o ,z o ) + G bess (P bess ,z bess ) + G renew (P pv ,P w )

[0130] - G load (z load ) + H de (P pv ,P w ) + H lp (z load );

[0131] In the formula, G load represents the load revenue, including two parts: electricity sales revenue and interruptible load subsidy cost. G bess represents the operating cost of the energy storage unit. G o represents the operating cost of the diesel power generation system. G renew represents the operating cost of the renewable energy power generation system. H de represents the penalty term for wind and light curtailment. H lp represents the penalty term for load shedding.

[0132] Among them, the calculation formula for the cost item is as follows:

[0133]

[0134]

[0135]

[0136]

[0137] where z o,s,i (t), z o,d,i (t) respectively represent the start-stop switching variables of diesel generator set i in the t-th period. z o,s,i (t) represents that diesel generator set i switches from the shutdown state to the startup state in the t-th period. z o,d,i (t) represents that diesel generator set i switches from the startup state to the shutdown state in the t-th period. f o,s , f o,d respectively represent the startup cost and shutdown cost of the diesel generator set; f ouel represents the fuel cost function of the diesel power generation system; g ouel represents the environmental conversion cost function of the diesel power generation system. f bess,g represents the operating cost coefficient of the energy-type energy storage, with the unit of yuan / kilowatt; f bess,inv represents the initial investment cost of the energy-type energy storage; r bess,y represents the life loss ratio within the optimization scheduling duration T; f w,g represents the operating cost coefficient of the wind power generation system, f pv,g represents the operating cost coefficient of the photovoltaic power generation system; G load represents the revenue of the load during the scheduling period T; n load represents the number of loads in the independent microgrid; f load,sale represents the selling electricity price of the microgrid; f load,c represents the compensation coefficient of the interruptible load; P load,d (t) represents the power of the d-th load at the t-th moment; ΔP cload,d represents the interruptible load reduction amount of the d-th load at the t-th moment.

[0138] Among them, the penalty term consists of the wind curtailment and PV curtailment penalty term and the load power shortage penalty term, and the calculation formula is as follows:

[0139]

[0140]

[0141] where respectively represent the short-term predicted powers of PV and wind power. λ de , λ lp respectively represent the penalty coefficients of the wind curtailment and PV curtailment penalty term and the load power shortage penalty term.

[0142] S5. Determine whether the global optimal solution or the number of iterations reaches the termination condition. If so, output the output plans of the energy storage, diesel generator, wind power, and photovoltaic system, as well as the switching plan of the load, for the next day predicted in the global optimal solution. Otherwise, update the particle population based on the power three-phase unbalance degree.

[0143] S51. Update the particle velocity and position. The formulas are as follows:

[0144] v i,p (t + 1)= ε(t)v i,p (t)+ c1(t)m1[b i,p (t)- x i,p (t)]+ c2(t)m2[b ap (t)- x i,p (t)];

[0145] x i,p (t + 1)= x i,p (t)+ v i,p (t + 1);

[0146] In the formula, v i,p (t) represents the velocity of the i-th particle in the t-th generation; ε represents the inertia factor; c1 and c2 represent the acceleration factors; m1 and m2 represent random numbers obeying the uniform normal distribution; b i,p (t) represents the individual optimal position found by particle i during the search process; b ap (t) represents the globally optimal position found by the entire particle swarm during the entire search process; x i,p (t) represents the position of the i-th particle in the t-th generation;

[0147] In the system, the three-phase unbalance is generally satisfied using hard constraints. This situation easily leads to it being difficult to generate a solution set that satisfies the SOC constraint conditions, and it easily causes the algorithm to converge prematurely and fall into a local optimal solution. Therefore, to address this problem, a probability fuzzy rule particle swarm optimization algorithm based on the SOC value is proposed. This method can guide the movement direction of the energy storage charge and discharge particles, enabling the population to reach the optimal solution quickly and efficiently under the constraint conditions.

[0148] Among them, the probability fuzzy rule particle swarm method includes the following steps:

[0149] S511. The fuzzy rule membership function based on the probability fuzzy positive of the SOC value is as Figure 4 shown. In the figure, the Y coordinate represents the probability value P of the particle generating power for the energy storage during the generation process dis ; the coordinate SOC represents the SOC value after this solution calculation; the coordinate SOC vary represents the SOC change rate T generated due to the power value vary .

[0150] S512. When the energy storage SOC value is too low, the particle population tends to the energy storage charging direction. At the same time, the probability fuzzy rule correction method will consider the SOC change rate T caused by the solution set at this time va , when the change rate T va is relatively large, slightly reduce the P dis value. When the change rate T va is relatively small, slightly increase the P dis value. This method can reduce the fluctuation of the energy storage SOC value.

[0151]

[0152] In the formula, κ represents the energy storage probability adjustment coefficient, taking 0.2; when T va >0, the energy storage behavior trend is charging at this time; T va <0 indicates that the energy storage tends to the discharging state at this time.

[0153] S513. When the energy storage SOC is within the set range, according to the change rate T va trend, keep the energy storage SOC as flat as possible: when T va >0, Prob tends to the discharging side, and at this time P dis is greater than 0.5. When T va <0, P dis tends to the charging side, that is, P dis is greater than 0.5;

[0154] S514. When the energy storage SOC value is too high, the direction generated by the particle swarm will tend to the energy storage discharging direction. At the same time, this method will consider the SOC change rate T in the solution set at this time va , when the change rate T va is relatively large, slightly reduce the Prob value.

[0155] S52. Optimize the optimal position b of the particle swarm using the chaos rule ap , and calculate the feasible solution vector corresponding to the optimal solution

[0156] S521. Map the optimal position:

[0157]

[0158] In the formula, b ap represents the optimal position that the entire particle swarm can search to currently; represents the upper bound of the value range of the particle position variable; represents the lower bound of the value range of the particle position variable;

[0159] S522. Use the running equation of a typical chaotic system for yt Perform K iterations to generate a chaotic sequence

[0160]

[0161] S523. Perform an inverse mapping on the chaotic sequence back to the original solution space to obtain a feasible solution sequence b of the chaotic variables ap =(b ap,1 , b ap,2 ,..., b ap,K );

[0162] S524. By calculating the fitness value of each feasible solution vector in the feasible solution sequence and retaining the feasible solution vector corresponding to the optimal solution, denoted as

[0163] S53. Randomly select a particle in the current subgroup and replace the position vector of the selected particle with the position vector of the feasible solution vector

[0164] S54. When the calculated fitness variance is less than the given preset variance value, re-initialize the current population and return to step S51; where the method for calculating the population fitness variance is

[0165]

[0166]

[0167] In the formula, f g , f i , respectively represent the normalization factor, fitness value, and average fitness value of the current particle population of particle i

[0168] S55. Determine whether the current operation meets the stop condition. If so, terminate the operation and output the optimal result

[0169] Embodiment 2

[0170] Please refer to Figure 2 , a microgrid dispatching terminal 1 of a hybrid energy storage system, including a memory 2, a processor 3, and a computer program stored on the memory 2 and executable on the processor 3. When the processor 3 executes the computer program, it implements each step of the microgrid dispatching method of a hybrid energy storage system in the above embodiment

[0171] In summary, the microgrid scheduling method and terminal of the hybrid energy storage system provided by the present invention conduct research on energy optimization scheduling for the slow dynamic component in the power generation and consumption energy deviation of the independent microgrid system, including proposing a prediction model based on historical data for source-load data, adding source-load uncertainty to the prediction results and calculating the relevant three-phase unbalance, researching the establishment and solution method of the day-ahead optimization scheduling mathematical model with multiple energy storages, which can improve the scheduling accuracy of the microgrid of the hybrid energy storage system. During the optimization scheduling process, on the basis of the particle swarm, it is avoided that the particle swarm falls into local or marginal areas; by guiding the moving direction of the particle swarm, the population can quickly and efficiently meet this constraint and find the optimal solution, thereby improving the scheduling calculation efficiency.

[0172] Among them, since the unbalanced power interfering with the system stability is non-linear, improved empirical mode decomposition is used for its adaptive decomposition, and the energy value of the aliased part is calculated by summing the absolute values of adjacent IMF components to determine the frequency division frequency, so as to reduce the aliasing phenomenon between frequencies and make the initial frequency allocation more accurate and reliable; then, aiming at the problem that the large power density of the power-type energy storage is likely to cause the SOC to cross the line, a fuzzy controller is designed to perform secondary optimization on the initial power command of the hybrid energy storage system, reduce the possibility of the power-type energy storage reaching the upper and lower limits, and realize the optimal allocation of the hybrid energy storage power. Compared with the traditional hybrid energy storage power allocation method, the allocation of the unbalanced power of the hybrid energy storage system in this embodiment is more accurate and reliable, not only bringing out the best performance of the hybrid energy storage system, thus quickly suppressing the power fluctuation of the hybrid energy storage system, but also reducing the loss of the energy-type energy storage and prolonging its service life.

[0173] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in the relevant technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A microgrid scheduling method for a hybrid energy storage system, characterized in that, Including the steps: S1. Predict the renewable energy power and load power for the next day based on the historical data of the hybrid energy storage system, and conduct uncertainty analysis on the renewable energy power according to the prediction data, where the renewable energy power includes photovoltaic power and wind power; S2. Calculate the power three-phase unbalance degree at each moment without energy storage participation for the prediction data; S3. Take the renewable energy power, energy storage power, load switching power, and diesel generator power as particles, and initialize the particles according to the constraint conditions to obtain a particle population; S4. Iterate the particle population, calculate the local optimal solution of the particle population for each iteration. If the local optimal solution of the particle population in the current iteration is better than that in the previous iteration, then take the local optimal solution of the particle population in the current iteration as the global optimal solution; otherwise, take the local optimal solution of the particle population in the previous iteration as the global optimal solution; S5. Judge whether the global optimal solution or the number of iterations reaches the termination condition. If so, output the output plans of the energy storage, diesel generator, wind power, photovoltaic system, and the load switching plan for the next day predicted in the global optimal solution; otherwise, update the particle population based on the power three-phase unbalance degree; The specific step S5 Includes the following steps: S51. Update the particle velocity and position, and the formula is as follows: ； ； Wherein, represents the velocity of the li-th particle in the t-th generation; represents the inertia factor; c1 and c2 represent the acceleration factors; m1 and m2 represent random numbers subject to a uniform normal distribution; represents the individual optimal position found by the particle li during the search process; represents the optimal position in the global state found by the entire particle swarm during the entire search process; represents the position of the li-th particle in the t-th generation; S52. Optimize the optimal position of the particle swarm using the chaos rule, and calculate the feasible solution vector corresponding to the optimal solution; S53. Randomly select a particle in the current subgroup, and replace the position vector of the selected particle with the position vector of the feasible solution vector; S54. When the calculated fitness variance is less than the given preset variance value, re-initialize the current population. The calculation method of the population fitness variance is: ； ； where f g , f li , and represent the normalization factor, fitness value, and average fitness value of the current particle population of particle li, respectively; S55. Judge whether the current operation meets the stop condition. If it meets, terminate the operation and output the optimal result.

2. The microgrid scheduling method for a hybrid energy storage system according to claim 1, characterized in that The prediction of the renewable energy power and load power for the next day based on the historical data of the hybrid energy storage system includes: S11. Use the normal probability distribution to simulate the wind speed prediction error distribution, and predict the total wind power for the next day; S12. Calculate the prediction error of the photovoltaic output power through the light intensity, and predict the total photovoltaic power for the next day; S13. Use the normal distribution to simulate the load prediction error, and predict the load power for the next day.

3. The microgrid scheduling method for a hybrid energy storage system according to claim 1, characterized in that The taking the renewable energy power, energy storage power, load switching power, and diesel generator power as particles includes: Establish a set P of planned curves and a set z of operating state values for the energy storage power, load switching power, and diesel generator power; P = {P o,i , P bess,j , P pv,k , P w,l}, z = {z o,i , z bess,j , z sload,q , z cload,r}; ； Where, P o 、P bess 、P pv 、P w respectively represent the output plan matrices of the diesel generator set, the energy storage unit, the photovoltaic power generation system, and the wind power generation system; z o 、z bess respectively represent the operation state matrices of the diesel generator set and the energy storage unit; z sload 、z cload respectively represent the switching plan matrices of the secondary load and the interruptible load participating in the demand-side response; P o,i represents the power generation plan curve of the i-th diesel generator set in the next 24 hours; P bess,j represents the charge and discharge plan curve of the j-th energy storage in the next 24 hours; P pv,k represents the output plan of the k-th photovoltaic power generation system in the next 24 hours; P w,l represents the output plan of the l-th wind power generation system in the next 24 hours; z o,i represents the start-stop plan of the i-th diesel generator set in the next 24 hours; z bess,j represents the start-stop plan of the j-th energy storage in the next 24 hours; z sload,q represents the switching plan of the q-th secondary load in the next 24 hours; z cload,r represents the switching plan of the r-th interruptible load participating in the demand-side response; n o 、n bess 、n pv 、n w 、n sload 、n cload respectively represent the quantities of the diesel generator set, the energy storage unit, the photovoltaic power generation system, the wind power generation system, the secondary load, and the interruptible load participating in the demand-side response in the independent microgrid.

4. A microgrid scheduling method for a hybrid energy storage system according to claim 1, characterized in that, The initialization of the particles according to the constraint conditions includes: Under the system power balance constraint, the operation constraint of the energy-type energy storage system, the operation constraint of the diesel power generation system, the operation constraint of the renewable energy power generation system, the reserve capacity constraint, and the three-phase unbalance constraint, generate the initial population of the particle swarm.

5. A microgrid scheduling method for a hybrid energy storage system according to claim 1, characterized in that, After the iteration of the particle population includes: Calculate the fitness of each particle, and the fitness includes an operating cost item and a decision penalty item; The operating cost items include the cost of the renewable energy power generation system, the operating cost of the diesel generator system, the operating cost of the energy storage unit, and the load revenue; The penalty items include the penalty item for wind and light curtailment and the penalty item for load loss.

6. The microgrid scheduling method for a hybrid energy storage system according to claim 1, characterized in that, The specific steps of step S52 include the following steps: S521. Map the optimal location: ； In the formula, represents the optimal position that the entire particle swarm can currently search to; represents the upper bound of the value range of the particle position variable; represents the lower bound of the value range of the particle position variable; S522. Using the operating equation of a typical chaotic system, perform K iterations to generate a chaotic sequence ; S523. Inverse map the chaotic sequence back to the original solution space to obtain a feasible solution sequence of chaotic variables ; S524. By calculating the fitness value of each feasible solution vector in the feasible solution sequence and retaining the feasible solution vector corresponding to the optimal solution, denoted as .

7. A microgrid dispatching terminal of a hybrid energy storage system, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the microgrid scheduling method of a hybrid energy storage system according to any one of claims 1-6 above.

Images