Generator set load optimal distribution method based on epidemic virus propagation algorithm

Abstract

The invention discloses a generator set load optimal distribution method based on an epidemic virus propagation algorithm, and the method specifically comprises the steps: building a scheduling model with the lowest unit operation cost as a target, and forming a system operation constraint condition; according to an epidemic virus propagation algorithm, solving an optimal solution and a worst solution of each operation cost under the condition of meeting operation constraint conditions of equality and inequality of the power system; and then a comprehensive sorting index is adopted to sort a large number of obtained Pareto optimal solutions, and the highest composite sorting index is used as an optimal scheme after multiple costs are weighed. According to the generator set load optimal distribution method, intelligent optimization of power load distribution is realized by simulating a virus transmission mechanism, the utilization rate of various heterogeneous power supplies can be effectively improved, and a power system can better accept wind power, photovoltaic power, hydroelectric power and other clean energy through optimal distribution of loads, so that the power generation efficiency is improved. And the method has important significance for assisting in realizing the dual-carbon target.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy-saving power generation optimization dispatching of electric power systems, and in particular relates to a method for optimizing load distribution of generator sets based on an epidemic virus propagation algorithm. Background Art

[0002] Under the background of dual carbon goals and clean and low-carbon energy transformation, the energy-saving power generation optimization dispatching technology of the power system faces new challenges and opportunities. Energy conservation and emission reduction is a feasible way. Considering the significant differences in energy conservation and emission reduction among various power generation technologies, the power system is required to maximize the energy utilization efficiency during the dispatching process. In addition, the reasonable allocation of loads of each unit can significantly improve the flexibility of system operation, ensure load demand, reduce energy consumption, and improve power generation efficiency, thereby achieving energy conservation, emission reduction and maximizing economic benefits. With the promotion of clean and low-carbon energy transformation reform, large-scale renewable energy is connected to the grid, and wind and photovoltaic power generation have strong volatility and uncontrollability, which makes the optimization allocation of loads and energy-saving power generation dispatching problems more complicated. Therefore, the power system needs to optimize the coordinated operation of traditional power generation and renewable energy through advanced dispatching algorithms and real-time response mechanisms, and maximize the absorption of clean energy such as wind and solar energy.

[0003] With the large-scale uncertain renewable energy connected to the power system, the number and types of units will increase, and the corresponding variables and constraints will increase accordingly. The energy-saving power generation optimization scheduling model will become increasingly complex and face a serious dilemma of dimensionality curse. Although traditional power system optimization methods, such as dynamic programming, mixed integer linear programming, Lagrangian relaxation method, etc., are widely used in energy-saving power generation scheduling, they often face problems such as slow convergence speed, strong parameter sensitivity, and easy to fall into local optimal solutions when dealing with large-scale, multi-objective and complex nonlinear problems. And when facing a large number of uncertain factors in modern new power systems, they show obvious lack of flexibility and adaptability, and the effect is limited. In recent years, intelligent optimization algorithms (such as particle swarm optimization, genetic algorithm, ant colony algorithm, bee colony optimization algorithm, sparrow search algorithm, etc.) have been widely used in power system load optimization. They achieve global optimization by simulating biological behaviors in nature (such as animal foraging, genetic inheritance, etc.), and can handle nonlinear, multi-objective complex optimization problems, which can overcome the defects of traditional optimization methods to a certain extent. However, these algorithms still have shortcomings in solution efficiency and global search capabilities. Summary of the invention

[0004] The purpose of the present invention is to provide a method for optimizing load distribution of generator sets based on a popular virus propagation algorithm, which can effectively balance local and global search capabilities and improve the accuracy and efficiency of load distribution.

[0005] The technical solution adopted by the present invention is a method for optimizing the load distribution of a generator set based on an epidemic virus propagation algorithm, which is specifically implemented according to the following steps:

[0006] Step 1: Establish a dispatch model with the goal of minimizing the unit operation cost. The cost includes the coal cost, start-up and shutdown cost, wind and solar power abandonment penalty cost, and environmental emission penalty cost of each generating unit, and refine the dispatch model from the perspective of system operation constraints;

[0007] Step 2: After step 1, according to the epidemic virus propagation algorithm, under the constraints of the power system equation and inequality operation constraints, the optimal solution and the worst solution of each operation cost are solved; then a comprehensive sorting index is used to sort a large number of Pareto optimal solutions, and the highest composite sorting index is used as the best solution after weighing multiple costs.

[0008] The present invention is also characterized in that:

[0009] In step 1, the objective function is to minimize the total operating cost of the unit, as shown in formula (1):

[0010]

[0011] Where: F represents the total operating cost of each unit in the power generation system; is the coal burning cost of thermal power units; is the start-up and shutdown cost of thermal power units; Penalty costs for curtailing wind and solar power; Penalty costs for the environment;

[0012] The coal burning cost of thermal power units is shown in formula (2):

[0013]

[0014] Where: P Gi,t is the active power output of thermal power unit i at time t; a i 、b i 、c i is the coal cost coefficient corresponding to thermal power unit i; N G is the total number of thermal power units; e i and h i is the valve point effect cost coefficient of thermal power unit i; P Gi,min is the minimum output power of thermal power unit i; T represents the 24h power generation scheduling time series;

[0015] The start-up and shutdown cost of the unit is shown in formula (3):

[0016]

[0017] Among them, S i,tis the operating state variable of thermal power unit i at time t; F i,t The cost incurred during the start and stop of thermal power unit i during period t;

[0018] The penalty cost for wind and solar power abandonment is shown in formula (4):

[0019]

[0020] in: N is the penalty cost coefficient for wind and solar power abandonment; W 、N Pv is the total number of wind farms and photovoltaic power stations; is the actual output power of wind power and photovoltaic generator set i at time t; is the predicted output of wind power and photovoltaic generator set i at time t;

[0021] The environmental emission penalty cost function is shown in equation (5) and equation (6);

[0022]

[0023] The environmental emission penalty function is shown in formula (7):

[0024]

[0025] Where: γ represents the environmental penalty coefficient; is the penalty variable of the system at time t; is the carbon emission function of thermal power units at time t; E max is the maximum carbon emission allowed for thermal power units; i 、m i 、n i Represents the carbon emission coefficient of thermal power unit i.

[0026] The power system operation constraints include system power balance constraints, spinning reserve constraints, output power constraints of each unit, thermal power unit ramping and start-stop time constraints, hydropower flow and storage capacity constraints, pumped storage operation and energy balance constraints;

[0027] The system power balance constraint is shown in equation (8):

[0028]

[0029] Where: N H is the total number of hydropower units; N PH is the total number of pumped storage units; P Hi,t is the output power of the i-th hydropower unit at time t; is the output power of the i-th pumped storage unit at time t; u i,tis the operating state variable of the i-th pumped storage unit at time t; P D,t is the load demand of the system at time t;

[0030] The positive and negative spinning reserve constraints of the system are shown in formula (9):

[0031]

[0032] Where: P Gi,max , P Gi,min is the maximum and minimum output power of thermal power unit i; P Hi,max , P Hi,min is the maximum and minimum output power of hydropower unit i; is the maximum and minimum power generation of pumped storage unit i; is the spinning reserve required by the system at time t;

[0033] The output power constraints of each unit are shown in formula (10);

[0034]

[0035] in: is the maximum and minimum pumping power of pumped storage unit i; is the maximum and minimum output power of wind turbine i; is the maximum and minimum output power of photovoltaic generator set i;

[0036] The climbing constraint of thermal power unit is shown in formula (11):

[0037] -R Gi,down Δt≤P Gi,t+1 -P Gi,t ≤R Gi,up Δt (11);

[0038] Where: R Gi,down , R Gi,up It indicates the rate of increase and decrease of the output power of the thermal power unit; Δt is the interval time;

[0039] The minimum start-up and shutdown time constraint of thermal power units is shown in formula (12):

[0040]

[0041] in: is the number of hours that thermal power unit i is continuously turned on and off during the t-1 period; is the minimum start-up and shutdown time of thermal power unit i;

[0042] The nonlinear relationship between hydropower generation power, turbine flow and reservoir capacity is expressed by a binary quadratic function, as shown in equations (13) and (14):

[0043] P Hi,t =f H (V t ,Q Hi,t ) (13);

[0044]

[0045] Where: α1, α2, α3, α4, α5, α6 are hydropower generation coefficients; V t is the storage capacity of the upstream reservoir of the hydropower station during period t; Q Hi,t is the power generation reference flow of the hydropower unit during period t;

[0046] The water flow constraint used when the hydropower unit generates electricity is shown in formula (15):

[0047] Q Hi，min ≤Q Hi,t ≤Q Hi，max (15);

[0048] Where: Q Hi,min , Q Hi,max They are the maximum and minimum flow rates allowed for hydroelectric generating units, respectively;

[0049] The reservoir capacity constraint of the hydropower station is shown in formula (16):

[0050] V t+1 =V t +V in -Q Hi,t -SP t (16);

[0051] Where: V in Indicates the amount of water flowing into the upstream reservoir of the hydropower station in each scheduling period; SP t It represents the amount of water evaporated or overflowed from the reservoir during period t;

[0052] Pumped storage unit operating condition constraints: Pumped storage units cannot perform both power generation and pumping at the same time, as shown in formula (17):

[0053] P gi,t ×P pi,t =0(17);

[0054] Where: P gi,t represents the power generation of pumped storage unit i in period t; P pi,t represents the pumping power of pumped storage unit i in period t;

[0055] The reservoir energy balance constraint is shown in equations (18) and (19):

[0056]

[0057] 0≤E t ≤E t,max (19);

[0058] Where: E t Represents the reservoir capacity; E t，max Represents the maximum storage capacity of the reservoir; η g , η p They represent the power generation efficiency and pumping efficiency of the pumped storage unit respectively.

[0059] In step 2, specifically:

[0060] Step 2.1, by simulating the process of the epidemic virus spreading group gradually changing from susceptible to herd immunity, and mapping these states to the solution space of the optimization problem, the complex optimization problem is iteratively optimized; the specific steps are:

[0061] Step 2.1.1, parameter initialization stage, set the dimension of the search space to D, which is the number of independent variables; the size of the population is N Pop ; Initial number of iterations k = 0, maximum number of iterations I ter,max ; Mutation rate M R ; V min 、V max is the minimum and maximum value range of the variable in the objective function;

[0062] Step 2.1.2, randomly initialize the solution population, each solution represents the optimal output plan of each unit in the scheduling period, that is, the optimal distribution of load; evaluate each solution, and arrange all solutions in ascending order according to the objective function, the first solution is the optimal solution, and the last solution is the worst solution;

[0063] Step 2.1.3, based on the viral replication stage of frameshift technology, new proteins in the population are generated by frameshift technology to form a new optimal solution set;

[0064] Step 2.1.4, apply the distance crossover operator to simulate immune behavior, and arbitrarily select gene fragments from the optimal immune individuals for crossover mutation to generate new mutation solutions;

[0065] Step 2.1.5, evaluate the objective function value of the new solution after mutation and update the population;

[0066] Step 2.1.6, repeat steps 2.1.2-2.1.5 for the obtained new population until the termination condition of the maximum number of iterations is reached;

[0067] Step 2.1.7, output the optimal solution and the worst solution, and obtain the output plan of each generator set under the conditions of the lowest and highest unit operating cost;

[0068] Step 2.2, design comprehensive ranking indicators, further optimize each generator set under the condition of lowest unit operating cost obtained in step 2.1, and form the best load distribution plan; specifically:

[0069] Step 2.2.1, for the problem of minimizing the unit operating cost, the membership function is shown in formula (22):

[0070]

[0071] Then the average fuzzy membership value can be expressed as formula (23):

[0072]

[0073] Where: μ m,n is the membership value of the mth allocation scheme under the nth cost function; F m,n is the mth solution of the nth cost function; F n,max 、F n,min is the optimal solution and the worst solution of the nth cost function; N is the total number of cost functions; w n is the weight factor of the objective function, and ∑w n =1;

[0074] Step 2.2.2, based on the ranking index of the TOPSIS method, each objective function value is first normalized, and then the normalized value is multiplied by the corresponding weight to obtain the normalized value, as shown in formula (24):

[0075]

[0076] Among them: m,n Normalized value for the mth allocation scheme under the nth cost function;

[0077] According to the optimal and worst solutions obtained by the popular virus propagation algorithm, the optimal solution is defined as The worst solution is defined as Calculate the Euclidean distance between each allocation scheme and the optimal solution and the worst solution, as shown in formula (25):

[0078]

[0079] in: is the Euclidean distance between the mth allocation scheme and the optimal solution and the worst solution;

[0080] The closeness of each allocation scheme to the optimal solution is calculated. The higher the closeness, the more obvious the superiority of the scheme. The optimal load allocation scheme can be determined by sorting, as shown in formula (26):

[0081]

[0082] Step 2.2.3, since the optimization performance of each indicator is different, the optimal load allocation scheme for the mth load is finally determined through the composite indicator, as shown in formula (27):

[0083]

[0084] According to the ranking results of the composite index, the highest composite ranking index is used as the optimal load distribution plan that weighs multiple costs; if the result after the ranking index is the same as the optimal solution obtained by the algorithm in step 2.1, it is retained, otherwise it is updated to the optimal solution after sorting.

[0085] In step 2.1.3, the specific process is:

[0086] Using the +1 shifting technique, the value of the initialized optimal load distribution solution is shifted to the right by 1, and the value of the first position is set to [V min ,V max ] range, the specific process is shown in formula (20):

[0087]

[0088] Where: Z k represents the current optimal solution generated by the kth generation, X is the initial optimal solution, and D is the dimension of the optimization problem.

[0089] In step 2.1.4, the specific process is:

[0090] Apply the distance crossover operator to simulate immune behavior for generating new optimal solution sequences, select gene fragments from the optimal immune individuals for crossover mutation, and evaluate the fitness of the new mutation solution;

[0091] The distance crossover operator is divided according to the social distance between individuals into infected individuals, susceptible individuals and optimal immune individuals, which correspond to the following formula (21):

[0092]

[0093] Where: X i (t) is the solution before crossover mutation; X i (t+1) is the solution after crossover mutation; r is [V min ,V max ]A random value in the range; X i,bestis the optimal position found for the current i-th individual; X i,a1 , X i,a2 , X i,a3 is the position of the i-th individual in the search space; R is the distance parameter.

[0094] In step 2.1.5, the specific process is:

[0095] Compare the solution obtained after the mutation in step 2.1.4 with the optimal solution sequence obtained in step 2.1.3. If the value is smaller, it means that the optimization ability is stronger, that is, the operating cost of each unit in the power generation system is lower, then the original optimal solution sequence in step 2.1.3 is replaced; otherwise, the individual optimization ability after mutation is weaker, that is, the operating cost of each unit in the power generation system is increased, then the original optimal solution sequence in step 2.1.3 is retained.

[0096] The beneficial effects of the present invention are as follows: the load optimization distribution method of the generator set of the present invention, firstly, realizes the intelligent optimization of the power load distribution by simulating the propagation mechanism of the virus, and can effectively improve the utilization rate of various heterogeneous power sources. Secondly, it has outstanding performance in reducing the operating cost of the unit. By optimizing the output power of each generator set, the output power is guaranteed to be stable and safe when the proportion of new energy is increasing. At the same time, it is combined with energy storage technology, and through flexible power generation and flexible power use, instability is turned into controllable flexibility, and renewable energy is used more efficiently, thereby improving the overall economy. In addition, reasonable load distribution can reduce the risk of power outages and ensure the stability and reliability of the power system under different operating conditions. Especially in the face of emergencies or load fluctuations, the dynamic optimization capability based on the virus algorithm enables the system to adjust quickly and effectively respond to various challenges. Finally, the method also helps to promote the integration of renewable energy. Through the optimal distribution of loads, the power system can better accept clean energy such as wind power, photovoltaics, and hydropower, which is of great significance to helping achieve the dual carbon goals. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Figure 1 is a flow chart of a generator load optimization allocation model of a popular virus propagation algorithm of the present invention;

[0098] Figure 2 It is a data graph of load, wind power and photovoltaic output power on a typical day;

[0099] Figure 3 It is the optimal allocation scheme for the load of a typical daily energy-saving power generation system. DETAILED DESCRIPTION

[0100] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0101] Example 1

[0102] The method for optimizing the load distribution of a generator set based on the popular virus propagation algorithm of the present invention is specifically implemented according to the following steps:

[0103] Step 1: Establish a dispatch model with the goal of minimizing the total operating cost of various types of generator sets. The costs include the coal cost, start-up and shutdown costs, wind and solar power abandonment penalty costs, and environmental emission penalty costs of each generator set, and refine the dispatch model from the perspective of system operation constraints;

[0104] Specifically, the objective function is to minimize the total operating cost of the unit, which includes the coal cost, start-up and shutdown cost, environmental emission penalty cost, and wind and solar power abandonment penalty cost of thermal power units;

[0105] The total objective function of minimizing the unit operating cost is composed of multiple interrelated and mutually constrained single cost objectives, as shown in formula (1):

[0106]

[0107] Where: F represents the total operating cost of each unit in the power generation system; is the coal burning cost of thermal power units; is the start-up and shutdown cost of thermal power units; Penalty costs for curtailing wind and solar power; Penalty costs for the environment;

[0108] There is a nonlinear quadratic function relationship between the coal cost of a thermal power unit and the unit output, which is used to describe the fuel consumption of a thermal power unit under different load conditions. When the valve point effect of a thermal power unit is considered, the total coal cost is shown in formula (2):

[0109]

[0110] Where: P Gi,t is the active power output of thermal power unit i at time t; a i , b i 、c i is the coal cost coefficient corresponding to thermal power unit i; N G is the total number of thermal power units; e i and h i is the valve point effect cost coefficient of thermal power unit i; P Gi,min is the minimum output power of thermal power unit i; T represents the 24h power generation scheduling time series;

[0111] Relying on the flexibility of each generator set to adjust the volatility of renewable energy will inevitably lead to frequent adjustment of the units. Hydropower units and pumped storage units have faster adjustment speeds, and it only takes a few minutes from no load to full load, while the scheduling period interval of the present invention is 1 hour, so their start-up and shutdown costs can be ignored. Therefore, the start-up and shutdown costs of the units are mainly caused by the frequent start-up and shutdown adjustments of thermal power units, as shown in formula (3):

[0112]

[0113] Among them, S i,t is the operating state variable of thermal power unit i at time t (S i,t =1, the thermal power unit is turned on or off; S i,t =0, the thermal power unit has not started or stopped); F i,t It is the cost incurred during the start and stop of thermal power unit i during period t.

[0114] In order to maximize the use of wind power and photovoltaic power to improve the low-carbon economy of energy-saving power generation systems, the penalty cost of wind and solar power abandonment is introduced to restrict the abandonment of wind and solar power. The penalty cost of wind and solar power abandonment is shown in formula (4):

[0115]

[0116] in: N is the penalty cost coefficient for wind and solar power abandonment; W 、N Pv is the total number of wind farms and photovoltaic power stations; is the actual output power of wind power and photovoltaic generator set i at time t; is the predicted output of wind power and photovoltaic generator set i at time t.

[0117] It should be noted that the present invention only considers the situation when wind power and photovoltaic power generation are underestimated. Because when wind power is overestimated, the main task of energy-saving power generation scheduling is to make up for the power gap by increasing the output of conventional thermal power units and hydropower units, which will generate additional operating costs for the units, which is contrary to our objective function of minimizing the operating costs of the units. Therefore, the present invention does not consider the penalty costs when wind power and photovoltaic power are overestimated.

[0118] Renewable energy, conventional hydropower and pumped storage are all clean energy sources, and power generation will not pollute the environment. Pollutant emissions from the power system mainly come from the combustion of fossil fuels in thermal power units. In order to mitigate the greenhouse effect and achieve the "dual carbon" goal, it is crucial to reduce pollutant emissions from thermal power units. When pollutant emissions exceed the limit value, the penalty function will be activated. The environmental emission penalty cost function is shown in formula (5);

[0119]

[0120] The environmental emission penalty function can also be approximately fitted into a quadratic function curve, as shown in the following formula (7):

[0121]

[0122] Where: γ represents the environmental penalty coefficient; is the penalty variable of the system at time t; is the carbon emission function of thermal power units at time t; E max is the maximum carbon emission allowed for thermal power units; i 、m i 、n i Represents the carbon emission coefficient of thermal power unit i.

[0123] The operating constraints of the power system include system power balance constraints, spinning reserve constraints, output power constraints of each unit, ramping and start-stop time constraints of thermal power units, flow and storage capacity constraints of hydropower, and pumped storage operation and energy balance constraints.

[0124] The system power balance constraint is shown in equation (8):

[0125]

[0126] Where: N H is the total number of hydropower units; N PH is the total number of pumped storage units; P Hi,t is the output power of the i-th hydropower unit at time t; is the output power of the i-th pumped storage unit at time t; u i,t is the operating state variable of the i-th pumped storage unit at time t (1 means power generation, -1 means pumping, and 0 means the unit is idle); P D,t is the load demand of the system at time t.

[0127] The positive and negative spinning reserve constraints of the system are shown in formula (9):

[0128]

[0129] Where: P Gi,max , P Gi,min is the maximum and minimum output power of thermal power unit i; P Hi,max , P Hi,min is the maximum and minimum output power of hydropower unit i; is the maximum and minimum power generation of pumped storage unit i; is the spinning reserve required by the system at time t.

[0130] The output power constraints of each unit are shown in formula (10);

[0131]

[0132] in: is the maximum and minimum pumping power of pumped storage unit i; is the maximum and minimum output power of wind turbine i; is the maximum and minimum output power of photovoltaic generator set i;

[0133] The climbing constraint of thermal power unit is shown in formula (11):

[0134] -R Gi,down Δt≤P Gi,t+1 -P Gi,t ≤R Gi,up Δt (11);

[0135] Where: R Gi,down , R Gi,up It indicates the rate of increase and decrease of the output power of the thermal power unit; Δt is the interval time.

[0136] The minimum start-up and shutdown time constraint of thermal power units is shown in formula (12):

[0137]

[0138] in: is the number of hours that thermal power unit i is continuously turned on and off during the t-1 period; is the minimum startup and shutdown time of thermal power unit i.

[0139] The nonlinear relationship between the power generation of conventional hydropower and the turbine flow and reservoir capacity is expressed by a binary quadratic function, as shown in equations (13) and (14):

[0140] P Hi,t =f H (V t ,Q Hi,t ) (13);

[0141]

[0142] Where: α1, α2, α3, α4, α5, α6 are hydropower generation coefficients; V t is the storage capacity of the upstream reservoir of the hydropower station during period t; Q Hi,t is the power generation reference flow of the hydropower unit during period t.

[0143] The water flow constraint used when the hydropower unit generates electricity is shown in formula (15):

[0144] Q Hi，min ≤QHi,t ≤Q Hi，max (15);

[0145] Where: Q Hi,min , Q Hi,max They are respectively the maximum and minimum flow rates allowed for hydropower generation.

[0146] The reservoir capacity constraint of the hydropower station is shown in formula (16):

[0147] V t+1 =V t +V in -Q Hi,t -SP t (16);

[0148] Where: V in Indicates the amount of water flowing into the upstream reservoir of the hydropower station in each scheduling period; SP t It represents the amount of water evaporated or overflowed from the reservoir during period t;

[0149] Pumped storage unit operating condition constraints: Pumped storage units cannot perform both power generation and pumping at the same time, as shown in formula (17):

[0150] P gi,t ×P pi,t =0(17);

[0151] Where: P gi,t represents the power generation of pumped storage unit i in period t; P pi,t It represents the pumping power of pumped storage unit i in period t.

[0152] The reservoir energy balance constraint is shown in equations (18) and (19):

[0153]

[0154] 0≤E t ≤E t,max (19);

[0155] Where: E t Represents the reservoir capacity; E t，max Represents the maximum storage capacity of the reservoir; η g , η p They represent the power generation efficiency and pumping efficiency of the pumped storage unit respectively.

[0156] Step 2: After step 1, according to the epidemic virus propagation algorithm, under the constraints of the power system equation and inequality operation constraints, the optimal solution and the worst solution of each operation cost are solved; then a comprehensive sorting index is used to sort a large number of Pareto optimal solutions, and the highest composite sorting index is used as the best solution after weighing multiple costs, such as Figure 1 As shown, the details are as follows:

[0157] Step 2.1, the epidemic virus propagation algorithm is an optimization algorithm that simulates the virus propagation and infection process. This optimization algorithm draws on the three different states of virus propagation, infection and immunity in the group. It simulates the process of the virus propagation group gradually changing from susceptible to group immunity, and maps these states to the solution space of the optimization problem, and then iterates and optimizes the complex optimization problem. The specific steps are:

[0158] Step 2.1.1, parameter initialization stage, set the dimension of the search space to D, which is the number of independent variables; the size of the population is N Pop ; Initial number of iterations k = 0, maximum number of iterations I ter,max , as the criterion for the end of algorithm iteration; mutation rate M R ; V min 、V max is the minimum and maximum value range of the variable in the objective function.

[0159] Step 2.1.2, randomly initialize the solution population, each solution represents the optimal output plan of each unit in the scheduling period, that is, the optimal distribution of load. Evaluate each solution, and then sort all solutions in ascending order according to the objective function, then the first solution is the optimal solution, and the last solution is the worst solution.

[0160] Step 2.1.3, based on the virus replication stage of frameshift technology, new proteins in the population are generated by frameshift technology to form a new optimal solution set; the specific process is:

[0161] The code shifting technique can optimize the position of each solution in the solution space by shifting the code of each solution, so that the solution can move more effectively towards the high-quality area. This process helps to improve the fitness of the solution, thereby accelerating the convergence and optimization process of the algorithm.

[0162] Specifically, each encoded bit (or feature) in the solution is shifted in a certain direction, i.e., the value is increased or decreased. During the shifting process, it is ensured that the encoding still meets the constraints of the variable.

[0163] Using the +1 shifting technique, the value of the initialized optimal load distribution solution is shifted to the right by 1, and the value of the first position is set to [V min ,V max ] range, the specific process is shown in formula (20):

[0164]

[0165] Where: Z k represents the kth generated current optimal solution, X is the initial optimal solution, and D is the dimension of the optimization problem (the number of variables in the solution). The result of the code shift represents a new current optimal solution sequence.

[0166] Step 2.1.4, apply the distance crossover operator to simulate immune behavior, and select gene fragments from the optimal immune individuals for crossover mutation to generate new mutation solutions. The specific process is:

[0167] The distance crossover operator is applied to generate new optimal solution sequences to simulate immune behavior, and gene fragments are selected from the optimal immune individuals for crossover mutation, and the fitness of the new mutation solution is evaluated. This can effectively improve the global search ability of the algorithm and enhance the diversity of solutions, thereby better coping with complex optimization problems.

[0168] When crossing, the distance between individuals is considered, and crossover is performed only when the distance between two individuals is less than a certain threshold. This mechanism can prevent the generation of overly similar solutions and increase the diversity of solutions. The distance crossover operator can be divided into infected individuals, susceptible individuals, and optimal immune individuals according to the social distance between individuals, corresponding to the following equations (21):

[0169]

[0170] Where: X i (t) is the solution before crossover mutation; X i (t+1) is the solution after crossover mutation; r is [V min ,V max ]A random value in the range; X i,best is the optimal position (scheduling optimal solution) found for the current i-th individual; X i,a1 , X i,a2 , X i,a3 is the position of the i-th individual in the search space; R is the distance parameter, which is related to the mutation rate M R Jointly control the process of virus spread and herd immunity behavior.

[0171] Step 2.1.5, evaluate the objective function value of the new solution after mutation and update the population; the specific process is:

[0172] Compare the solution obtained after the mutation in step 2.1.4 with the optimal solution sequence obtained in step 2.1.3. If the value is smaller, it means that the optimization ability is stronger, that is, the operating cost of each unit in the power generation system is lower, then the original optimal solution sequence in step 2.1.3 is replaced; otherwise, the individual optimization ability after mutation is weaker, that is, the operating cost of each unit in the power generation system is increased, then the original optimal solution sequence in step 2.1.3 is retained.

[0173] Step 2.1.6, repeat steps 2.1.2-2.1.5 for the obtained new population until the termination condition of the maximum number of iterations is reached.

[0174] Step 2.1.7, output the optimal solution and the worst solution, and obtain the output plan of each generator set under the conditions of the lowest and highest unit operating cost;

[0175] Step 2.2, design comprehensive ranking indicators, further optimize each generator set under the condition of lowest unit operating cost obtained in step 2.1, and form the best load distribution plan;

[0176] Considering the complexity of multi-cost coordination of the system, it is difficult to ensure that the minimum value of each optimization cost is the best load distribution solution. Since different indicators present different results, the present invention proposes a comprehensive ranking indicator, which obtains the correct overall evaluation of the solution of the optimal load distribution problem of the generator set by following different optimization mechanisms, and the solution corresponding to the highest comprehensive ranking indicator value is taken as the best load distribution solution;

[0177] Step 2.2.1, the ranking index based on the average fuzzy membership reflects the average satisfaction of multiple cost solutions. For the problem of minimizing the unit operating cost, the membership function is shown in formula (22):

[0178]

[0179] Then the average fuzzy membership value can be expressed as formula (23):

[0180]

[0181] Where: μ m,n is the membership value of the mth allocation scheme under the nth cost function; F m,n is the mth solution of the nth cost function; F n,max 、F n,min is the optimal solution and the worst solution of the nth cost function; N is the total number of cost functions; w n is the weight factor of the objective function, and ∑w n =1.

[0182] Step 2.2.2, the ranking index based on the TOPSIS method, is a multi-attribute decision-making method that aims to select the solution that is closest to the optimal solution and farthest from the worst solution.

[0183] In TOPSIS, each objective function value is first normalized, and then the normalized value is multiplied by the corresponding weight to obtain the normalized value, as shown in formula (24):

[0184]

[0185] Among them: m,n Normalize the value of the mth allocation under the nth cost function.

[0186] According to the optimal and worst solutions obtained by the popular virus propagation algorithm, the optimal solution is defined as The worst solution is defined as Calculate the Euclidean distance between each allocation scheme and the optimal solution and the worst solution, as shown in formula (25):

[0187]

[0188] in: is the Euclidean distance between the mth allocation scheme and the optimal solution and the worst solution.

[0189] The closeness of each allocation scheme to the optimal solution is calculated. The higher the closeness, the more obvious the superiority of the scheme. The optimal load allocation scheme can be determined by sorting, as shown in formula (26):

[0190]

[0191] Step 2.2.3, since the optimization performance of each indicator is different, the optimal load allocation scheme for the mth load is finally determined through the composite indicator, as shown in formula (27):

[0192]

[0193] According to the ranking results of the composite index, the highest composite ranking index is used as the optimal load distribution solution that weighs multiple costs. If the result after ranking the index is the same as the optimal solution obtained by the algorithm in step 2.1, it is retained, otherwise it is updated to the optimal solution after ranking.

[0194] Example 2

[0195] In order to verify the effectiveness of the load optimization allocation method for generator sets based on the epidemic virus propagation algorithm proposed in this invention, an improved 4-machine test system including wind power, photovoltaic power, thermal power, hydropower and pumped storage is used as an example for simulation analysis. There are 4 thermal power units with a total installed capacity of 1730MW, 1 wind farm with a total installed capacity of 680MW, 1 photovoltaic power station with a total installed capacity of 500MW, 2 single-unit 180MW hydropower units, and the total installed capacity of pumped storage is 150MW. The parameters of each thermal power unit are shown in Table 1, Table 2, and Table 3.

[0196] Table 1 Parameters of each thermal power unit

[0197]

[0198] Table 2 Main parameters of conventional hydropower stations

[0199] <![CDATA[P Hi,max / P Hi,min (MW)]]> 300 / 0 <![CDATA[α3]]> 0.0300 <![CDATA[Q Hi,max / Q Hi,min (×10 4 m 3 )]]> 15 / 5 <![CDATA[α4]]> 0.9000 <![CDATA[α1]]> -0.0042 <![CDATA[α5]]> 10 <![CDATA[α2]]> -0.4200 <![CDATA[α6]]> -50

[0200] Table 3 Main parameters of pumped storage power station

[0201] <![CDATA[P g,max ]]> <![CDATA[P g,min ]]> <![CDATA[P p,max ]]> <![CDATA[P p,min ]]> <![CDATA[E max ]]> <![CDATA[E ini ]]> <![CDATA[η d ]]> <![CDATA[η c ]]> 150MW 0 150MW 0 1200MWh 600MWh 85％ 87％

[0202] A typical day is selected as the dispatching scenario, and the coordinated operation optimization of the energy-saving power generation system is analyzed. The load, wind power output, and photovoltaic output curves of a typical day are as follows: Figure 2 As shown in Table 4, the optimal output of each generator set in the dispatch cycle on a typical day in January obtained by the epidemic virus propagation algorithm is shown in Table 4. The optimal dispatch output bar chart is shown in Figure 3 As shown;

[0203] Table 4 Optimal dispatch output values ​​of energy-saving power generation system in each period of a typical day

[0204]

[0205] It can be seen from the above table that under the regulation of flexible power sources such as hydropower, thermal power and pumped storage, wind power and photovoltaic power generation are fully connected to the grid during the entire dispatching cycle, and there is no wind and light abandonment, so there is no penalty cost. Hydropower and pumped storage cooperate with each other to jointly stabilize the fluctuation of renewable energy. Pumped storage not only maintains rated power pumping conditions during the valley load period, but also generates electricity at rated power during the peak load period, exerting its ultimate regulation ability. The hydropower unit not only stops running during the valley load period to ensure the minimum output regulation of the thermal power unit and reduce the start-up and shutdown costs of the thermal power unit; but also cooperates with the pumped storage unit during the peak load period to jointly meet the peak power load demand and avoid load shedding or power shortage in the power system. In general, the energy-saving power generation system load distribution model established by the present invention can make full use of the natural characteristics of different energy sources to achieve mutual assistance and complementarity, thereby improving the operating characteristics of each power source and greatly reducing the negative impact of wind power and photovoltaic randomness on system operation. The optimal load distribution of the system under the optimization goal of optimal economic benefits and the most stable operation is achieved.

[0206] Example 3

[0207] In order to verify the effectiveness and feasibility of the epidemic virus propagation algorithm in solving the optimal allocation problem of system load. Other meta-inspiration algorithms, including particle swarm optimization algorithm, genetic algorithm, and ant colony algorithm, were also used under the same model and computing platform, and the obtained results were compared with those of the epidemic virus propagation algorithm. In addition to meta-inspiration, mixed integer quadratic programming was also used for comparison. The specific comparison results are shown in Table 5 below;

[0208] Table 5 Comparison of statistical results of different algorithms

[0209]

[0210] From the comparison of the results in Table 5, it is found that the minimum, maximum and average values ​​obtained by the epidemic virus propagation algorithm are significantly lower than those obtained by other metaheuristic algorithms and mixed integer quadratic programming, and even its average value is significantly lower than the optimal solution searched by other algorithms, which fully demonstrates that the optimization algorithm of the present invention is superior to other methods in terms of solution quality and optimization performance. This verifies the effectiveness and feasibility of the application of the epidemic virus propagation algorithm in the field of energy-saving power generation optimization scheduling technology.

Claims

1. A method for optimizing the load distribution of generator sets based on an epidemic virus propagation algorithm, characterized in that: Follow the steps below to implement it: Step 1: Establish a dispatch model with the goal of minimizing the unit operation cost. The cost includes the coal cost, start-up and shutdown cost, wind and solar power abandonment penalty cost, and environmental emission penalty cost of each generating unit, and refine the dispatch model from the perspective of system operation constraints; Step 2: After step 1, according to the epidemic virus propagation algorithm, the optimal solution and the worst solution of each operating cost are solved under the constraints of the power system equation and inequality operation constraints; Then, a comprehensive ranking index is used to sort a large number of Pareto optimal solutions, and the highest composite ranking index is used as the best solution after weighing multiple costs.

2. The method for optimizing load distribution of generator sets based on epidemic virus propagation algorithm according to claim 1, characterized in that: In step 1, the objective function is to minimize the total operating cost of the unit, as shown in formula (1): Where: F represents the total operating cost of each unit in the power generation system; is the coal burning cost of thermal power units; is the start-up and shutdown cost of thermal power units; Penalty costs for curtailing wind and solar power; Penalty costs for the environment; The coal burning cost of thermal power units is shown in formula (2): Where: P Gi,t is the active power output of thermal power unit i at time t; a i 、b i 、c i N is the coal cost coefficient corresponding to thermal power unit i; G is the total number of thermal power units; e i and h i is the valve point effect cost coefficient of thermal power unit i; P Gi,min is the minimum output power of thermal power unit i; T represents the 24h power generation scheduling time series; The start-up and shutdown cost of the unit is shown in formula (3): Among them, S i,t is the operating state variable of thermal power unit i at time t; F i,t The cost incurred during the start and stop of thermal power unit i during period t; The penalty cost for wind and solar power abandonment is shown in formula (4): in: N is the penalty cost coefficient for wind and solar power abandonment; W 、N Pv is the total number of wind farms and photovoltaic power stations; is the actual output power of wind power and photovoltaic generator set i at time t; is the predicted output of wind power and photovoltaic generator set i at time t; The environmental emission penalty cost function is shown in equation (5) and equation (6); The environmental emission penalty function is shown in formula (7): Where: γ represents the environmental penalty coefficient; is the penalty variable of the system at time t; is the carbon emission function of thermal power units at time t; E max is the maximum carbon emission allowed for thermal power units; i 、m i 、n i Represents the carbon emission coefficient of thermal power unit i.

3. The method for optimizing load distribution of generator sets based on epidemic virus propagation algorithm according to claim 2, characterized in that: The power system operation constraints include system power balance constraints, spinning reserve constraints, output power constraints of each unit, thermal power unit ramping and start-stop time constraints, hydropower flow and storage capacity constraints, pumped storage operation and energy balance constraints; The system power balance constraint is shown in equation (8): Where: N H is the total number of hydropower units; N PH is the total number of pumped storage units; P Hi,t is the output power of the i-th hydropower unit at time t; is the output power of the i-th pumped storage unit at time t; u i,t is the operating state variable of the i-th pumped storage unit at time t; P D,t is the load demand of the system at time t; The positive and negative spinning reserve constraints of the system are shown in formula (9): Where: P Gi,max , P Gi,min is the maximum and minimum output power of thermal power unit i; P Hi,max , P Hi,min is the maximum and minimum output power of hydropower unit i; is the maximum and minimum power generation of pumped storage unit i; is the spinning reserve required by the system at time t; The output power constraints of each unit are shown in formula (10); in: is the maximum and minimum pumping power of pumped storage unit i; is the maximum and minimum output power of wind turbine i; is the maximum and minimum output power of photovoltaic generator set i; The climbing constraint of thermal power unit is shown in formula (11): -R Gi,down Δt≤P Gi,t+1 -P Gi,t ≤R Gi,up Δt (11); Where: R Gi,down , R Gi,up It indicates the rate of increase and decrease of the output power of the thermal power unit; Δt is the interval time; The minimum start-up and shutdown time constraint of thermal power units is shown in formula (12): in: is the number of hours that thermal power unit i is continuously turned on and off during the t-1 period; is the minimum start-up and shutdown time of thermal power unit i; The nonlinear relationship between hydropower generation power, turbine flow and reservoir capacity is expressed by a binary quadratic function, as shown in equations (13) and (14): P Hi,t =f H (V t ,Q Hi,t ) (13); Where: α1, α2, α3, α4, α5, α6 are hydropower generation coefficients; V t is the storage capacity of the upstream reservoir of the hydropower station during period t; Q Hi,t is the power generation reference flow of the hydropower unit during period t; The water flow constraint used when the hydropower unit generates electricity is shown in formula (15): Q Hi，min ≤Q Hi,t ≤Q Hi，max (15); Where: Q Hi,min , Q Hi,max They are the maximum and minimum flow rates allowed for hydroelectric generating units, respectively; The reservoir capacity constraint of the hydropower station is shown in formula (16): V t+1 =V t +V in -Q Hi,t -SP t (16); Where: V in Indicates the amount of water flowing into the upstream reservoir of the hydropower station in each scheduling period; SP t It represents the amount of water evaporated or overflowed from the reservoir during period t; Pumped storage unit operating condition constraints: Pumped storage units cannot perform both power generation and pumping at the same time, as shown in formula (17): P gi,t ×P pi,t =0(17); Where: P gi,t represents the power generation of pumped storage unit i in period t; P pi,t represents the pumping power of pumped storage unit i in period t; The reservoir energy balance constraint is shown in equations (18) and (19): 0≤E t ≤E t,max (19); Where: E t Represents the reservoir capacity; E t，max Represents the maximum storage capacity of the reservoir; η g , η p They represent the power generation efficiency and pumping efficiency of the pumped storage unit respectively.

4. The method for optimizing load distribution of generator sets based on epidemic virus propagation algorithm according to claim 3, characterized in that: In the step 2, specifically: Step 2.1, by simulating the process of the epidemic virus spreading group gradually changing from susceptible to herd immunity, and mapping these states to the solution space of the optimization problem, the complex optimization problem is iteratively optimized; the specific steps are: Step 2.1.1, parameter initialization stage, set the dimension of the search space to D, which is the number of independent variables; the size of the population is N Pop ; Initial number of iterations k = 0, maximum number of iterations I ter,max ; Mutation rate M R ; V min 、V max is the minimum and maximum value range of the variable in the objective function; Step 2.1.2, randomly initialize the solution population, each solution represents the optimal output plan of each unit in the scheduling period, that is, the optimal distribution of load; evaluate each solution, and arrange all solutions in ascending order according to the objective function, the first solution is the optimal solution, and the last solution is the worst solution; Step 2.1.3, based on the viral replication stage of frameshift technology, new proteins in the population are generated by frameshift technology to form a new optimal solution set; Step 2.1.4, apply the distance crossover operator to simulate immune behavior, and arbitrarily select gene fragments from the optimal immune individuals for crossover mutation to generate new mutation solutions; Step 2.1.5, evaluate the objective function value of the new solution after mutation and update the population; Step 2.1.6, repeat steps 2.1.2-2.1.5 for the obtained new population until the termination condition of the maximum number of iterations is reached; Step 2.1.7, output the optimal solution and the worst solution, and obtain the output plan of each generator set under the conditions of the lowest and highest unit operating cost; Step 2.2, design comprehensive ranking indicators, further optimize each generator set under the condition of lowest unit operating cost obtained in step 2.1, and form the best load distribution plan; specifically: Step 2.2.1, for the problem of minimizing the unit operating cost, the membership function is shown in formula (22): Then the average fuzzy membership value can be expressed as formula (23): Where: μ m,n is the membership value of the mth allocation scheme under the nth cost function; F m,n is the mth solution of the nth cost function; F n,max 、F n,min is the optimal solution and the worst solution of the nth cost function; N is the total number of cost functions; w n is the weight factor of the objective function, and ∑w n =1; Step 2.2.2, based on the ranking index of the TOPSIS method, each objective function value is first normalized, and then the normalized value is multiplied by the corresponding weight to obtain the normalized value, as shown in formula (24): Among them: m,n Normalized value for the mth allocation scheme under the nth cost function; According to the optimal and worst solutions obtained by the popular virus propagation algorithm, the optimal solution is defined as The worst solution is defined as Calculate the Euclidean distance between each allocation scheme and the optimal solution and the worst solution, as shown in formula (25): in: is the Euclidean distance between the mth allocation scheme and the optimal solution and the worst solution; The closeness of each allocation scheme to the optimal solution is calculated. The higher the closeness, the more obvious the superiority of the scheme. The optimal load allocation scheme can be determined by sorting, as shown in formula (26): Step 2.2.3, since the optimization performance of each indicator is different, the optimal load allocation scheme for the mth load is finally determined through the composite indicator, as shown in formula (27): According to the ranking results of the composite index, the highest composite ranking index is used as the optimal load distribution scheme that weighs multiple costs; if the result after the ranking index is the same as the optimal solution obtained by the algorithm in step 2.1, it is retained, otherwise it is updated to the optimal solution after sorting.

5. The method for optimizing load distribution of generator sets based on epidemic virus propagation algorithm according to claim 4, characterized in that: In step 2.1.3, the specific process is: Using the +1 shifting technique, the value of the initialized optimal load distribution solution is shifted to the right by 1, and the value of the first position is set to [V min ,V max ] range, the specific process is shown in formula (20): Where: Z k represents the current optimal solution generated by the kth generation, X is the initial optimal solution, and D is the dimension of the optimization problem.

6. The method for optimizing load distribution of generator sets based on epidemic virus propagation algorithm according to claim 5, characterized in that: In step 2.1.4, the specific process is: Apply the distance crossover operator to simulate immune behavior for generating new optimal solution sequences, select gene fragments from the optimal immune individuals for crossover mutation, and evaluate the fitness of the new mutation solution; The distance crossover operator is divided according to the social distance between individuals into infected individuals, susceptible individuals and optimal immune individuals, which correspond to the following formula (21): Where: X i (t) is the solution before crossover mutation; X i (t+1) is the solution after crossover mutation; r is [V min ,V max ]A random value in the range; X i,best is the optimal position found for the current i-th individual; X i,a1 , X i,a2 , X i,a3 is the position of the i-th individual in the search space; R is the distance parameter.

7. The method for optimizing load distribution of generator sets based on epidemic virus propagation algorithm according to claim 6, characterized in that: In step 2.1.5, the specific process is: Compare the solution obtained after the mutation in step 2.1.4 with the optimal solution sequence obtained in step 2.1.

3. If the value is smaller, it means that the optimization ability is stronger, that is, the operating cost of each unit in the power generation system is lower, then the original optimal solution sequence in step 2.1.3 is replaced; otherwise, the individual optimization ability after mutation is weaker, that is, the operating cost of each unit in the power generation system is increased, then the original optimal solution sequence in step 2.1.3 is retained.

Images