An elevator group control scheduling method based on improved NSGA-II algorithm

Abstract

The application discloses an elevator group control scheduling method based on an improved NSGA-II algorithm and belongs to the field of elevator group control scheduling methods. The method comprises the following steps: obtaining elevator call data and elevator state data, establishing an elevator operation mathematical model, taking total passenger waiting time and system total energy consumption as optimization targets, and constructing a fitness function of the targets; introducing heuristic population initialization, self-adaptive crossover mutation rate and other methods to improve the NSGA-II algorithm, solving multiple targets through the improved NSGA-II algorithm, and obtaining a pareto solution set; automatically judging a current traffic mode according to the distribution of the pareto solution set; and making a decision on the pareto solution set according to the traffic mode, the fitness value of the solution and a decision scheme to obtain an optimal solution, namely, an elevator dispatching scheme. The method can automatically judge the traffic mode in the solving process, can solve the dependence of an optimization algorithm on the traffic mode before calculation, and can quickly converge and jump out of a local optimum.

Description

Technical Field

[0001] This invention relates to an elevator group control and scheduling method based on an improved NSGA-II algorithm, belonging to the field of elevator scheduling and control methods. Background Technology

[0002] In urban high-rise buildings with high passenger transport demand, multiple elevator cars are typically installed at the same call location to reduce the workload of a single elevator and improve transport efficiency. However, when passenger flow is high, how to design control strategies that consider both transport efficiency and system energy consumption to achieve efficient coordinated scheduling of multiple elevator cars has always been a research challenge.

[0003] Current intelligent optimization algorithms, when constructing the objective function, require the availability of traffic patterns as a prerequisite. Different traffic patterns are used to assign different weights to each objective, and the objectives are linearly weighted to form a single objective function. This dependence on traffic patterns limits the applicability of this method and prevents it from adapting to various architectural scenarios. Furthermore, most intelligent algorithms suffer from local convergence and premature convergence when solving the objective function. Therefore, designing a scheduling optimization algorithm that can both adaptively determine traffic patterns for different scenarios and achieve global optimum is of great significance. Summary of the Invention

[0004] To improve the operating efficiency and service level of multi-car elevator systems, this invention provides an elevator group control scheduling optimization method based on an improved NSGA-II algorithm. It offers a new scheduling approach that can adaptively determine traffic patterns to solve the scheduling of multi-car elevators, making elevator scheduling more efficient and energy-saving.

[0005] The technical solution adopted in this invention has the following steps:

[0006] Step 1: Obtain elevator floor operation information, including the total number of staircases, the number of cars, the floor and direction called by passengers, the destination floor of passengers in the elevator, the floor and direction of elevator operation, and the total number of internal and external calls;

[0007] Step 2: Based on the obtained elevator floor operation information, establish a mathematical model of elevator operation, and construct a multi-objective value function based on the mathematical model;

[0008] Step 3: Based on the obtained elevator floor operation information, encode each individual using real number encoding to obtain an individual code. The length of the code is equal to the number of passenger calls to the elevator. Each element in the code takes a value between 1 and N, where N is the number of elevator cars.

[0009] Step 4: Based on the individual codes in Step 3, a parent population is generated using a combination of heuristic initialization and random initialization. Then, based on the multi-objective value function in Step 2, the multi-objective values ​​of the individuals in the parent population are calculated, and non-dominated sorting is performed based on the multi-objective values ​​to obtain the non-dominated sorting level of the parent population.

[0010] Step 5: Combining the non-dominated sorting of the parent population, perform selection, crossover, and mutation operations on the parent population to generate the offspring population. Merge the parent and offspring populations, calculate crowding and perform non-dominated sorting, retain the best individuals, and generate a new generation of parent population.

[0011] Step 6: Return the new generation of parent population generated in Step 5 as input to the iteration in Step 5. After each iteration, increment the iteration count (iter) of Step 5 by 1. If the iteration count after incrementing by 1 is greater than Iter... thred The iteration ends when the iteration ends. After the iteration ends, the final obtained parent population and non-dominated sorting level is output. Based on the final obtained parent population and non-dominated sorting level, the Pareto non-dominated solution set is obtained, where It thred This is the threshold for the end of the iteration count;

[0012] Step 7: Determine the traffic pattern based on the distribution of solutions in the Pareto non-dominated solution set;

[0013] Step 8: Based on the determined traffic pattern and fitness value function, make a decision on the Pareto non-dominated solution set and select the best solution as the optimal ladder dispatch scheme.

[0014] Furthermore, the multi-objective construction in step 2 includes: establishing the total waiting time function F. wt Total elevator travel time function F rt The total energy consumption function of the system F e According to F wt F rt and F e Establish a multi-objective optimization objective min[F wrt F e ], where F wrt =F wt +F rt This represents the total waiting and riding time for the elevator.

[0015] The heuristic initialization in step 4 includes forward interval allocation elevator initialization and forward minimum waiting time allocation elevator initialization, as follows:

[0016] (1) The initialization of the forward interval allocation of elevators specifically includes: sorting all passenger calls for elevators by floor and direction, elevator operation floor and direction, in ascending and descending order and floor height in a cyclical manner, and all passenger calls between two adjacent cars are allocated to the upcoming forward car in the adjacent car.

[0017] (2) The initialization of the elevator dispatching with the minimum waiting time in the forward direction includes: sorting all passenger elevator calls by floor and direction, elevator operation floor and direction, in a cyclical manner according to up and down and floor height, starting from the first passenger call in front of each car, and assigning the passenger elevator call to the car with the shortest waiting time in the forward direction car that is about to arrive.

[0018] During population initialization, introducing individuals generated heuristically can improve the quality of the initial population.

[0019] The selection, crossover, and mutation operations in step 5 are as follows:

[0020] (1) Based on the non-dominant ranking of the parent population, select the individuals and their number in the parent population that require crossover and mutation operations using a binary tournament selection strategy. The selection of the number of individuals for crossover and mutation is as follows:

[0021]

[0022] P m =1-P c

[0023] Where P c P is the crossover rate. cmin To minimize the crossover rate, P cmax The maximum crossover rate is given by P, F1 is the number of solutions in the non-dominated solution set, npop is the number of individuals in the population, and P is the number of individuals in the population. m To adapt to the mutation rate, when the F1 value in the population is large, the individuals in the population tend to be consistent. At this time, the crossover operation is not very meaningful. By increasing the proportion of mutated individuals, the probability of population evolution can be increased.

[0024] P cnum =npop*P c P mnum =npop*P m

[0025] Among them, P cnum For the number of individuals in the crossover operation, P mnum The number of individuals subjected to the mutation operation;

[0026] (2) Perform crossover and mutation operations on the selected individuals to generate offspring populations.

[0027] The specific steps for determining the traffic mode in step 7 are as follows:

[0028] (1) Obtain the objective function F in the Pareto nondominated solution set. wrt The minimum value F wrt_min ;

[0029] (2) According to F wrt_min The magnitude of the value determines the traffic pattern (TP), specifically:

[0030]

[0031] Among them, f low_th and f high_th Based on the elevator model corresponding to the building, and according to common upward peak, downward peak, and inter-floor patterns in elevator group control and scheduling, the target value F corresponding to different traffic modes is simulated. wrt The threshold, where f high_th The objective value F in the non-dominated solution set for the upward and downward peaks. wrt The minimum or mean value of f low_th The objective value F in the non-dominated solution set corresponding to the inter-layer pattern. wrt The minimum or average value is used to classify the peak, normal and low-peak modes.

[0032] The specific steps of step 8 are as follows:

[0033] (1) For each solution in the obtained Pareto non-dominated solution set, the F corresponding to the solution wrt and F e Dividing by the total number of inbound and outbound calls respectively yields the F corresponding to each solution in the Pareto non-dominated solution set. wrt and F e The average value is calculated, and then the corresponding fitness value is calculated. The fitness value function is as follows:

[0034]

[0035] Among them, M all f represents the total number of inbound and outbound calls. wrt f is the fitness function of the average waiting time for elevators. e The fitness value function is the average energy consumption of the system.

[0036] (2) Based on the traffic mode TP obtained in step 7, obtain F. wrt and F e The weighting coefficients corresponding to the average fitness value can be selected as follows for different traffic patterns:

[0037] TP <![CDATA[w1]]> <![CDATA[w2]]> Peak mode 0.8 0.2 Normal mode 0.5 0.5 Off-peak mode 0.2 0.8

[0038] Where w1 is the average waiting time for the elevator, F wrt The fitness value corresponds to the weighting coefficient, and w2 is the system average energy consumption F. e The weight coefficients corresponding to the fitness values;

[0039] (3) Based on the weighting coefficients obtained in step (2), perform a weighted summation of the two fitness value functions obtained in step (1) to obtain the total fitness f of each solution under the current traffic mode. all ,as follows:

[0040] f all =w1f wrt +w2f e

[0041] (4) Calculate f for each solution in the Pareto non-dominated solution set. all Choose total adaptation f all The solution with the largest value is the optimal solution, i.e., the best Pythagorean theorem.

[0042] Compared with the prior art, the present invention has the following advantages:

[0043] The method of this invention can solve the problem of dependence on traffic modes before solving existing elevator group control and scheduling algorithms by adaptively judging traffic modes during the multi-objective optimization process, thereby improving the applicability and occasions of the algorithm. In addition, by improving the NSGA-II algorithm, the algorithm can escape local optima during the solution process, thereby improving the quality of the solution. Attached Figure Description

[0044] Figure 1 The flowchart shows the group control elevator scheduling method based on the improved NSGA-II algorithm in the invention.

[0045] Figure 2 Flowchart for solving the improved NSGA-II algorithm;

[0046] Figure 3 A comparison of the Pareto non-dominated solution set distributions before and after the NSGA-II algorithm improvement;

[0047] Figure 4 This is a graph showing the judgment of different traffic modes under the Pareto solution set distribution;

[0048] Figure 5 A comparison chart showing the fitness of the final solutions obtained before and after the improvement of the NSGA-II algorithm;

[0049] Figure 6 This is a fitness comparison chart showing the results of the genetic algorithm GA optimizing a linearly weighted single objective and the multi-objective optimization using the improved NSGA-II algorithm. Detailed Implementation

[0050] This invention addresses the problem that existing scheduling algorithms rely heavily on elevator traffic patterns, which limits their application. It proposes an elevator group control scheduling optimization method based on an improved NSGA-II algorithm. This method adaptively determines traffic patterns to reduce user waiting time and elevator energy consumption.

[0051] The specific implementation steps are as follows:

[0052] like Figure 1 As shown, the elevator group control and scheduling task is divided into a data acquisition module, a scheduling algorithm module, and a group control elevator dispatching module. The data acquisition module includes: collecting the floor and direction of the elevator call, the number of cars, the direction of car travel and the floor it is currently on, and the floor the caller is calling from inside the car. The scheduling algorithm module includes: establishing a mathematical model for multi-elevator operation, solving the scheduling optimization algorithm, and a traffic pattern recognition and decision-making module. The group control elevator dispatching module includes: updating the car dispatching task and a single elevator operation module.

[0053] The algorithm solution part is as follows Figure 2 As shown, the specific steps are as follows:

[0054] Step 1: Obtain elevator floor operation information, including the total number of staircases, the number of cars, the floor and direction called by passengers, the destination floor of passengers in the elevator, the floor and direction of elevator operation, and the total number of internal and external calls;

[0055] Step 2: Based on the obtained elevator floor operation information, establish a mathematical model of elevator operation and construct a multi-objective value function;

[0056] The multi-elevator model is composed of multiple single-elevator models. For the mathematical model of single-elevator operation, the forward-intercepting scheduling principle is adopted. Based on the current destination floor, outward call floor, and the forward-intercepting principle, the waiting time, riding time, and system energy consumption of each passenger to complete all tasks of the current inward and outward calls are calculated. The function for multiple objective values ​​is constructed as follows:

[0057] Establish a total elevator waiting time model, where waiting time is the time from when a passenger presses the elevator call button to when the elevator car arrives at that floor. The specific model is as follows:

[0058]

[0059] Where N represents the total number of cars, M represents the total number of external call signals from all cars, w(i,j) represents the waiting time for car i to respond to the j-th call signal, and F wt Total waiting time;

[0060] Establish a total elevator travel time model, where travel time is the time from when a passenger enters the elevator to when they reach their destination floor. The specific model is as follows:

[0061]

[0062] Among them, w r (i,j) represents the elevator travel time during the period corresponding to the j-th external call signal in car i; w s (i,j) represents the elevator stop time during the passenger's ride corresponding to the j-th external call signal in car i, F rt Total elevator travel time;

[0063] Establish a system total energy consumption model. System energy consumption: the energy consumption generated during the process of completing all outbound and inbound call destination layers at a certain moment. The specific model is as follows:

[0064] F e =n r C r +n s C s

[0065] Where n r This represents the number of floors traversed during this period, n. s Indicates the number of stops, C r C represents the average energy consumption per layer of operation. s F represents the average energy consumption per stop. e This represents the total energy consumption of the system.

[0066] The multi-objective model to be optimized is established as follows:

[0067] min[F wrt F e ]

[0068] Among them, F wrt =F wt +F rt F represents the total waiting time for the elevator. e This represents the total energy consumption of the system.

[0069] Step 3: Based on the obtained elevator floor operation information, encode each individual using real number encoding. The length of the code is equal to the number of passenger calls to the elevator. Each element in the code takes a value between 1 and N, where N is the number of elevator cars.

[0070] For example, corresponding to 3 elevator cars and 8 external call signals, the corresponding codes are {2,3,1,1,2,3,1,2}. From left to right, this means that car 2 responds to external call 1, car 3 responds to external call 2, car 1 responds to external call 3, and so on, until car 2 responds to external call 8. That is, car 1 responds to calls 3, 4, and 7, car 2 responds to calls 1, 5, and 8, and car 3 responds to calls 2 and 6. The calls numbered 1-8 correspond to the floor and direction information of each elevator.

[0071] Step 4: Based on the individual codes in Step 3, a combination of heuristic initialization and random initialization is used to generate the parent population. Then, based on the multi-objective value function in Step 2, the multi-objective values ​​of the individuals in the parent population are calculated, and non-dominated sorting is performed based on the multi-objective values ​​to obtain the non-dominated sorting level of the parent population.

[0072] (1) The initialization of the forward interval allocation of elevators specifically includes: sorting all passenger calls for elevators by floor and direction, elevator operation floor and direction, in ascending and descending order and floor height in a cyclical manner, and all passenger calls between two adjacent cars are allocated to the upcoming forward car in the adjacent car.

[0073] (2) The initialization of the elevator dispatching with the minimum waiting time in the forward direction includes: sorting all passenger elevator calls by floor and direction, elevator operation floor and direction, in a cyclical manner according to up and down and floor height, starting from the first passenger call in front of each car, and assigning the passenger elevator call to the car with the shortest waiting time in the forward direction car that is about to arrive.

[0074] Step 5: Combining the non-dominated ranking level of the parent population, perform selection, crossover, and mutation operations on the parent population to generate offspring populations. Merge the parent and offspring populations to obtain a merged population. Calculate the crowding degree and perform non-dominated ranking on the merged population to obtain its non-dominated ranking level. Retain the best individuals from the merged population to generate a new generation of parent populations. Perform non-dominated ranking on the new generation of parent populations to obtain their non-dominated ranking level.

[0075] The specific operations for selection, crossover, and mutation are as follows:

[0076] (1) Based on the non-dominant ranking of the parent population, select the individuals and their number in the parent population that require crossover and mutation operations using a binary tournament selection strategy. The selection of the number of individuals for crossover and mutation is as follows:

[0077]

[0078] P m =1-P c

[0079] Where P c P is the crossover rate. cmin To minimize the crossover rate, P cmax The maximum crossover rate is given by P, F1 is the number of solutions in the non-dominated solution set, npop is the number of individuals in the population, and P is the number of individuals in the population. m To adapt to the mutation rate, when the F1 value in the population is large, the individuals in the population tend to be consistent. At this time, the crossover operation is not very meaningful. By increasing the proportion of mutated individuals, the probability of population evolution can be increased.

[0080] P cnum =npop*P c P mnum =npop*P m

[0081] Among them, P cnum For the number of individuals in the crossover operation, P mnum The number of individuals subjected to the mutation operation;

[0082] (2) Perform crossover and mutation operations on the selected individuals based on the selected individuals, and perform crossover and mutation operations on the chromosome positions to generate offspring populations.

[0083] Step 6: Return the new generation of parent population generated in Step 5 as input to the iteration in Step 5. After each iteration, increment the iteration count (iter) of Step 5 by 1. If the iteration count after incrementing by 1 is greater than Iter... thred The iteration ends when the iteration ends. After the iteration ends, the final obtained parent population and non-dominated sorting level is output. Based on the final obtained parent population and non-dominated sorting level, the Pareto non-dominated solution set is obtained, where It thred This is the threshold for the end of the iteration count;

[0084] The distribution of Pareto non-dominated solution sets obtained by the NSGA-II algorithm before and after the improvement are compared as follows: Figure 3 As shown, the improved algorithm can obtain a better Pareto non-dominated solution set.

[0085] Step 7: Based on the distribution of solutions in the Pareto non-dominated solution set, determine the traffic pattern as follows:

[0086] Find F in the Pareto nondominated solution set wrt Find the minimum value of F. wrt_min According to F wrt_mi The magnitude of the value determines the traffic pattern (TP), specifically:

[0087]

[0088] Among them, f low_th and f high_th Based on the elevator model corresponding to the building, and according to common upward peak, downward peak, and inter-floor patterns in elevator group control and scheduling, the target value F corresponding to different traffic modes is simulated. wrt The threshold, where f high_th The objective value F in the non-dominated solution set for the upward and downward peaks. wrt The minimum or mean value of f low_th The objective value F in the non-dominated solution set corresponding to the inter-layer pattern. wrt The minimum or mean value, such as Figure 4 As shown, it is divided into peak mode, normal mode and off-peak mode.

[0089] Step 8: Based on the traffic pattern determined in Step 7, the constructed fitness function, and the decision scheme, make a decision on the Pareto non-dominated solution set and select the optimal solution, which is the best ladder dispatch scheme, as follows:

[0090] (1) Divide the two objective values ​​corresponding to each solution in the Pareto non-dominated solution set obtained in step 6 by the total number of inbound and outbound calls to obtain the average value of the two objectives in the Pareto non-dominated solution set. Then calculate the fitness value corresponding to the average value. The fitness value function is as follows:

[0091]

[0092] Among them, M all f represents the total number of inbound and outbound calls. wrt f is the fitness function of the average waiting time for elevators. e The fitness value function is the average energy consumption of the system.

[0093] (2) Based on the traffic mode TP obtained in step 7, the weight coefficients of the two objective functions are obtained. The selection of weight coefficients under different traffic modes is as follows:

[0094] TP <![CDATA[w1]]> <![CDATA[w2]]> Peak mode 0.8 0.2 Normal mode 0.5 0.5 Off-peak mode 0.2 0.8

[0095] Where w1 is the weight corresponding to the fitness value function of the average waiting time for elevators, and w2 is the weight corresponding to the fitness value function of the average energy consumption of the system.

[0096] (3) Based on the weighting coefficients obtained in step (2), perform a weighted summation of the two fitness value functions obtained in step (1) to obtain the total fitness f of each solution under the current traffic mode. all ,as follows:

[0097] f all =w1f wrt +w2f e

[0098] (4) Calculate f for each solution in the Pareto non-dominated solution set. all Choose total adaptation f all The solution with the largest value is the optimal solution, i.e., the best Pythagorean theorem. The fitness of the final solution is as follows: Figure 5 , Figure 6 As shown, in addition to being able to adaptively determine traffic patterns, the solutions obtained by this algorithm also have higher fitness.

[0099] The group control system assigns elevator call signal response tasks to individual elevators, which then respond individually to ensure efficient operation. Simultaneously, each elevator transports passengers to their destination floor based on the call requests from within the car.

Claims

1. An elevator group control and scheduling method based on an improved NSGA-II algorithm, characterized in that, Includes the following steps: Step 1: Obtain elevator floor operation information, including the total number of staircases, the number of cars, the floors and directions called by passengers, the destination floors of passengers in the elevator, the floor and direction where the elevator is currently operating, and the total number of internal and external calls; Step 2: Based on the obtained elevator floor operation information, establish a mathematical model of elevator operation, and construct a multi-objective value function based on the mathematical model of elevator operation. The multi-objective value function includes the total waiting time function and the total energy consumption function of the system. Step 3: Based on the obtained elevator floor operation information, encode each individual using real number encoding to obtain an individual code. The length of the code is equal to the number of passenger calls to the elevator. Each element in the code takes a value between 1 and N, where N is the number of elevator cars. Step 4: Based on the individual codes in Step 3, a parent population is generated using a combination of heuristic and random initialization. Then, based on the multi-objective value function in Step 2, the multi-objective values ​​of the individuals in the parent population are calculated, and non-dominated sorting is performed based on the multi-objective values ​​to obtain the non-dominated sorting level of the parent population. Step 5: Combining the non-dominated ranking level of the parent population, perform selection, crossover, and mutation operations on the parent population to generate the offspring population. Merge the parent and offspring populations to obtain a merged population. Calculate the crowding degree and perform non-dominated ranking on the merged population to obtain its non-dominated ranking level. Retain the best individuals from the merged population to generate a new generation of parent populations. Perform non-dominated ranking on the new generation of parent populations to obtain their non-dominated ranking level. Step 6: Return the new generation of parent population generated in Step 5 as input to the iteration in Step 5. After each iteration, increment the iteration count (iter) of Step 5 by 1. If the iteration count after incrementing by 1 is greater than the previous one, then the iteration count is considered complete. The iteration ends when the iteration ends. After the iteration ends, the final obtained parent population and non-dominated sorting level are output. The Pareto non-dominated solution set is obtained based on the final obtained parent population and non-dominated sorting level. This is the threshold for the end of the iteration count; Step 7: Based on the distribution of solutions in the Pareto non-dominated solution set, determine the traffic mode. Based on the determined traffic mode and fitness value function, make a decision on the Pareto non-dominated solution set and select the best solution as the optimal ladder dispatch scheme.

2. The elevator group control and scheduling method based on the improved NSGA-II algorithm as described in claim 1, characterized in that... In step 4, heuristic initialization includes forward interval allocation ladder initialization and forward minimum waiting time allocation ladder initialization.

3. The elevator group control and scheduling method based on the improved NSGA-II algorithm as described in claim 2, characterized in that: The initialization of forward interval allocation and dispatching specifically includes: By sorting all passenger calls for elevator floors and directions, as well as the floor and direction where the elevator is running, in a cyclical manner according to upward and downward directions and floor height, all passenger calls between two adjacent cars are assigned to the upcoming car in the same direction within that adjacent car. The initialization of the forward minimum waiting time allocation for elevator dispatch specifically includes: sorting all passenger calls to the elevator floor and direction, as well as the floor and direction where the elevator is running, in a cyclical manner according to up / down and floor height, starting from the first passenger call in front of each car, and assigning the passenger call to the car with the shortest waiting time among the forward cars that are about to arrive.

4. The elevator group control and scheduling method based on the improved NSGA-II algorithm as described in claim 1, characterized in that... Step 5 involves selection, crossover, and mutation operations on the parent population, specifically including: (1) Based on the non-dominant ranking of the parent population, a binary tournament selection strategy is used to select from the parent population. Individuals that need to perform cross-operations, and Individuals to be mutated, among which and The size is determined as follows: in For adaptive crossover rate, To minimize the crossover rate, To achieve the maximum crossover rate, The number of solutions in the non-dominated solution set. The number of individuals in the population. To adapt the mutation rate, The number of individuals to be cross-operated, The number of individuals to be mutated; (2) Perform crossover and mutation operations on the selected individuals to generate offspring populations.

5. The elevator group control and scheduling method based on the improved NSGA-II algorithm as described in claim 1, characterized in that... In step 7, based on the distribution of solutions in the Pareto non-dominated solution set, traffic patterns are determined. Based on the determined traffic patterns and fitness function, a decision is made on the Pareto non-dominated solution set to select the optimal solution, which is the best ladder dispatching scheme. This includes: (1) Obtain the corresponding solutions in the Pareto non-dominated solution set. minimum value , This refers to the total waiting and boarding time for the elevator. (2) According to The magnitude of the value determines the traffic pattern (TP), specifically: in To simulate the peak mode and normal mode The corresponding threshold, To simulate the difference between normal and off-peak modes The corresponding threshold; (3) The total waiting time for each solution in the obtained Pareto non-dominated solution set. and total system energy consumption Dividing by the total number of inbound and outbound calls respectively yields the corresponding solutions in the Pareto non-dominated solution set. and The average value is calculated, and then the fitness value corresponding to the average value is obtained. and The fitness value corresponding to the average value; (4) Based on the traffic pattern determined in (2), obtain and The weighting coefficient of the fitness value corresponding to the average value and ,in, Total waiting time for elevator The weighting coefficients corresponding to the fitness values ​​of the average value. Total system energy consumption The weighting coefficient corresponding to the fitness value of the average value; Based on the obtained weighting coefficients, in step (3) and We take the weighted sum of the fitness values ​​corresponding to the average values ​​of the solutions and obtain the total fitness value of each solution in the Pareto non-dominated solution set under the current traffic mode. We select the solution with the largest total fitness value as the best solution and use it as the optimal dispatching scheme.

Images