A cloud computing task scheduling method

Abstract

The present invention discloses a cloud computing task scheduling method, which includes constructing a mathematical model, initializing quantum positions and parameters of a quilt search mechanism and a particle swarm evolution mechanism, calculating and sorting quilt fitness values, calculating particle fitness values, finding a global optimal quantum position, updating the quilt quantum position, determining a new generation of quilt quantum positions, updating the particle quantum position, calculating and new particle fitness values, determining a new generation of particle quantum positions, a local optimal quantum position, and a global optimal quantum position, judging whether information interaction exists; if so, performing information interaction between the particle swarm and the quilt group, and re-determining the global optimal position of the particle swarm; if not, continuing execution; iteratively updating the quantum position to a maximum number of iterations, and obtaining a scheduling strategy based on a mapping matrix of the optimal quantum position in the last generation corresponding to the task and a virtual machine allocation matrix. The present invention achieves a minimum total task execution time, total execution cost, and load imbalance under budget and deadline constraints.

Description

Technical Field

[0001] The present invention belongs to the field of cloud computing and relates to a cloud computing task scheduling method, in particular to a cloud computing task scheduling method based on the co-evolution mechanism of quantum quilt swarm and particle swarm. Background Art

[0002] With the rapid adoption and widespread application of cloud computing technology in my country, the number of users and tasks on cloud computing platforms has gradually expanded. Therefore, the key research areas in cloud computing are how to scientifically, rationally, and effectively schedule tasks, improve task allocation efficiency and resource utilization, minimize task execution time, meet other service requirements raised by cloud computing platform users, and generate additional profits for cloud service providers. Therefore, an efficient scheduling method is needed to achieve the desired goals in cloud environments. Numerous past research results have demonstrated that cloud computing task scheduling methods based on intelligent optimization algorithms offer superior results compared to traditional methods. Currently, efficient and practical task scheduling has become a hot research topic in the field of cloud computing.

[0003] Scholars have applied intelligent optimization algorithms, such as the particle swarm optimization algorithm and the whale optimization algorithm, to the optimization problem of task scheduling in cloud computing environments. However, these algorithms have drawbacks, such as slow convergence, poor global search capabilities, and an inability to easily escape local optima. Furthermore, existing cloud computing task scheduling methods primarily prioritize the quality of service for cloud computing users, optimizing for single objectives such as time or cost. With the development of cloud computing, it is necessary to consider the complex and ever-changing application requirements of cloud computing, such as the economic benefits and resource utilization of cloud service providers.

[0004] A search of existing literature revealed that Fan Ying et al., in their paper "A New Method for Cloud Computing Task Scheduling Based on Optimization of Invasive Tumor Growth under Dynamic Queues and BPNN," published in Computer Applications and Software, combined an invasive tumor growth optimization algorithm with a backpropagation neural network algorithm, independently optimizing the system's average waiting time, computational cost, and resource utilization. However, while optimizing one objective, they failed to consider the impact on other objectives, resulting in some unmet requirements. Chen Xuan et al., in their paper "Research on an Improved Bird Flock Algorithm for Task Scheduling in Mobile Cloud Computing," published in Computer Application Research, employed an improved bird flock algorithm to achieve some time savings and energy reductions, but neglected the key metric of cost. Summary of the Invention

[0005] In view of the above-mentioned existing technologies, the technical problem to be solved by the present invention is to provide a cloud computing task scheduling method based on the co-evolution mechanism of quantum quilt swarm and particle swarm, so as to ensure that the total execution time, total execution cost and load imbalance of tasks are smaller under the constraints of budget and deadline.

[0006] To solve the above technical problems, a cloud computing task scheduling method of the present invention includes the following steps:

[0007] Step 1: Establish the total execution time of the task, the total execution cost of the task and the load imbalance and construct a mathematical model;

[0008] Step 2: Initialize the quantum positions of the quill search mechanism and the particle swarm evolution mechanism and set the parameters;

[0009] Step 3: Calculate the fitness value of each tunicate, sort all tunicates, and calculate the fitness value of each particle to find the global optimal quantum position of the particle;

[0010] Step 4: Generate quantum rotation angles based on the quill group search mechanism and use a simulated simplified quantum rotation gate to update the quill's quantum position;

[0011] Step 5: Apply the greedy strategy to sort all tunicates according to their fitness values and determine the quantum position of the new generation of tunicate groups;

[0012] Step 6: Generate quantum rotation angles based on the particle swarm evolution mechanism and use a simulated simplified quantum rotation gate to update the quantum position of the particle;

[0013] Step 7: Calculate the fitness value of each particle and the new particle, and determine the quantum position, local optimal quantum position, and global optimal quantum position of the new generation of particles;

[0014] Step 8: Determine whether g+1=5k1 is reached. If it is not reached, then directly execute step nine. If it is reached, then the particle swarm and the tunic swarm are exchanged with each other, and then the tunic swarms in the tunic swarm are re-sorted according to the fitness value, and the global optimal position of the particle swarm is re-determined.

[0015] Step 9: Determine whether the current number of iterations has reached the maximum number of iterations G. If not, set g = g + 1 and return to step 4. If it has, terminate the iteration loop and obtain the final task scheduling strategy based on the allocation matrix of tasks and virtual machines corresponding to the mapping positions of the optimal quantum positions in all populations in the last generation.

[0016] Furthermore, in step 1, the total execution time of the task, the total execution cost of the task and the load imbalance are established and a mathematical model is constructed as follows:

[0017] Define the allocation matrix of tasks and virtual machines as Among them, U is the number of tasks, V is the number of virtual machines, each task can only be assigned to one virtual machine for execution. If task u is assigned to virtual machine v for execution, then Z vu =1; if task u is not assigned to virtual machine v for execution, then Z vu =0, u=1,2,…,U, v=1,2,…,V;

[0018] (1) Let the instruction length vector of the task be L = [L1, L2, ..., L U ], the execution speed vector of the virtual machine is K=[K1,K2,…,K V ], the execution time matrix of the task on the virtual machine is The execution time of task u on virtual machine v is expressed as u=1,2,…,U,v=1,2,…,V,the execution time vector of the virtual machine is D=[D1,D2,…,D V ], where the execution time of the virtual machine v can be expressed as The total execution time of a task is the maximum execution time obtained after all virtual machines execute the assigned tasks. The total execution time of a task is Z vu ∈Z, u=1,2,…,U, v=1,2,…,V, where max represents the maximum value function, satisfying the constraint λ is the cutoff time, expressed as λ = D L +r1(D H -D L ), D L The time obtained by assigning the tasks executed on the virtual machine with the longest execution time to the virtual machine with the fastest execution speed, D H The time obtained by assigning all tasks executed on the virtual machine with the longest execution time to the virtual machine with the slowest running speed, r1 is a uniform random number between [0,1];

[0019] (2) The execution cost matrix of the task on the virtual machine is: The unit execution cost matrix of the virtual machine is W = [W1, W2, ..., W V ], where the execution cost of task u on virtual machine v is expressed as C vu =E vu W v , u=1,2,…,U,v=1,2,…,V,when all tasks are assigned to virtual machines for execution, the total cost is u=1,2,…,U,v=1,2,…,V,satisfy the constraint B(Z)≤β, and the budget β is expressed as: β=B min +r2(Bmax -B min ), where B max is the highest cost obtained by assigning all tasks to the virtual machine with the highest cost, B min It is the lowest cost obtained by assigning all tasks to the virtual machine with the lowest cost, and r2 is a uniform random number between [0,1].

[0020] (3) Load imbalance is Among them, D max is the maximum execution time of all virtual machines, D min is the minimum execution time of all virtual machines, is the average execution time of all virtual machines, and min represents the minimum function;

[0021] (4) The total execution time of the task The total execution cost B(Z) and load imbalance f3(Z) of the task are normalized, and the total execution time of the task is normalized to The total execution cost of the task is normalized to The load imbalance is normalized to Then the linear weighted method is used to design the objective function, which is Among them, α1, α2 and α3 are inertia weights, α1+α2+α3=1.

[0022] Furthermore, in step 2, the quantum positions of the quill search mechanism and the particle swarm evolution mechanism are initialized and the parameters are set, including:

[0023] First, the population size of the quilt group is set to N1, the population size of the particle group is set to N2, the maximum number of iterations is G, the iteration number is labeled g, g∈[1,G], and the quantum position of the nth quilt in the U-dimensional search space at the g-th iteration is n=1,2,…,N1; the quantum position of the mth particle in the U-dimensional search space is The local optimal quantum position is m=1,2,…,N2; in the first generation, let g=1, initialize each dimension of the quantum position of the capsule group and the quantum position of the particle group to a uniform random number between [0,1], and let the quantum position of the particle be the local optimal quantum position in the first generation.

[0024] Furthermore, in step 3, the fitness value of each tunic is calculated, all tunicates are sorted, and the fitness value of each particle is calculated to find the global optimal quantum position of the particle.

[0025] Mapping the quantum position of the quilt group to position, the position of the nth quilt in the g-th iteration is The mapping formula is n=1,2,…,N1, u=1,2,…,U, where, is the u-th dimension of the position of the n-th quilt at the g-th iteration, is the u-th dimension of the quantum position of the n-th capsule at the g-th iteration, is the upper limit of the u-th dimension of the tunicate position, is the lower limit of the u-th dimension of the tunicate position, Represents the nearest integer function; the value of each dimension of the quilt's position coordinate is an integer in the range [1, V]. The position of each quilt is assigned to the task and virtual machine allocation matrix Z. The assignment rule is: the value of the u-th dimension of the position of the n-th quilt in the g-th iteration is v, that is, n=1,2,…,N1,u=1,2,…,U,v={1,2,…,V},meaning that task u is assigned to virtual machine v. Then in the task and virtual machine assignment matrix Z, Z vu = 1 and the element of the u-th column of the matrix Z is divided by Z vu = 1, and the values of the remaining elements in the column are 0; the fitness function of the position of the nth quilt in the g-th iteration is The fitness values of the position and quantum position of the nth capsule in the gth iteration obtained by the fitness function can be obtained by Indicates that all quilts are arranged in order from small to large according to their fitness values, and the quantum position of the quilt with the smallest fitness value is defined as the optimal quantum position. The optimal quantum position in the quilt group in the g-th iteration is The corresponding minimum fitness value is Map the local optimal quantum position of the particle swarm to the local optimal position. The local optimal position of the mth particle in the gth iteration is The local optimal position is assigned to the allocation matrix Z of tasks and virtual machines. The fitness function of the local optimal position of the mth particle in the gth iteration is The fitness values of the local optimal position of the mth particle and the local optimal quantum position of the gth iteration obtained by the fitness function can be obtained by Indicates that the local optimal quantum position with the smallest fitness value in the g-th iteration particle swarm is the global optimal quantum position, and the global optimal quantum position of the g-th iteration particle swarm is The corresponding minimum fitness value is

[0026] Furthermore, in step 4, the quantum rotation angle is generated according to the quill group search mechanism, and the quantum position of the quill is updated using the simulated simplified quantum rotation gate, including:

[0027] (1) The u-th dimension of the position of the n-th bladder in the g-th iteration is the water flow advection in the deep sea when the jet propulsion behavior is performed. Its own gravity is The conflicting forces between the tunicates are Where r3 is a uniform random number between [0,1]. After avoiding conflicts between quilts, the quilt will move towards the optimal quantum position. The position of the nth quilt after it moves towards the optimal quantum position is u=1,2,…,U,n=1,2,…,N1,for the position of the nth quilt after it moves towards the optimal quantum position, a uniform random number r4 is generated between [0,1]. If r4<0.5, the distance between the u-th dimension of the quantum position of the nth quilt and the u-th dimension of the quantum position of the r-th quilt randomly selected from the first n quilts is The u-th dimension of the position of the n-th capsule after it moves closer to the quantum position of the r-th capsule randomly selected from the first n capsules is If r4 ≥ 0.5, the distance between the u-th dimension of the quantum position of the n-th capsule and the u-th dimension of the optimal quantum position is The u-th dimension of the position of the nth capsule after it moves towards the optimal quantum position is in, is the u-th dimension of the optimal quantum position at the g-th iteration, is the u-th dimension of the quantum position of the r-th quilt randomly selected from the first n quilts in the g-th iteration, r∈{1,2,…,n}, r5 is a uniform random number between [0,c1], and c1 is a constant greater than 1;

[0028] (2) The quantum rotation angle vector corresponding to the updated quantum position of the nth tunicate in the population is n=1,2,…,N1; for the generation of the quantum rotation angle vector of the nth quilt, a uniform random number r6 is generated between [0,1]; if r6<0.5, the u-th dimension of the quantum rotation angle of the nth quilt in the g+1th iteration is n=1,2,…,N1,u=1,2,…,U;if r6≥0.5,the u-th dimension of the rotation angle of the n-th capsule quantum in the g+1th iteration is n=1,2,…,N1,u=1,2,…,U,r7 is a uniform random number between [2,c2], and c2 is a constant greater than 2;

[0029] (3) Using the simplified quantum rotating gate simulation, the quantum position of the nth capsule is updated as The u-th dimension of the quantum position of the n-th capsule is n=1,2,…,N1,u=1,2,…,U;the quantum positions of all N1 capsules in the population are updated in the U-dimensional search space according to the above rules.

[0030] Furthermore, in step 5, a greedy strategy is applied to sort all tunicates according to their fitness values, and the quantum positions of the new generation of tunicates are determined as follows:

[0031] The updated quantum position of the capsule is mapped to the position, and the position is assigned to the task and the virtual machine allocation matrix Z. The fitness function is used to calculate the corresponding fitness value, and the updated quantum position is sorted from small to large according to the fitness value. The quantum position of the nth capsule after sorting is The corresponding fitness value and the quantum position of the nth capsule in the previous generation The corresponding fitness values are compared, and the quantum position of the capsule with the smaller fitness value is selected as the quantum position of the nth capsule in the g+1th iteration. n=1,2,…,N1, the quantum positions of all N1 quilts are selected according to the above rules. The quantum position of the quilt with the smallest fitness value is defined as the optimal quantum position. The optimal quantum position in the quilt group at the g+1th iteration is The corresponding minimum fitness value is The quantum position of the capsule with the largest fitness value is The corresponding maximum fitness value is

[0032] Furthermore, in step 6, the quantum rotation angle is generated according to the particle swarm evolution mechanism, and the quantum position of the particle is updated using a simulated simplified quantum rotation gate, including:

[0033] (1) Generate a uniform random number r8 between [0,1]. If r8 < 0.5, the u-th dimension of the velocity of the m-th particle in the g+1th iteration is in, is the u-th dimension of the local optimal quantum position of the m-th particle in the g-th iteration, is the u-th dimension of the global optimal quantum position of the particle swarm at the g-th iteration, c3 is the individual learning factor of the particle, c4 is the social learning factor of the particle, r9 and r 10 is a uniform random number between [0,1], ω is the inertia coefficient, and ω can be expressed as where ω max is the maximum value of inertia weight, ω min is the minimum value of the inertia weight. The u-th dimension of the m-th particle velocity is limited to where γ min is the minimum value of the particle velocity, γ max is the maximum value of the particle velocity; if r8≥0.5, the u-th dimension of the m-th particle velocity in the g+1th iteration is where r 11 is a uniform random number between [0, c5], c5 is a constant greater than 1, and a uniform random number r is generated between [0, 1] 12,like Generate a new particle with the same speed as the mth particle, and the uth dimension of the new particle position is Among them, r 13 is a uniform random number between [0,1], is the u-th dimension of the local optimal quantum position of the randomly selected s-th particle, and are coefficients that vary with the number of iterations and can be expressed as and Among them, c6 and c7 are constants between [0,1];

[0034] (2) The quantum rotation angle vector corresponding to the updated quantum position of the mth particle in the population is m=1,2,…,N2, the u-th dimension of the quantum rotation angle vector of the m-th particle is If r8≥0.5 and No. The u-th dimension of the quantum rotation angle vector corresponding to the updated quantum position of the new particle is

[0035] (3) Using the simplified quantum rotating gate to simulate the update of the quantum position of the mth particle is: The u-th dimension of the quantum position of the m-th particle is m=1,2,…,N2,u=1,2,…,U,if r8≥0.5 and Using a simplified quantum revolving door simulation to update The quantum position of the new particle is Among them The u-th dimension of the quantum position of the new particle is u=1,2,…,U, the quantum positions of all N2 particles in the population are updated in the U-dimensional search space according to the above rules.

[0036] Furthermore, in step 7, the fitness values of each particle and the new particle are calculated to determine the quantum position, local optimal quantum position, and global optimal quantum position of the new generation of particles, including:

[0037] The quantum position of the updated particle and the quantum position of the new particle are mapped to positions. The positions are assigned to the task and virtual machine allocation matrix Z, which is then fed into the fitness function to calculate the corresponding fitness value. The quantum position of the mth particle is As the quantum position of the mth particle in the g+1th iteration If r8≥0.5 and The quantum position of the mth particle The corresponding fitness value is The quantum position of a new particle Compare the corresponding fitness values and select the quantum position with the smaller fitness value as the quantum position of the mth particle in the g+1th iteration The quantum positions of all N2 particles are selected according to the above rules, and the quantum position of the mth particle is The corresponding fitness value and the local optimal quantum position of the mth particle in the previous generation The corresponding fitness values are compared and the quantum position with the smaller fitness value is selected as the local optimal quantum position of the mth particle in the g+1th iteration. m=1,2,…,N2, the local optimal quantum positions of all N2 particles are selected according to the above rules; the fitness values corresponding to the updated local optimal quantum positions of all particles are compared, and the local optimal quantum position with the smallest fitness value is the global optimal quantum position of the particle swarm in the g+1th iteration The corresponding minimum fitness value is The local optimal quantum position with the largest fitness value is The corresponding maximum fitness value is The fitness value corresponding to the global optimal quantum position of the particle swarm in the g+1 generation The fitness value corresponding to the optimal quantum position of the tunicate group By comparison, the one with the smaller fitness value is the optimal quantum position among all populations in the g+1 generation. The corresponding fitness value is

[0038] Furthermore, in step eight, information is exchanged between the particle swarm and the tunicate swarm, and then the tunicates in the tunicate swarm are reordered according to the fitness value, and the global optimal position of the particle swarm is re-determined, including:

[0039] The quantum position of the capsule with the largest fitness value in the g+1th iteration and its corresponding fitness value Replaced by the global optimal quantum position of the particle swarm and its corresponding fitness value The local optimal quantum position of the particle with the largest fitness value in the g+1th iteration and its corresponding fitness value Replaced by the optimal quantum position of the tunic group and its corresponding fitness value After information exchange, the capsules in the capsule group are rearranged in order from small to large according to their fitness values, and the quantum position of the capsule with the smallest fitness value is selected as the optimal quantum position of the capsule group in the g+1th iteration. The corresponding minimum fitness value is Then compare the local optimal quantum positions of the particle swarm again, and select the local optimal quantum position with the smallest fitness value as the global optimal quantum position of the particle swarm in the g+1th iteration. The corresponding minimum fitness value is

[0040] The present invention addresses the issue of existing cloud computing task scheduling methods, which often fail to consider the benefits of cloud computing service providers when optimizing for a single objective, such as time or cost, that users care about. This invention proposes a cloud computing task scheduling method based on the coevolutionary mechanism of quantum quilt swarms and particle swarms. This method employs a linear weighted objective function with a penalty mechanism and employs the coevolutionary mechanism of quantum quilt swarms and particle swarms to achieve multi-objective optimization. This method considers the time and cost concerns of users and the load imbalance concerns of cloud computing service providers, from the perspectives of satisfying user service quality and protecting the interests of cloud computing service providers. The method minimizes the total execution time, total execution cost, and load imbalance of tasks, keeping costs within budget and total task execution time within deadlines. The algorithm exhibits excellent convergence performance and, within the constraints of budget and deadline, achieves a relatively small total execution time, total execution cost, and load imbalance, overcoming the limitations of existing methods. This method addresses the problem that existing cloud computing task scheduling methods primarily optimize for a single objective, such as time or cost, while ignoring resource utilization, which is of concern to cloud computing service providers. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a schematic diagram of the cloud computing task scheduling method designed by the present invention based on the co-evolution mechanism of quantum quilt swarm and particle swarm.

[0042] Figure 2 This is a curve diagram of the convergence relationship between the fitness value and the number of iterations when the number of tasks is 100 and the number of virtual machines is 15.

[0043] Figure 3 This is a curve diagram of the convergence relationship between the fitness value and the number of iterations when the number of tasks is 1000 and the number of virtual machines is 15. DETAILED DESCRIPTION

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] Step 1: Establish the total execution time of the task, the total execution cost of the task and the load imbalance and build a mathematical model.

[0046] Define the allocation matrix of tasks and virtual machines as Among them, U is the number of tasks, V is the number of virtual machines, each task can only be assigned to one virtual machine for execution. If task u is assigned to virtual machine v for execution, then Z vu =1; if task u is not assigned to virtual machine v for execution, then Z vu =0, u = 1, 2, ..., U, v = 1, 2, ..., V. Establish the total execution time of the task, the total execution cost of the task and the load imbalance and construct a mathematical model. The specific process is as follows:

[0047] (1) Let the instruction length vector of the task be L = [L1, L2, ..., L U ], the execution speed vector of the virtual machine is K=[K1,K2,…,K V ], the execution time matrix of the task on the virtual machine is The execution time of task u on virtual machine v can be expressed as u=1,2,…,U,v=1,2,…,V,the execution time vector of the virtual machine is D=[D1,D2,…,D V ], where the execution time of the virtual machine v can be expressed as The total execution time of a task is the maximum execution time obtained after all virtual machines execute the assigned tasks. The total execution time of a task is Z vu ∈Z, u=1,2,…,U, v=1,2,…,V, where max represents the maximum value function. The task should be completed before the deadline, that is, the total execution time of the task The deadline λ cannot be exceeded. The deadline can be expressed as λ = D L +r1(D H -D L ), that is, satisfy the constraint Among them, D L The time obtained by assigning the tasks executed on the virtual machine with the longest execution time to the virtual machine with the fastest execution speed, D H The time obtained by assigning all tasks executed on the virtual machine with the longest execution time to the virtual machine with the slowest running speed, r1 is a uniform random number between [0,1].

[0048] (2) The execution cost matrix of the task on the virtual machine is: The unit execution cost matrix of the virtual machine is W = [W1, W2, ..., W V ], where the execution cost of task u on virtual machine v can be expressed as C vu =E vu W v , u=1,2,…,U,v=1,2,…,V,when all tasks are assigned to virtual machines for execution, the total cost is u=1,2,…,U,v=1,2,…,V,the total cost B(Z) of all tasks executed on the virtual machine should not exceed the budget β, which can be expressed as β=B min +r2(B max -B min ), that is, satisfying the constraint B(Z)≤β, where B max is the highest cost obtained by assigning all tasks to the virtual machine with the highest cost, B min It is the lowest cost obtained by assigning all tasks to the virtual machine with the lowest cost, and r2 is a uniform random number between [0,1].

[0049] (3) Load imbalance is Among them, D max is the maximum execution time of all virtual machines, D min is the minimum execution time of all virtual machines, is the average execution time of all virtual machines, and min represents the minimum function.

[0050] (4) The total execution time of the task The total execution cost B(Z) and load imbalance f3(Z) of the task are normalized, and the total execution time of the task is normalized to The total execution cost of the task is normalized to The load imbalance is normalized to Then the linear weighted method is used to design the objective function, which is Among them, α1, α2 and α3 are inertia weights, α1+α2+α3=1.

[0051] Step 2: Initialize the quantum positions of the quill search mechanism and the particle swarm evolution mechanism and set the parameters.

[0052] First, the population size of the quilt group is set to N1, the population size of the particle group is set to N2, the maximum number of iterations is G, the iteration number is labeled g, g∈[1,G], and the quantum position of the nth quilt in the U-dimensional search space at the g-th iteration is n=1,2,…,N1; the quantum position of the mth particle in the U-dimensional search space is The local optimal quantum position is m = 1, 2, ..., N2. In the first generation, let g = 1, initialize each dimension of the quantum position of the quilt group and the quantum position of the particle group to a uniform random number between [0, 1], and let the quantum position of the particle be the local optimal quantum position in the first generation.

[0053] Step 3: Calculate the fitness value of each tunicate, sort all tunicates, and calculate the fitness value of each particle to find the global optimal quantum position of the particle.

[0054] Mapping the quantum position of the quilt group to position, the position of the nth quilt in the g-th iteration is The mapping formula is n=1,2,…,N1, u=1,2,…,U, where, is the u-th dimension of the position of the n-th quilt at the g-th iteration, is the u-th dimension of the quantum position of the n-th capsule at the g-th iteration, is the upper limit of the u-th dimension of the tunicate position, is the lower limit of the u-th dimension of the tunicate position, Represents the nearest integer function. The value of each dimension of the quilt position coordinate is an integer in the range [1, V]. The position of each quilt is assigned to the task and virtual machine allocation matrix Z, where the allocation rule is: the value of the u-th dimension of the position of the n-th quilt in the g-th iteration is v, that is, n=1,2,…,N1,u=1,2,…,U,v={1,2,…,V},meaning that task u is assigned to virtual machine v. Then in the task and virtual machine assignment matrix Z, Z vu = 1 and the element of the u-th column of the matrix Z is divided by Z vu = 1, and the values of the remaining elements in the column are 0. The fitness function of the position of the nth quilt in the g-th iteration is The fitness values of the position and quantum position of the nth capsule in the gth iteration obtained by the fitness function can be obtained by Indicates that all quilts are arranged in order from small to large according to their fitness values, and the quantum position of the quilt with the smallest fitness value is defined as the optimal quantum position. The optimal quantum position in the quilt group in the g-th iteration is The corresponding minimum fitness value is Map the local optimal quantum position of the particle swarm to the local optimal position. The local optimal position of the mth particle in the gth iteration is The local optimal position is assigned to the allocation matrix Z of tasks and virtual machines. The fitness function of the local optimal position of the mth particle in the gth iteration is The fitness values of the local optimal position of the mth particle and the local optimal quantum position of the gth iteration obtained by the fitness function can be obtained by Indicates that the local optimal quantum position with the smallest fitness value in the g-th iteration particle swarm is the global optimal quantum position, and the global optimal quantum position of the g-th iteration particle swarm is The corresponding minimum fitness value is

[0055] Step 4: Generate quantum rotation angles based on the quill group search mechanism and use a simulated simplified quantum rotation gate to update the quantum position of the quill. The specific steps are as follows.

[0056] (1) The u-th dimension of the position of the n-th bladder in the g-th iteration is the water flow advection in the deep sea when the jet propulsion behavior is performed. Its own gravity is The conflicting forces between the tunicates are Where r3 is a uniform random number between [0,1]. After avoiding conflicts between quilts, the quilts will move towards the optimal quantum position. The position of the nth quilt after it moves towards the optimal quantum position is u=1,2,…,U,n=1,2,…,N1,for the position of the nth quilt after it moves towards the optimal quantum position, a uniform random number r4 is generated between [0,1]. If r4<0.5, the distance between the u-th dimension of the quantum position of the nth quilt and the u-th dimension of the quantum position of the r-th quilt randomly selected from the first n quilts is The u-th dimension of the position of the n-th capsule after it moves closer to the quantum position of the r-th capsule randomly selected from the first n capsules is If r4 ≥ 0.5, the distance between the u-th dimension of the quantum position of the n-th capsule and the u-th dimension of the optimal quantum position is The u-th dimension of the position of the nth capsule after it moves towards the optimal quantum position is in, is the u-th dimension of the optimal quantum position at the g-th iteration, is the u-th dimension of the quantum position of the r-th quilt randomly selected from the first n quilts in the g-th iteration, r∈{1,2,…,n}, r5 is a uniform random number between [0,c1], and c1 is a constant greater than 1.

[0057] (2) The quantum rotation angle vector corresponding to the updated quantum position of the nth tunicate in the population is n=1,2,…,N1. For the generation of the quantum rotation angle vector of the nth quill, a uniform random number r6 is generated between [0,1]. If r6<0.5, the u-th dimension of the quantum rotation angle of the nth quill in the g+1th iteration is n=1,2,…,N1,u=1,2,…,U;if r6≥0.5,the u-th dimension of the rotation angle of the n-th capsule quantum in the g+1th iteration is n=1,2,…,N1,u=1,2,…,U,r7 is a uniform random number between [2,c2], and c2 is a constant greater than 2.

[0058] (3) Using the simplified quantum rotating gate simulation, the quantum position of the nth capsule is updated as The u-th dimension of the quantum position of the n-th capsule is n=1,2,…,N1,u=1,2,…,U。 The quantum positions of all N1 capsules in the population are updated in the U-dimensional search space according to the above rules.

[0059] Step 5: Apply the greedy strategy to sort all tunicates according to their fitness values and determine the quantum position of the new generation of tunicate groups.

[0060] The updated quantum position of the capsule is mapped to the position, and the position is assigned to the task and the virtual machine allocation matrix Z. The fitness function is used to calculate the corresponding fitness value, and the updated quantum position is sorted from small to large according to the fitness value. The quantum position of the nth capsule after sorting is The corresponding fitness value and the quantum position of the nth capsule in the previous generation The corresponding fitness values are compared, and the quantum position of the capsule with the smaller fitness value is selected as the quantum position of the nth capsule in the g+1th iteration. n=1,2,…,N1, the quantum positions of all N1 quilts are selected according to the above rules. The quantum position of the quilt with the smallest fitness value is defined as the optimal quantum position. The optimal quantum position in the quilt group at the g+1th iteration is The corresponding minimum fitness value is The quantum position of the capsule with the largest fitness value is The corresponding maximum fitness value is

[0061] Step 6: Generate quantum rotation angles based on the particle swarm evolution mechanism and use a simulated simplified quantum rotation gate to update the quantum position of the particle. The specific steps are as follows.

[0062] (1) Generate a uniform random number r8 between [0,1]. If r8 < 0.5, the u-th dimension of the velocity of the m-th particle in the g+1th iteration is in, is the u-th dimension of the local optimal quantum position of the m-th particle in the g-th iteration, is the u-th dimension of the global optimal quantum position of the particle swarm at the g-th iteration, c3 is the individual learning factor of the particle, c4 is the social learning factor of the particle, r9 and r 10 is a uniform random number between [0,1], ω is the inertia coefficient, and ω can be expressed as where ω max is the maximum value of inertia weight, ω min is the minimum value of the inertia weight. The u-th dimension of the m-th particle velocity is limited to where γ min is the minimum value of the particle velocity, γ maxis the maximum value of the particle velocity; if r8≥0.5, the u-th dimension of the m-th particle velocity in the g+1th iteration is where r 11 is a uniform random number between [0, c5], c5 is a constant greater than 1, and a uniform random number r is generated between [0, 1] 12 ,like Generate a new particle with the same speed as the mth particle, and the uth dimension of the new particle position is Among them, r 13 is a uniform random number between [0,1], is the u-th dimension of the local optimal quantum position of the randomly selected s-th particle, and are coefficients that vary with the number of iterations and can be expressed as and Where c6 and c7 are constants between [0,1].

[0063] (2) The quantum rotation angle vector corresponding to the updated quantum position of the mth particle in the population is m=1,2,…,N2, the u-th dimension of the quantum rotation angle vector of the m-th particle is If r8≥0.5 and No. The u-th dimension of the quantum rotation angle vector corresponding to the updated quantum position of the new particle is

[0064] (3) Using the simplified quantum rotating gate to simulate the update of the quantum position of the mth particle is: The u-th dimension of the quantum position of the m-th particle is m=1,2,…,N2,u=1,2,…,U,if r8≥0.5 and Using a simplified quantum revolving door simulation to update The quantum position of the new particle is Among them The u-th dimension of the quantum position of the new particle is u=1,2,…,U. The quantum positions of all N2 particles in the population are updated in the U-dimensional search space according to the above rules.

[0065] Step 7: Calculate the fitness value of each particle and the new particle, and determine the quantum position, local optimal quantum position, and global optimal quantum position of the new generation of particles.

[0066] The quantum position of the updated particle and the quantum position of the new particle are mapped to positions. The positions are assigned to the task and virtual machine allocation matrix Z, which is then fed into the fitness function to calculate the corresponding fitness value. The quantum position of the mth particle is As the quantum position of the mth particle in the g+1th iteration If r8≥0.5 and The quantum position of the mth particle The corresponding fitness value is The quantum position of a new particle Compare the corresponding fitness values and select the quantum position with the smaller fitness value as the quantum position of the mth particle in the g+1th iteration The quantum positions of all N2 particles are selected according to the above rules. The quantum position of the mth particle is The corresponding fitness value and the local optimal quantum position of the mth particle in the previous generation The corresponding fitness values are compared and the quantum position with the smaller fitness value is selected as the local optimal quantum position of the mth particle in the g+1th iteration. m=1,2,…,N2, the local optimal quantum position of all N2 particles is selected according to the above rules. The fitness values corresponding to the local optimal quantum positions of all particles after the update are compared, and the local optimal quantum position with the smallest fitness value is the global optimal quantum position of the particle swarm in the g+1th iteration. The corresponding minimum fitness value is The local optimal quantum position with the largest fitness value is The corresponding maximum fitness value is The fitness value corresponding to the global optimal quantum position of the particle swarm in the g+1 generation The fitness value corresponding to the optimal quantum position of the tunicate group By comparison, the one with the smaller fitness value is the optimal quantum position among all populations in the g+1 generation. The corresponding fitness value is

[0067] Step 8: Determine whether g+1=5k1 is reached. If it is not reached, go directly to step nine. If it is reached, the particle group and the tuft group will exchange information and the quantum position of the tuft with the largest fitness value in the g+1th iteration will be and its corresponding fitness value Replaced by the global optimal quantum position of the particle swarm and its corresponding fitness value The local optimal quantum position of the particle with the largest fitness value in the g+1th iteration and its corresponding fitness value Replaced by the optimal quantum position of the tunic group and its corresponding fitness value After information exchange, the capsules in the capsule group are rearranged in order from small to large according to their fitness values, and the quantum position of the capsule with the smallest fitness value is selected as the optimal quantum position of the capsule group in the g+1th iteration. The corresponding minimum fitness value is Then compare the local optimal quantum positions of the particle swarm again, and select the local optimal quantum position with the smallest fitness value as the global optimal quantum position of the particle swarm in the g+1th iteration. The corresponding minimum fitness value is

[0068] Step 9: Determine whether the current number of iterations has reached the maximum number of iterations G. If not, set g = g + 1 and return to step 4. If it has, terminate the iteration loop and obtain the final task scheduling strategy based on the allocation matrix of tasks and virtual machines corresponding to the mapping positions of the optimal quantum positions in all populations in the last generation.

[0069] exist Figure 2 and Figure 3 In the paper, the cloud computing task scheduling method based on the quantum quilt swarm and particle swarm coevolution mechanism designed by the present invention is denoted as QPTSA-CCS, the cloud computing task scheduling method based on the particle swarm algorithm is denoted as PSO-CCS, the cloud computing task scheduling method based on the whale optimization algorithm is denoted as WOA-CCS, and the cloud computing task scheduling method based on the quilt swarm algorithm is denoted as TSA-CCS. The simulation experiment parameters are set as follows: N1=25, N2=25, G=500, c1=2, c2=3, c3=1, c4=1, c5=2,ω max =0.9,ω min =0.4, c6=0.9, c7=0.8. In the simulation experiment, the quantum tuft swarm and particle swarm co-evolution mechanism uses the same population initial parameters as the particle swarm algorithm, whale optimization algorithm, and tuft swarm algorithm. Other parameter settings of the particle swarm algorithm refer to "Task Scheduling using Modified PSO Algorithm in Cloud Computing Environment" published by Solmaz Abdi et al., other parameter settings of the whale optimization algorithm refer to "The Whale Optimization Algorithm" published by Seyedali Mirjalili et al., and other parameter settings of the tuft swarm algorithm refer to "Tunicate Swarm Algorithm: A new bio-inspired based metaheuristic paradigm for global optimization" published by Satnam Kaur et al. In order to overcome the contingency, 100 independent experiments were conducted on the four algorithms, and the average value was taken. The cloud computing task scheduling method based on the quantum quilt group and particle swarm co-evolution mechanism was compared with the cloud computing task scheduling method based on the particle swarm algorithm, the cloud computing task scheduling method based on the whale optimization algorithm and the cloud computing task scheduling method based on the quilt group algorithm. The convergence curves of the fitness values of the quantum quilt group and particle swarm co-evolution mechanism, the particle swarm algorithm, the whale optimization algorithm and the quilt group algorithm were obtained as shown in the figure. Figure 2 and Figure 3 .

[0070] from Figure 2 and Figure 3 It can be seen from the results that the cloud computing task scheduling method based on the quantum quilt swarm and particle swarm co-evolution mechanism designed by the present invention is effective and superior. The fitness values obtained by the quantum quilt swarm and particle swarm co-evolution mechanism under different task numbers are all smaller than the fitness values obtained by the particle swarm algorithm, the whale optimization algorithm and the quilt swarm algorithm. Under the constraints of cost and deadline, the cloud computing task scheduling method based on the quantum quilt swarm and particle swarm co-evolution mechanism reduces the total execution time of the task, the total execution cost of the task and the load imbalance, achieves better task scheduling effect, and improves the utilization rate of cloud computing resources, which proves the effectiveness and superiority of the cloud computing task scheduling method based on the quantum quilt swarm and particle swarm co-evolution mechanism.

Claims

1. A cloud computing task scheduling method, characterized in that: The following steps are involved: Step 1: Establish the total execution time of the task, the total execution cost of the task and the load imbalance and construct a mathematical model; Step 2: Initialize the quantum positions of the quill search mechanism and the particle swarm evolution mechanism and set the parameters; Step 3: Calculate the fitness value of each tunicate, sort all tunicates, and calculate the fitness value of each particle to find the global optimal quantum position of the particle; Step 4: Generate quantum rotation angles based on the quill group search mechanism and use a simulated simplified quantum rotation gate to update the quill's quantum position; Step 5: Apply the greedy strategy to sort all tunicates according to their fitness values and determine the quantum position of the new generation of tunicate groups; Step 6: Generate quantum rotation angles based on the particle swarm evolution mechanism and use a simulated simplified quantum rotation gate to update the quantum position of the particle; Step 7: Calculate the fitness value of each particle and the new particle, and determine the quantum position, local optimal quantum position, and global optimal quantum position of the new generation of particles; Step 8: Determine whether g+1=5k1 is reached. If it is not reached, then directly execute step nine. If it is reached, then the particle swarm and the tunic swarm are exchanged with each other, and then the tunic swarms in the tunic swarm are re-sorted according to the fitness value, and the global optimal position of the particle swarm is re-determined. Step 9: Determine whether the current number of iterations has reached the maximum number of iterations G. If not, set g = g + 1 and return to step 4. If it is reached, the iterative loop is terminated, and the final task scheduling strategy is obtained according to the allocation matrix of tasks and virtual machines corresponding to the mapping positions of the optimal quantum positions in all populations in the last generation; The total execution time, total execution cost and load imbalance of the task are established and the mathematical model is constructed as follows: The allocation matrix between tasks and virtual machines is defined as: Among them, U is the number of tasks, V is the number of virtual machines, each task can only be assigned to one virtual machine for execution. If task u is assigned to virtual machine v for execution, then Z vu =1; if task u is not assigned to virtual machine v for execution, then Z vu =0, u=1,2,…,U, v=1,2,…,V.

2. A cloud computing task scheduling method according to claim 1, characterized in that: The step 1 is specifically as follows: (1) Let the instruction length vector of the task be L = [L1, L2, ..., L U ], the execution speed vector of the virtual machine is K=[K1,K2,…,K V ], the execution time matrix of the task on the virtual machine is The execution time of task u on virtual machine v is expressed as u=1,2,…,U,v=1,2,…,V,the execution time vector of the virtual machine is D=[D1,D2,…,D V ], where the execution time of the virtual machine v can be expressed as The total execution time of a task is the maximum execution time obtained after all virtual machines execute the assigned tasks. The total execution time of a task is Z vu ∈Z, u=1,2,…,U, v=1,2,…,V, where max represents the maximum value function, satisfying the constraint λ is the cutoff time, expressed as λ = D L +r1(D H -D L ), D L The time obtained by assigning the tasks executed on the virtual machine with the longest execution time to the virtual machine with the fastest execution speed, D H The time obtained by assigning all tasks executed on the virtual machine with the longest execution time to the virtual machine with the slowest running speed, r1 is a uniform random number between [0,1]; (2) The execution cost matrix of the task on the virtual machine is: The unit execution cost matrix of the virtual machine is W = [W1, W2, ..., W V ], where the execution cost of task u on virtual machine v is expressed as C vu =E vu W v , u=1,2,…,U,v=1,2,…,V,when all tasks are assigned to virtual machines for execution, the total cost is u=1,2,…,U,v=1,2,…,V,satisfy the constraint B(Z)≤β, and the budget β is expressed as: β=B min +r2(B max -B min ), where B max is the highest cost obtained by assigning all tasks to the virtual machine with the highest cost, B min It is the lowest cost obtained by assigning all tasks to the virtual machine with the lowest cost, and r2 is a uniform random number between [0,1]. (3) Load imbalance is Among them, D max is the maximum execution time of all virtual machines, D min is the minimum execution time of all virtual machines, is the average execution time of all virtual machines, and min represents the minimum function; (4) The total execution time of the task The total execution cost B(Z) and load imbalance f3(Z) of the task are normalized, and the total execution time of the task is normalized to The total execution cost of the task is normalized to The load imbalance is normalized to Then the linear weighted method is used to design the objective function, which is Among them, α1, α2 and α3 are inertia weights, α1+α2+α3=1.

3. The cloud computing task scheduling method according to claim 1, characterized in that: The initialization of the quantum positions of the quill search mechanism and the particle swarm evolution mechanism and setting of the parameters in step 2 include: First, the population size of the quilt group is set to N1, the population size of the particle group is set to N2, the maximum number of iterations is G, the iteration number is labeled g, g∈[1,G], and the quantum position of the nth quilt in the U-dimensional search space at the g-th iteration is n=1,2,…,N1; the quantum position of the mth particle in the U-dimensional search space is The local optimal quantum position is m=1,2,…,N2; in the first generation, let g=1, initialize each dimension of the quantum position of the capsule group and the quantum position of the particle group to a uniform random number between [0,1], and let the quantum position of the particle be the local optimal quantum position in the first generation.

4. The cloud computing task scheduling method according to claim 1, wherein: Step 3 calculates the fitness value of each tunicate, sorts all tunicates, and calculates the fitness value of each particle to find the global optimal quantum position of the particle. Mapping the quantum position of the quilt group to position, the position of the nth quilt in the g-th iteration is The mapping formula is n=1,2,…,N1, u=1,2,…,U, where, is the u-th dimension of the position of the n-th quilt at the g-th iteration, is the u-th dimension of the quantum position of the n-th capsule at the g-th iteration, is the upper limit of the u-th dimension of the tunicate position, is the lower limit of the u-th dimension of the tunicate position, Represents the nearest integer function; the value of each dimension of the quilt's position coordinate is an integer in the range [1, V]. The position of each quilt is assigned to the task and virtual machine allocation matrix Z. The assignment rule is: the value of the u-th dimension of the position of the n-th quilt in the g-th iteration is v, that is, n=1,2,…,N1,u=1,2,…,U,v={1,2,…,V},meaning that task u is assigned to virtual machine v. Then in the task and virtual machine assignment matrix Z, Z vu = 1 and the element of the u-th column of the matrix Z is divided by Z vu = 1, and the values of the remaining elements in the column are 0; the fitness function of the position of the nth quilt in the g-th iteration is The fitness values of the position and quantum position of the nth capsule in the gth iteration obtained by the fitness function can be obtained by Indicates that all quilts are arranged in order from small to large according to their fitness values, and the quantum position of the quilt with the smallest fitness value is defined as the optimal quantum position. The optimal quantum position in the quilt group in the g-th iteration is The corresponding minimum fitness value is Map the local optimal quantum position of the particle swarm to the local optimal position. The local optimal position of the mth particle in the gth iteration is The local optimal position is assigned to the allocation matrix Z of tasks and virtual machines. The fitness function of the local optimal position of the mth particle in the gth iteration is The fitness values of the local optimal position of the mth particle and the local optimal quantum position of the gth iteration obtained by the fitness function can be obtained by Indicates that the local optimal quantum position with the smallest fitness value in the g-th iteration particle swarm is the global optimal quantum position, and the global optimal quantum position of the g-th iteration particle swarm is The corresponding minimum fitness value is 5. The cloud computing task scheduling method according to claim 1, characterized in that: In step 4, the quantum rotation angle is generated according to the quill group search mechanism, and the quantum position of the quill is updated using the simulated simplified quantum rotation gate, which includes: (1) The u-th dimension of the position of the n-th bladder in the g-th iteration is the water flow advection in the deep sea when the jet propulsion behavior is performed. Its own gravity is The conflicting forces between the tunicates are Where r3 is a uniform random number between [0,1]. After avoiding conflicts between quilts, the quilt will move towards the optimal quantum position. The position of the nth quilt after it moves towards the optimal quantum position is u=1,2,…,U,n=1,2,…,N1,for the position of the nth quilt after it moves towards the optimal quantum position, a uniform random number r4 is generated between [0,1]. If r4<0.5, the distance between the u-th dimension of the quantum position of the nth quilt and the u-th dimension of the quantum position of the r-th quilt randomly selected from the first n quilts is The u-th dimension of the position of the n-th capsule after it moves closer to the quantum position of the r-th capsule randomly selected from the first n capsules is If r4 ≥ 0.5, the distance between the u-th dimension of the quantum position of the n-th capsule and the u-th dimension of the optimal quantum position is The u-th dimension of the position of the nth capsule after it moves towards the optimal quantum position is in, is the u-th dimension of the optimal quantum position at the g-th iteration, is the u-th dimension of the quantum position of the r-th quilt randomly selected from the first n quilts in the g-th iteration, r∈{1,2,…,n}, r5 is a uniform random number between [0,c1], and c1 is a constant greater than 1; (2) The quantum rotation angle vector corresponding to the updated quantum position of the nth tunicate in the population is n=1,2,…,N1; for the generation of the quantum rotation angle vector of the nth quilt, a uniform random number r6 is generated between [0,1]; if r6<0.5, the u-th dimension of the quantum rotation angle of the nth quilt in the g+1th iteration is n=1,2,…,N1,u=1,2,…,U;if r6≥0.5,the u-th dimension of the rotation angle of the n-th capsule quantum in the g+1th iteration is n=1,2,…,N1,u=1,2,…,U,r7 is a uniform random number between [2,c2], and c2 is a constant greater than 2; (3) Using the simplified quantum rotating gate simulation, the quantum position of the nth capsule is updated as The u-th dimension of the quantum position of the n-th capsule is n=1,2,…,N1,u=1,2,…,U;the quantum positions of all N1 capsules in the population are updated in the U-dimensional search space according to the above rules.

6. The cloud computing task scheduling method according to claim 1, characterized in that: In step 5, the greedy strategy is applied to sort all tunicates according to their fitness values, and the quantum positions of the new generation of tunicates are determined as follows: The updated quantum position of the capsule is mapped to the position, and the position is assigned to the task and the virtual machine allocation matrix Z. The fitness function is used to calculate the corresponding fitness value, and the updated quantum position is sorted from small to large according to the fitness value. The quantum position of the nth capsule after sorting is The corresponding fitness value and the quantum position of the nth capsule in the previous generation The corresponding fitness values are compared, and the quantum position of the capsule with the smaller fitness value is selected as the quantum position of the nth capsule in the g+1th iteration. n=1,2,…,N1, the quantum positions of all N1 quilts are selected according to the above rules, and the quantum position of the quilt with the smallest fitness value is defined as the optimal quantum position. The optimal quantum position in the quilt group in the g+1th iteration is The corresponding minimum fitness value is The quantum position of the capsule with the largest fitness value is The corresponding maximum fitness value is 7. The cloud computing task scheduling method according to claim 1, characterized in that: In step 6, the quantum rotation angle is generated according to the particle swarm evolution mechanism, and the quantum position of the particle is updated using a simulated simplified quantum rotation gate, including: (1) Generate a uniform random number r8 between [0,1]. If r8 < 0.5, the u-th dimension of the velocity of the m-th particle in the g+1th iteration is in, is the u-th dimension of the local optimal quantum position of the m-th particle in the g-th iteration, is the u-th dimension of the global optimal quantum position of the particle swarm at the g-th iteration, c3 is the individual learning factor of the particle, c4 is the social learning factor of the particle, r9 and r 10 is a uniform random number between [0,1], ω is the inertia coefficient, and ω can be expressed as where ω max is the maximum value of inertia weight, ω min is the minimum value of the inertia weight, and the u-th dimension of the m-th particle velocity is limited to where γ min is the minimum value of the particle velocity, γ max is the maximum value of the particle velocity; if r8≥0.5, the u-th dimension of the m-th particle velocity in the g+1th iteration is where r 11 is a uniform random number between [0, c5], c5 is a constant greater than 1, and a uniform random number r is generated between [0, 1] 12 ,like Generate a new particle with the same speed as the mth particle, and the uth dimension of the new particle position is Among them, r 13 is a uniform random number between [0,1], is the u-th dimension of the local optimal quantum position of the randomly selected s-th particle, and are coefficients that vary with the number of iterations and can be expressed as and Among them, c6 and c7 are constants between [0,1]; (2) The quantum rotation angle vector corresponding to the updated quantum position of the mth particle in the population is m=1,2,…,N2, the u-th dimension of the quantum rotation angle vector of the m-th particle is If r8≥0.5 and No. The u-th dimension of the quantum rotation angle vector corresponding to the updated quantum position of the new particle is (3) Using the simplified quantum rotating gate to simulate the update of the quantum position of the mth particle is: The u-th dimension of the quantum position of the m-th particle is m=1,2,…,N2,u=1,2,…,U,if r8≥0.5 and Using a simplified quantum revolving door simulation to update The quantum position of the new particle is Among them The u-th dimension of the quantum position of the new particle is u=1,2,…,U, the quantum positions of all N2 particles in the population are updated in the U-dimensional search space according to the above rules.

8. The cloud computing task scheduling method according to claim 1, wherein: Calculating the fitness value of each particle and the new particle in step 7, and determining the quantum position, local optimal quantum position, and global optimal quantum position of the new generation of particles include: The quantum position of the updated particle and the quantum position of the new particle are mapped to positions. The positions are assigned to the task and virtual machine allocation matrix Z, which is then fed into the fitness function to calculate the corresponding fitness value. The quantum position of the mth particle is As the quantum position of the mth particle in the g+1th iteration If r8≥0.5 and The quantum position of the mth particle The corresponding fitness value is The quantum position of a new particle Compare the corresponding fitness values and select the quantum position with the smaller fitness value as the quantum position of the mth particle in the g+1th iteration The quantum positions of all N2 particles are selected according to the above rules, and the quantum position of the mth particle is The corresponding fitness value and the local optimal quantum position of the mth particle in the previous generation The corresponding fitness values are compared and the quantum position with the smaller fitness value is selected as the local optimal quantum position of the mth particle in the g+1th iteration. m=1,2,…,N2, the local optimal quantum positions of all N2 particles are selected according to the above rules; the fitness values corresponding to the updated local optimal quantum positions of all particles are compared, and the local optimal quantum position with the smallest fitness value is the global optimal quantum position of the particle swarm in the g+1th iteration The corresponding minimum fitness value is The local optimal quantum position with the largest fitness value is The corresponding maximum fitness value is The fitness value corresponding to the global optimal quantum position of the particle swarm in the g+1 generation The fitness value corresponding to the optimal quantum position of the tunicate group By comparison, the one with the smaller fitness value is the optimal quantum position among all populations in the g+1 generation. The corresponding fitness value is 9. The cloud computing task scheduling method according to claim 1, characterized in that: Step 8, in which information is exchanged between the particle swarm and the tunicate swarm, and the tunicates in the tunicate swarm are reordered according to their fitness values, and the global optimal position of the particle swarm is re-determined, includes: The quantum position of the capsule with the largest fitness value in the g+1th iteration and its corresponding fitness value Replaced by the global optimal quantum position of the particle swarm and its corresponding fitness value The local optimal quantum position of the particle with the largest fitness value in the g+1th iteration and its corresponding fitness value Replaced by the optimal quantum position of the tunic group and its corresponding fitness value After information exchange, the capsules in the capsule group are rearranged in order from small to large according to their fitness values, and the quantum position of the capsule with the smallest fitness value is selected as the optimal quantum position of the capsule group in the g+1th iteration. The corresponding minimum fitness value is Then compare the local optimal quantum positions of the particle swarm again, and select the local optimal quantum position with the smallest fitness value as the global optimal quantum position of the particle swarm in the g+1th iteration. The corresponding minimum fitness value is

Images