Control method and system based on region segmentation
Control method and system based on region segmentation
 CN 106,408,082 A
 Filed: 09/23/2016
 Published: 02/15/2017
 Est. Priority Date: 09/23/2016
 Status: Active Application
First Claim
1. a kind of control method based on region segmentation is it is characterised in that methods described includes：
 Obtain region of search S and object function f_{it}；
Obtain the iterationses setting, correlation coefficient and region segmentation border, described correlation coefficient includes the iterationses setting；
The Euclidean distance of described object function particle history optimal location and global optimum position is obtained in described region of search,Particle inertia weight is calculated according to described Euclidean distance；
Particle rapidity is updated according to described inertia weight；
Particle position is updated according to described particle rapidity；
Region segmentation is carried out according to the particle rapidity after updating and the particle position after renewal；
Judge whether the particle after region segmentation meets TSP question condition；
If being unsatisfactory for, return the described step that particle rapidity is updated according to inertia weight；
If meeting, TSP question is carried out to described particle；
Speed according to the particle after described TSP question and location updating individuality history optimal solution pbest and colony'"'"'s history areExcellent solution gbest；
Judge whether current iterationses are less than the iterationses value of described setting；
If so, described current iteration number of times is added 1, return the described step that particle rapidity is updated according to inertia weight；
If it is not, globally optimal solution is calculated according to the individual history optimal solution after updating and colony'"'"'s history optimal solution.
Chinese PRB Reexamination
Abstract
The invention discloses a control method and a control system based on region segmentation. The control method comprises the steps of: acquiring a Euclidean distance between a historical optimal position of a function particle and a global optimal position in a searching region, and calculating a particle inertia weight according to the Euclidean distance; updating particle speeds and particle positions according to the inertia weight; conducting region segmentation according to the updated particle speeds and particle positions; subjecting particles after region segmentation to selfadaptive mutation; and finally acquiring a global optimal solution. By adopting the control method, the exponential declined inertia weight of the single particle is designed, then the region segmentation is conducted through information cross optimization, the selfadaptive mutation of the particle is carried out by means of a Cauchy operator, and the convergence speed is accelerated and the convergence precision is improved on the premise of keeping low computation complexity. The algorithm can be applied to wireless sensor network energy management, the optimal solution of the objective function is obtained, the power consumption of wireless sensor nodes in the InternetofThings system is reduced, and the service cycle of the wireless sensor nodes is prolonged.

3 Citations
A kind of nearinfrared method for detecting human face based on improvement Multiple Kernel Learning frame  
Patent #
CN 108,427,915 A
Filed 02/07/2018

Current Assignee

Photovoltaic cell multipeak maximum power tracking method and system based on population  
Patent #
CN 109,814,651 A
Filed 01/21/2019

Current Assignee

The improvement population MPPT algorithm of condition is restarted based on dynamic inertia weight and multithreshold  
Patent #
CN 108,170,200 A
Filed 01/03/2018

Current Assignee

No References
9 Claims

1. a kind of control method based on region segmentation is it is characterised in that methods described includes：

Obtain region of search S and object function f_{it}； Obtain the iterationses setting, correlation coefficient and region segmentation border, described correlation coefficient includes the iterationses setting； The Euclidean distance of described object function particle history optimal location and global optimum position is obtained in described region of search,Particle inertia weight is calculated according to described Euclidean distance； Particle rapidity is updated according to described inertia weight； Particle position is updated according to described particle rapidity； Region segmentation is carried out according to the particle rapidity after updating and the particle position after renewal； Judge whether the particle after region segmentation meets TSP question condition； If being unsatisfactory for, return the described step that particle rapidity is updated according to inertia weight； If meeting, TSP question is carried out to described particle； Speed according to the particle after described TSP question and location updating individuality history optimal solution pbest and colony'"'"'s history areExcellent solution gbest； Judge whether current iterationses are less than the iterationses value of described setting； If so, described current iteration number of times is added 1, return the described step that particle rapidity is updated according to inertia weight； If it is not, globally optimal solution is calculated according to the individual history optimal solution after updating and colony'"'"'s history optimal solution.


2. method according to claim 1 is it is characterised in that described acquisition correlation coefficient specifically includes：
 Obtain coefficient a,B, c, d, wherein a=0.5, b=50, c=0.35, d=0.8.

3. method according to claim 1 is it is characterised in that described calculating inertia weight specifically includes：
 Using formulaCalculate inertia weight；
Wherein,L_{ij}For jth particle history optimal location in ith iteration and global optimum positionThe Euclidean distance put, D represents region of search dimension, g_{k}Represent component in k dimension for the global optimum position, p_{jk}For jth particleComponent in k dimension for the history optimal location, ω
_{max}For 0.9, ω
_{min}For 0.4, L_{max}Represent region of search all particle Euclideans away fromFrom maximum, t_{i}For current iteration number of times, t_{max}For the iterationses setting.
 Using formulaCalculate inertia weight；

4. method according to claim 1 is it is characterised in that described specifically wrap according to inertia weight renewal particle rapidityInclude：

Using formula v_{ij}(t+1)=ω
_{ij}v_{ij}(t)+c_{1}r_{1}(pbest_{ij}(t)x_{ij}(t))+c_{2}r_{2}(gbest_{g}(t)x_{ij}(t)) more new particle speedDegree, wherein t is iterationses, c_{1}For itself aceleration pulse, c_{2}For overall aceleration pulse, r_{1}And r_{2}It is random between [0,1]Number, v_{ij}Component in j dimension for the speed of particle i, v in the t time iteration of (t) expression_{ij}(t+1) represent particle in the t+1 time iterationComponent in j dimension for the speed of i, ω
_{ij}For inertia weight, pbest_{ij}T () is that the individual history of particle i in the t time iteration is optimumSolution, gbest_{g}T () is colony'"'"'s history optimal solution in t iteration, x_{ij}T () is the position of particle i in the t time iteration in j dimensionComponent.


5. method according to claim 1 is it is characterised in that described renewal particle position specifically includes：
 Using formula x_{ij}(t+1)=x_{ij}(t)+v_{ij}(t+1) particle position, wherein v are updated_{ij}(t+1) represent the speed of particle i in the t+1 time iteration in jComponent in dimension, x_{ij}Component in j dimension for the position of particle i, x in the t time iteration of (t) expression_{ij}(t+1) represent the t+1 time repeatedlyComponent in j dimension for the position of particle i in generation.

6. method according to claim 1 is it is characterised in that the described region segmentation that carries out specifically includes：

(1) in described region of search, the Euclidean distance according to particle to global optimum position is ranked up, and is set with Euclidean distanceDefinite value is boundary, and particle search region is divided into two parts：
Boundary inner region and boundary exterior domain, the particle definition of boundary inner regionFor the first particle, the particle of boundary exterior domain is defined as the second particle；(2) two particle j and k are selected from described first particle, using formula V_{i}=aV_{j}+(1a)V_{k}And X_{i}=aX_{j}+(1a)X_{k}Carry out crossover operation, will the speed of particle j and particle k and position be intersected by default weight proportion, generate one newThe speed of particle i and positional information, wherein V_{j}For the speed of particle j, V_{k}For the speed of particle k, V_{i}For the speed of new particle i, X_{j}For the position coordinateses of particle j, X_{k}For the position coordinateses of particle k, X_{i}For the position coordinateses of new particle i, a is constant, [0,1] itBetween； (3) use the information of information one described second particle of replacement of new particle i； (4) execution step (2) and (3) are substituted until all described second particles, thus forming new population； (5) execution step (1)～
(4) repeatedly, repeated segmentation region of search reaches preset value to segmentation times.


7. method according to claim 1 is it is characterised in that described judge whether particle meets TSP question condition toolBody includes：

Judge whether particle j at least meets one of following condition： (1) colony'"'"'s history optimal solution gbest is all no improved in continuous b iteration, and the ideal values of b are：
Meet TSP question condition if meeting and being particle；(2) distance function s (l) of particle j and global optimum particle k meetsIf meet being grainGestational edema foot TSP question condition, whereinS (l) is the distance of particle j and global optimum particle kFunction, l_{kj}For the Euclidean distance between particle j and global optimum particle k, n is particle number, and c is the constant between [0,1], DFor region of search.


8. method according to claim 1 is it is characterised in that described carry out TSP question to described particle and specifically wrapInclude：
Using formulaWithTo particle jCarry out TSP question, whereinV_{j1}, it is speed before particle j variation, V_{j2}ForSpeed after particle j variation, X_{j1}For particle j variation front position coordinate, X_{j1}For position coordinateses after particle j variation, rand is [0,1]Between random value generating function, for produce one [0,1] random value.

9. a kind of system based on region segmentation is it is characterised in that described system includes：

Region of search and object function acquisition module, for obtaining region of search S and object function f_{it}； Iterationses, correlation coefficient and the region segmentation border acquisition module setting, for obtaining the iterationses setting, correlationCoefficient and region segmentation border； Distance calculation module, for calculating described object function particle history optimal location with the overall situation in described region of searchThe Euclidean distance of excellent position； Inertia weight computing module, for calculating inertia weight； Particle rapidity update module, for updating particle rapidity according to described inertia weight； Particle position update module, for updating particle position； Region segmentation module, is used for carrying out region segmentation； TSP question conditional judgment module, for judging whether particle meets TSP question condition； TSP question module, for when meeting described TSP question condition, carrying out TSP question to described particle； Optimal solution update module, for more new individual history optimal solution pbest and colony'"'"'s history optimal solution gbest； Whether iterationses judge module, for judging current iteration number of times less than the iterationses value setting； Iterationses control module, for when current iteration number of times is less than the iterationses value setting, by described current iterationNumber of times adds 1； Globally optimal solution computing module, for when current iteration number of times is not less than the iterationses value setting, calculating the overall situationExcellent solution.

Specification(s)