CN103856952A - Method for optimizing Pareto multiple target deployment of industrial wireless sensor network - Google Patents
Method for optimizing Pareto multiple target deployment of industrial wireless sensor network Download PDFInfo
- Publication number
- CN103856952A CN103856952A CN201410068269.3A CN201410068269A CN103856952A CN 103856952 A CN103856952 A CN 103856952A CN 201410068269 A CN201410068269 A CN 201410068269A CN 103856952 A CN103856952 A CN 103856952A
- Authority
- CN
- China
- Prior art keywords
- node
- leader cluster
- solution
- grid
- cluster node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention discloses a method for optimizing Pareto multiple target deployment of an industrial wireless sensor network, and relates to the field of industrial automation and intelligent computing. A new communication model is defined for the two-layer cluster structure industrial wireless sensor network commonly used in the industry, the method for optimizing the Pareto multiple target deployment is provided, the cluster head node deploying position is optimized through the Pareto human study optimization algorithm, and the multi-objective optimization of reliability, real-time performance, economical efficiency and expansibility of the industrial wireless sensor network is achieved by providing redundant cluster heads, optimizing and limiting the hop count to a base station from a sensor node, reducing the number of the cluster heads, balancing the cluster head charge number, and reserving network access ports needed by mobile nodes and upgrading and maintaining of a system.
Description
Technical field
The present invention relates to industrial automation and the large field of intelligent computation two, be specifically related to a kind of industrial wireless sensor network Pareto multiple target and dispose optimization method.
Background technology
Now, in order to improve process control efficiency, observe environmental regulations and to meet the demand of economic goal, industrial automation system is widely used in large scale industry system.In the past few decades, wired industrial communication, such as field bus system and wired HART, in factory automation and the extensive use of process automation field, has improved productivity and the efficiency of industrial system.But mounting arrangements cable is still very difficult and spend high in severe industrial environment.For industrial remote monitoring, the cost of mounting industrial cable is even taller than the cost of transducer itself especially.Meanwhile, due to the wiring restriction of traditional wired system, it is difficult to be applied to the detection of industrial target rotation.Therefore the application prospect of wireless industrial system is very tempting, industrial wireless communication make the setting of factory automation system become more easily, cheaper, more flexible.Due to the huge market application foreground of industrial wireless sensor network, world automation Heat & Control Pty Ltd., as Honeywell, Emerson, Siemens, ABB, Roger's Weir and YOKOGAWA, develop industry wireless sensor network system solution and relevant wireless product separately based on different wireless technologys such as Zigbee, wirelessHART, ISA100.11a.
Industrial wireless sensor network has become the research and application focus of industrial automation and control field.At present in practical application in industry, for the ease of management, assurance real-time, industrial wireless sensor network adopts double-layer separate clustering architecture conventionally, in double-layer separate clustering architecture, conventionally comprise two category nodes: sensor node and leader cluster node, wherein sensor node is responsible for the collection of field data, and the data that leader cluster node receiving sensor node sends over also send to base station by Data Integration.In practical application, industrial system needs industrial wireless sensor network to guarantee absolute reliability and the real-time of transfer of data, working properly to guarantee control system, avoids causing casualties, the production accident such as blast, loss of material.But in industrial environment, ubiquitous electromagnetic pollution, high burn into high humility, vibrations and dust etc. all may make sensor node break down, thereby interrupt network connects.For large-scale industrial application, the data of the sensor node of One's name is legion cooperation collection and propagation delay sensitivity within the fixing short period of being everlasting, make to guarantee that the reliability of data and real-time are more difficult.In addition, in long-term commercial Application, the node in sensor network also inevitably can produce fault.At present, researcher has proposed to comprise improved coding strategy, MAC agreement, ARQ agreement, the multiple strategy such as method for routing and dispatching algorithm is to improve real-time and the reliability of industrial wireless sensor network, but for large-scale industrial wireless sensor network application, it comprises node may cross ten thousand, only by improving coded system, MAC agreement and route are far from being enough, also must carry out from the angle of system the deployment of global optimization node, the carefully topological structure of design industrial wireless sensor network, this is the basis that guarantees whole industrial wireless sensing network system reliability and real-time.
In addition, sensor node generally all adopts powered battery, finite energy, and the system in process industry, once node energy exhausts, must change battery.If it is improper to dispose design, the battery loss difference of different nodes, bunch head is excessive, and unified replacing can cause unnecessary waste, will cause carrying out continually battery altering according to actual electric weight, and this obviously can increase the workload of network operation.For large-scale industry system, need the variable of detection numerous, often reach points up to ten thousand, therefore, need to optimize and dispose bunch head to reduce cost of investment, simplified network structure, is convenient to administer and maintain.Meanwhile, have large-scale hardware in actual industrial scene, these equipment can have a strong impact on the normal communication between node, therefore in node deployment, must consider that possible signal disturbs and shielding.In addition, in practical application, also there is the demands such as mobile node access industrial wireless sensor network, system upgrade expansion.For example, when plant maintenance is overhauled, special equipment needs interim access current network; When system reform upgrading, newly-increased measuring control point also needs to be linked in existing industry wireless sensor network system.In order not affect the normal work of industrial wireless sensor network, when building, industrial wireless sensor network just need to consider the access of mobile node and newly-increased node.These practical application request are disposed large-scale industry wireless sensor network node becomes the difficult optimization problem of multiple target NP, and exist the dimension of multiple optimization aim may be not identical, in practical application industrial wireless sensor network scale and application demand also different, be difficult to provide exactly the problems such as weight between each optimization aim.
Dispose in design based on cluster structured industrial wireless sensor network, lower floor's bunch inner sensor node only with this bunch of leader cluster node communication, leader cluster node is collected bunch data that inner sensor node sends and be sent to base station by all the other leader cluster nodes in the mode of multi-hop after data processing, and its structure as shown in Figure 1.Sensor node in Fig. 1 in 4 leader cluster nodes all only sends to this bunch of leader cluster node by the field data data that collect, and leader cluster node is sent to base station by all the other leader cluster nodes with the form of multi-hop by the data after data processing again.But, in the design of traditional industry wireless sensor network disposition method, sensor node or leader cluster node are deployed on grid intersection, while making actual deployment, do not allow node location to have deviation, increase the poor robustness in difficulty and the practical application of industrial wireless sensor network node actual installation, therefore do not met actual industrial wireless sensor network application demand.The present invention has defined a kind of new deployment model, any point that in new model, the position of setting sensor node and leader cluster node can be in grid, thus while making actual deployment, allow node location to have certain deviation, more realistic application characteristic.Meanwhile, the present invention adopts a kind of Pareto multiple target mankind to learn optimized algorithm and solves multiple target industrial wireless sensor network node deployment optimization problem, makes engineer to select ideal optimal solution according to practical application request.
Summary of the invention
The object of the invention is to the deficiency existing for prior art, provide a kind of industrial wireless sensor network Pareto multiple target to dispose optimization method, the method adopts the Pareto multiple target mankind to learn optimized algorithm the leader cluster node deployed position in industrial wireless sensor network is optimized, guaranteeing on the basis of system reliability and real-time, can further reduce construction cost and the maintenance cost of system, the difficulty that reduces industrial wireless sensor network node actual installation, has improved the robustness of practical application.
For achieving the above object, the present invention adopts following technical proposals:
A kind of industrial wireless sensor network Pareto multiple target is disposed optimization method, it is characterized in that the concrete steps of the method are as follows:
(1), first obtain industry spot real space, barrier size and location, wireless senser power that industrial wireless sensor network is disposed, determine grid units precision, monitored area is divided into
three-dimensional grid, X, Y, Z divide according to grid units precision the grid number obtaining in corresponding horizontal, vertical, perpendicular space coordinates respectively;
(2), record the sensor node number that need to dispose according to process requirements in each grid, generate sensor node array A, sensor node array A size is
, grid value is the sensor node number in each grid;
(3), according to industry spot, wireless signal noisy barrier is being divided
position dyspoiesis thing array B in three-dimensional network coordinate, barrier array B size is
, grid value has barrier for " 1 " represents on the corresponding field position of this grid, and grid value represents clear on the corresponding field position of this grid for " 0 ";
(4), whether there will be according to mobile node
on field position corresponding to three-dimensional grid, generate positions of mobile nodes array C, positions of mobile nodes array C size is
, grid value there will be mobile node for " 1 " represents on field position that this grid is corresponding, if grid value is " 0 ", represents to there will not be mobile node on field position that this grid is corresponding;
(5), the communication radius of setting sensor node is
, the communication radius of leader cluster node is
the position of sensor node and leader cluster node can be the optional position in cube grid, and the condition that sensor node can be communicated by letter with leader cluster node is the communication radius that the ultimate range between the residing cube grid of sensor node and the residing cube grid of leader cluster node is less than sensor node
, and between both communication links, there is not barrier, otherwise think sensor node can not with leader cluster node communication; The condition that leader cluster node can be communicated by letter with leader cluster node is the communication radius that the residing cube grid of leader cluster node and the ultimate range between the residing cube grid of the leader cluster node that will communicate by letter are less than leader cluster node
, and do not have barrier between communication link between two leader cluster nodes, otherwise think between two leader cluster nodes can not communication;
(6), the solution of problem is expressed as vector
,
,
,
, n is the clear grid sum that can be used to dispose leader cluster node,
represent that j clear grid is vacant,
be illustrated in j clear grid and dispose leader cluster node, No is the grid sum that has barrier;
(7), the setting Pareto multiple target mankind learn the parameter of optimized algorithm, comprise disaggregation scale NP, individual knowledge base size si, social knowledge storehouse size ss, incidental learning Probability p r, individual learning probability pi, the initial disaggregation X of random generation, deposits in individual knowledge base IKD the solution in initial disaggregation X as initial IKD disaggregation in;
(8), calculate sensor network construction cost index in the fitness function F that in disaggregation X, each individual corresponding industrial wireless sensor network node multiple-objection optimization is disposed
, maintenance cost index
, the real-time index of system
fitness value:
Wherein,
for the leader cluster node sum of disposing,
the traffic load variance of leader cluster node,
the load number of j leader cluster node, L
cHall leader cluster node load average numbers,
for sensor node is to the maximum hop count of base station,
the multi-hop jumping figure that k sensor node signal is transferred to base station,
it is the number of sensor node in industrial wireless sensor network;
(9), judge in disaggregation X, whether each solution meets system reliability constraint, real-time constraint, maintainable and extensibility constraint, if do not meet constraints, calculate its separated amount of restraint to running counter to the infeasible solution of constraint, continue to optimize otherwise jump to step (11), concrete steps are as follows:
(9-1), judge in disaggregation X, whether each solution meets the reliability constraint of system sensor node, and sensor node reliability constraint formula is as follows:
Wherein,
can the communicate by letter minimum value of leader cluster node number of all the sensors node in industrial wireless sensor network,
it is the I communication leader cluster node number of sensor node in order to guarantee system communication reliability defined;
If do not meet formula (5), calculate the separated amount of restraint of its sensor node reliability to running counter to the infeasible solution of constraint, it is as follows that sensor node reliability is disobeyed amount of restraint computing formula:
Wherein,
for sensor node reliability is disobeyed amount of restraint,
(k=1,2 ..., N
s) be the leader cluster node number that k sensor node can be communicated by letter;
(9-2), judge in disaggregation X, whether each solution meets the reliability constraint of system leader cluster node, and leader cluster node reliability constraint formula is as follows:
Wherein,
the minimum value of communicated by letter with the leader cluster node number of all leader cluster nodes in industrial wireless sensor network,
it is the I communication leader cluster node number of leader cluster node in order to guarantee system communication reliability defined;
If do not meet formula (7), calculate the separated amount of restraint of its leader cluster node reliability to running counter to the infeasible solution of constraint, it is as follows that leader cluster node reliability is disobeyed amount of restraint computing formula:
Wherein,
for leader cluster node reliability is disobeyed amount of restraint,
(j=1,2 ..., N
cH) be the leader cluster node number that j leader cluster node can be communicated by letter;
(9-3), judge in disaggregation X, whether each solution meets the reliability constraint of system mobile node, system mobile node reliability constraint formula is as follows:
Wherein,
the minimum value of communicated by letter with the leader cluster node number of all mobile nodes in industrial wireless sensor network,
it is the I communication leader cluster node number of mobile node in order to guarantee system communication reliability defined;
If do not meet formula (9), calculate the separated amount of restraint of its mobile node reliability to running counter to the infeasible solution of constraint, it is as follows that mobile node reliability is disobeyed amount of restraint computing formula:
Wherein,
for industrial wireless sensor network mobile node reliability is disobeyed amount of restraint,
(t=1,2 ..., N
m) be the leader cluster node number that in the grid there will be at t mobile node, mobile node can be communicated by letter, N
mthe sum of the grid there will be for mobile node, N
mcan calculate by positions of mobile nodes array C;
(9-4), judge in disaggregation X, whether each solution meets system real time constraint, and real-time constraint formulations is as follows:
Wherein,
for sensor node is to the maximum hop count of base station,
be in whole industry wireless sensor network system sensor node to the higher limit of base station maximum hop count, sensor node exceedes higher limit to the jumping figure between base station cannot be transferred to control system in time by the data that cause transducer collection, thereby makes the real-time of system cannot meet industrial requirement;
If do not meet formula (11), calculate the separated amount of restraint of its real-time to running counter to the infeasible solution of constraint, it is as follows that real-time is disobeyed amount of restraint computing formula:
Wherein,
for industrial wireless sensor network real-time is disobeyed amount of restraint,
for sensor node is to the maximum hop count of base station,
be in whole industry wireless sensor network system sensor node to the higher limit of base station maximum hop count;
(9-5), judge in disaggregation X, whether each solution meets system maintainability and extensibility constraint, and maintainability is as follows with extensibility constraint formulations:
Wherein,
it is the load number of j leader cluster node, MCL is the accessible maximum load nodes of leader cluster node, consider that industrial wireless sensor network must have the demand of test, maintenance and upgrading in structure and normal course of operation, each leader cluster node should at least be reserved 1 load access point, therefore in order to guarantee that the extensibility of system requires the actual maximum communication load of each leader cluster node to be less than MCL-1;
If do not meet formula (13), calculate its maintainable and separated amount of restraint of extensibility for the infeasible solution of running counter to constraint, the computing formula that maintainable and extensibility is disobeyed amount of restraint is as follows:
(14)
Wherein,
disobey amount of restraint for industrial wireless sensor network is maintainable with extensibility;
(10), infeasible solution in step (9) is punished, the fitness value after infeasible solution punishment is:
Wherein P is an abundant large positive number,
for total separated amount of restraint,
(i=1,2 ..., 5) and be the separated amount of restraint of i constraint of infeasible solution;
(11), according to the fitness value of the construction cost index of each solution in disaggregation X, maintenance cost index, system real time index and always disobey amount of restraint disaggregation X is carried out to non-bad sequence, before getting rank according to the Pareto grade of separating and the distance value that blocks up after sequence, ss solution, as initial social knowledge storehouse SKD disaggregation, deposits all Pareto optimal solutions in optimal solution set Xo in; Wherein, in non-bad sequence, two solutions are dominant and are defined as follows:
(a) if separating x is feasible solution, separating y is infeasible solution, thinks that separating x is dominant and separates y, separates the Pareto grade of x and is better than separating y;
(b) if separating x is infeasible solution, separating y is infeasible solution, separates y but total promise breaking bundle value of separating x is less than, and thinks that separating x is dominant and separates y, and the Pareto grade of separating x is better than separating y;
(c) if separating x is feasible solution, separating y is feasible solution, is dominant and separates y but separate x, thinks that separating x is dominant and separates y, separates the Pareto grade of x and is better than separating y;
(12), carry out the study operator that the Pareto multiple target mankind learn optimized algorithm and generate disaggregation of new generation
, the Pareto multiple target mankind learn optimized algorithm and produce new explanation, formula specific as follows with specific probability by implementing individual study, social learning and random search study:
Wherein, rand () represents the random number between 0 to 1, pr is the probability of carrying out random search study operator, Rand (0,1) represent the random binary bits 0 or 1 that produces, p is [1, si] between random integers, q is [1, ss] between random integers, p is for determining the individuality of individual knowledge base IKD for new explanation study, q is for determining the individuality of the storehouse SKD of social knowledge for new explanation study, pi is the probability of carrying out individual study operator, (1-pr-pi) is the probability of implementing social learning;
(13), calculate disaggregation of new generation according to the fitness function providing in step (8)-step (10) and constraints
in fitness function value corresponding to each solution and total amount of restraint of disobeying;
(14), the renewal multiple target mankind learn the individual knowledge base of optimized algorithm
, social knowledge storehouse
and optimal solution set Xo, concrete steps are as follows:
(14-1), by disaggregation of new generation
, the storehouse SKD of social knowledge and optimal solution set Xo merge, then the disaggregation obtaining after being combined is carried out non-bad sequence according to fitness function value and total amount of restraint of disobeying, before getting rank according to the Pareto grade of separating after sequence and the distance value that blocks up, ss solution is as the new storehouse SKD of social knowledge, and all Pareto optimal solutions deposit Xo in;
(14-2), upgrade individual knowledge base
time, first check newly-generated individuality solution
(i=1,2 ..., NP) whether be present in the storehouse SKD of social knowledge after renewal, if existed, use newly-generated individuality solution
replace the poorest individual solution of rank in corresponding individual knowledge base IKD, if there is no in the storehouse SKD of social knowledge, new explanation
after mixing with individuality solutions all in individual knowledge base, carry out non-bad sequence, get before rank si according to the Pareto grade of separating after sequence and the distance value that blocks up and separate as the individual knowledge base IKD after renewal;
(15), judge whether to reach the default Pareto multiple target mankind and learnt optimized algorithm maximum iteration time, stop iterative search if reached, all Pareto optimal solutions in output optimal solution set Xo, in Xo, select according to the actual requirements an optimal solution, if do not reach maximum iteration time, jump procedure (12) continues iterative search.
A kind of industrial wireless sensor network Pareto multiple target of the present invention is disposed optimization method compared with prior art, has following apparent outstanding substantive distinguishing features and remarkable advantage:
First, adopt new industrial wireless sensor network to dispose Optimized model, being different from other dispositions methods setting sensor node or leader cluster node is deployed on grid intersection, the any point that in new model, the position of setting sensor node and leader cluster node can be in grid, thereby while making actual deployment, allow node location to have certain deviation, more realistic application characteristic; Can the basis for estimation that simultaneously adopts 2 ultimate ranges between the grid of node place communicate by letter as node make communication distance have certain surplus, can effectively overcome that the distance of node communication in practical application is subject to industrial environmental influence and the fuctuation within a narrow range that causes or node installation position deviation cause cannot proper communication problem, improved the robustness of scheme;
Secondly, dimension for multiple optimization aim in industrial wireless sensor network node deployment optimization problem may be not identical, in practical application, industrial wireless sensor network scale and application demand are also different, be difficult to provide exactly the weight between each optimization aim, simultaneously design may change to the importance of different demands in practical application in industry process, the present invention adopts a kind of Pareto multiple target mankind to learn optimized algorithm and solves multiple target industrial wireless sensor network node deployment optimization problem, thereby can obtain one group of Pareto optimal solution, engineer can comprehensively be compared according to each Pareto optimal solution realistic objective functional value obtaining, and then select ideal optimal solution according to practical application request, in the time that changing, application demand can from the Pareto optimal solution obtaining, reselect neatly optimal case,
Finally, the present invention is in optimizing construction cost, maintenance cost and real-time, take into full account interference, on-the-spot mobile node access, the system self maintained of the barriers such as industry spot hardware to wireless signal and expanded the reality such as needs with upgrading, more met industry spot practical application feature and demand.
Accompanying drawing explanation
Fig. 1 is double-layer separate clustering architecture schematic diagram ▲ expression sensor node of industrial wireless sensor network, ● represent leader cluster node, ★ represents base station;
Fig. 2 is general industry network communication of wireless sensor model schematic diagram ▲ expression sensor node, ● represent leader cluster node;
Fig. 3 is improvement industrial wireless sensor network communication model schematic diagram ▲ expression sensor node that the present invention proposes, ● represent leader cluster node;
Fig. 4 is the coding schematic diagram with the 3 D Industrial wireless sensor network node deployment issue of obstacle in the present invention (triangle represents the leader cluster node of disposing, and black square represents that place grid is for there being barrier grid);
Fig. 5 is Pareto noninferior solution schematic diagram;
Fig. 6 blocks up apart from schematic diagram in the non-bad sequence of Pareto;
Fig. 7 is that a kind of industrial wireless sensor network Pareto multiple target of the present invention is disposed optimization method flow chart.
Embodiment
Below in conjunction with accompanying drawing, embodiments of the present invention are described in further detail.
A kind of industrial wireless sensor network Pareto multiple target of the present invention is disposed optimization method, and its operating procedure is as follows:
(1), first obtain industry spot real space, barrier size and location, the wireless senser power that industrial wireless sensor network is disposed, determine grid units precision in conjunction with industry spot size with wireless sensor node maximum communication distance, monitored area is divided into
three-dimensional grid, X, Y, Z divide according to grid units precision the grid number obtaining in corresponding horizontal, vertical, perpendicular space coordinates respectively.
(2), record the sensor node number that need to dispose according to process requirements in each grid, generate sensor node array A, sensor node array A size is
, grid value is the sensor node number in each grid, for example, { in 1,2,3} grid, have 3 sensor nodes, the elements A in array A [1,2,3]=3.
(3), according to industry spot, wireless signal noisy barrier is being divided
position dyspoiesis thing array B in three-dimensional network coordinate, barrier array B size is
, array B element value has barrier, array B element value to represent clear on the corresponding field position of this grid for " 0 " for " 1 " represents on the corresponding field position of this grid.
(4), whether there will be according to mobile node
on field position corresponding to three-dimensional grid, generate positions of mobile nodes array C, positions of mobile nodes array C size is
, grid value there will be mobile node for " 1 " represents on field position that this grid is corresponding, if grid value is " 0 ", represents to there will not be mobile node on field position that this grid is corresponding.
(5), general traditional dispositions method of being different from is as shown in Figure 2 deployed in sensor node or leader cluster node on grid intersection, in this method, setting sensor node and leader cluster node can be deployed in the optional position in grid, as shown in Figure 3.
In previous other industrial wireless sensor networks or wireless sensor network node deployment issue, deployment region is divided into grid as shown in Figure 2 conventionally, and the crosspoint of grid is the position candidate of node deployment; Obviously, the setting that node deployment can only be deployed in grid intersection does not conform to the on-the-spot application characteristic of conventional model and actual industrial, the poor robustness of the deployment scheme practical application based on crosspoint in addition.For example in Fig. 2,2 leader cluster nodes (CH1, CH2) and 1 sensor node (S1) are arranged on the crosspoint of grid,
with
represent respectively the communication radius of sensor node and leader cluster node.In theory, CH2 is in the communication range of S1, and CH1 is in the communication range of CH2, and the data that therefore sensor node S1 gathers can be sent to CH1 via CH2.But in actual applications, it is extremely difficult all sensor nodes and leader cluster node being accurately arranged on the position of grid intersection representative; In addition, the actual communication range of S1, CH1 and CH2 may be subject to the impact of the factors such as high temperature, low temperature, electromagnetic interference, heavy rain and slightly reduce in harsh industrial environment, thereby can cause the interim interruption of network link and the failure of transfer of data.In the improvement deployment model that the present invention proposes, as shown in Figure 3, sensor node and leader cluster node are as in a certain grid, think that its physical location can be any point in grid, the condition that sensor node can be communicated by letter with leader cluster node is the communication radius that the ultimate range between the residing grid of sensor node and the residing grid of leader cluster node is less than sensor node
, and between both communication links, there is not barrier, otherwise think sensor node can not with leader cluster node communication; The condition that leader cluster node can be communicated by letter with leader cluster node is the communication radius that the residing grid of leader cluster node and the ultimate range between the residing grid of the leader cluster node that will communicate by letter are less than leader cluster node
, and do not have barrier between communication link between two leader cluster nodes, otherwise think can not communication between two leader cluster nodes.Visible, in improved model, allow the actual bit of node to be equipped with small size movement, leave the surplus of communication distance thereby simultaneously determine with ultimate range whether node can communicate by letter mutually, more realistic application characteristic, has better robustness and feasibility.In new deployment strategy, the data that in Fig. 3, sensor node S1 can only gather it are sent to by the leader cluster node in the region of green dotted line, i.e. a bunch CH3; A bunch CH1 can communicate by letter with CH3 with the leader cluster node CH2 that is positioned at the region that blue solid lines surrounds.It should be noted that, in Fig. 2 and Fig. 3, in order to be described clearly, to facilitate, suppose that all nodes are positioned at sustained height, therefore adopted 2 dimensional plane figure to describe; In practical application, in the time that node location height is not identical, should adopt the cube grid shown in Fig. 4, communication distance between node should calculate according to the three-dimensional coordinate range formula providing in following formula, for example two nodes are respectively at { (x1, x1+1), (y1, y1+1), (z1, z1+1) } and { (x2, x2+1), (y2, y2+1), (z2, z2+1) } in these two three-dimensional cubic volume mesh, and x1<x2, y1<y2, z1<z2, the communication distance of these two nodes is:
Because sensor node in industrial system need to be determined according to technique, cannot arbitrarily change installation site, therefore industrial wireless sensor network disposes that to optimize be mainly to optimize the deployed position of leader cluster node, and then optimizes whole topology of networks and guarantee, improve the performance of whole network.In practical application in industry, the normal communication of the barrier meeting severe jamming nodes such as large-scale hardware, therefore in the present invention, leader cluster node is not deployed in the grid of barrier.
(6), the solution of problem is expressed as
,
,
,
, n is the clear grid sum that industry spot can be used to dispose leader cluster node,
represent that j clear grid is vacant,
be illustrated in j clear grid and dispose leader cluster node, No is the grid sum that has barrier, and the No array B that can break the barriers obtains.Fig. 4 provides the expression example of the solution of 4 × 4 × 2 node deployment problems, in figure, the grid of blacking indicates the grid of barrier, triangle represents that corresponding grid is deployed with a bunch head, and owing to there being 3 grids to have barrier in this example problem, the length of therefore separating is 4 × 4 × 2-3=29.
(7), set the Pareto multiple target mankind and learn the parameter of optimized algorithm, comprise disaggregation scale NP, individual knowledge base size si, social knowledge storehouse size ss, incidental learning Probability p r, individual learning probability pi, generate at random initial disaggregation
, the solution in initial disaggregation X is deposited in individual knowledge base IKD as initial IKD disaggregation.
(8) fitness value of each target in the fitness function F of the industrial wireless sensor network node multiple-objection optimization deployment issue scheme that, in calculating disaggregation X, each individuality is corresponding
Wherein,
for the leader cluster node sum of disposing, in the time of design industrial wireless sensor network deployment scheme, because sensor node quantity must be good according to system monitoring Location of requirement, so system constructing cost fluctuation depends primarily on the number of leader cluster node, therefore the present invention is using the number of leader cluster node as sensor network construction cost index
be optimized,
littlely show that required leader cluster node is fewer, the cost that builds network is lower,
the traffic load variance of leader cluster node,
the load number of j leader cluster node, L
cHall leader cluster node load average numbers, the battery of leader cluster node must be changed before using, can normally work to guarantee industrial wireless sensor network, due to leader cluster node process and transmission data need energy, drive the leader cluster node of more multi-load by faster the leader cluster node consumed energy than other, the unbalanced of leader cluster node load will cause field engineer need to enter continually the battery of industry spot replacing bunch head, thereby increase maintenance cost, therefore the present invention adopts the traffic load variance of leader cluster node to characterize the load balance degree of bunch head, and as maintenance cost index
be optimized,
for sensor node is to the maximum hop count of base station,
the multi-hop jumping figure that k sensor node signal is transferred to base station,
it is the number of sensor node in industrial wireless sensor network, in the industrial wireless sensor network based on double-layer separate clustering architecture, the transfer of data real-time of upper network layer is mainly determined by the jumping figure of communication, in network lower floor, bunch head sends to base station by the mode of multi-hop after reading termly each sensor node data processing, therefore in network lower floor as long as sensor node number is no more than the upper limit and can guarantees real-time, so and owing to being the problem that does not have optimization raising real-time according to technological requirement fixed cycle pick-up transducers nodal information, and for the upper strata of network, maximum hop count difference in different deployment schemes between sensor node and base station, therefore be necessary to reduce multi-hop jumping figure to guarantee, improve the real-time of transfer of data, so the maximum hop count of the present invention between all the sensors node and base station is as the real-time index of system
be optimized.
(9), check in disaggregation X, whether each solution runs counter to system reliability, real-time, maintainable and extensibility constraint, if do not meet constraints, calculate its separated amount of restraint for the infeasible solution of running counter to constraint, continue to optimize otherwise jump to step (11), concrete steps are as follows:
(9-1), judge in disaggregation X, whether each solution meets the reliability constraint of system sensor node, system sensor node reliability constraint formula is as follows:
(5)
Wherein,
can the communicate by letter minimum value of leader cluster node number of all the sensors node in industrial wireless sensor network,
it is the I communication leader cluster node number of sensor node in order to guarantee system communication reliability defined;
If do not meet formula (5), calculate the separated amount of restraint of its sensor node reliability to running counter to the infeasible solution of constraint, it is as follows that sensor node reliability is disobeyed amount of restraint computing formula:
(6)
Wherein,
for sensor node reliability is disobeyed amount of restraint,
(k=1,2 ..., N
s) be the leader cluster node number that k sensor node can be communicated by letter;
(9-2), judge in disaggregation X, whether each solution meets the reliability constraint of system leader cluster node, system leader cluster node reliability constraint formula is as follows:
Wherein,
the minimum value of communicated by letter with the leader cluster node number of all leader cluster nodes in industrial wireless sensor network,
it is the I communication leader cluster node number of leader cluster node in order to guarantee system communication reliability defined;
If do not meet formula (7), calculate the separated amount of restraint of its leader cluster node reliability to running counter to the infeasible solution of constraint, it is as follows that leader cluster node reliability is disobeyed amount of restraint computing formula:
Wherein,
for leader cluster node reliability is disobeyed amount of restraint,
(j=1,2 ..., NCH) and be the leader cluster node number that j leader cluster node can be communicated by letter;
(9-3), judge in disaggregation X, whether each solution meets the reliability constraint of system mobile node, and mobile node reliability constraint formula is as follows:
Wherein,
the minimum value of communicated by letter with the leader cluster node number of all mobile nodes in industrial wireless sensor network,
it is the I communication leader cluster node number of mobile node in order to guarantee system communication reliability defined, in order to guarantee the reliability of industrial wireless sensor network communication, each sensor node, leader cluster node and mobile node must be equipped with extra leader cluster node in advance as secondary node, in the time that original work leader cluster node breaks down, spare cluster head node will automatically switch to operating state transmission information to base station, make network can keep normal work, the using and the reserved Reliability Strategy for example generally adopting for actual industrial can be set L
sN=L
cN=L
mN=2;
If do not meet formula (9), calculate the separated amount of restraint of its mobile node reliability to running counter to the infeasible solution of constraint, it is as follows that mobile node reliability is disobeyed amount of restraint computing formula:
Wherein,
for industrial wireless sensor network mobile node reliability is disobeyed amount of restraint,
(t=1,2 ..., N
m) be the leader cluster node number that in the grid there will be at t mobile node, mobile node can be communicated by letter, N
mthe sum of the grid there will be for mobile node, N
mcan calculate by positions of mobile nodes array C;
(9-4), judge in disaggregation X, whether each solution meets system real time constraint, and real-time constraint formulations is as follows:
Wherein,
for sensor node is to the maximum hop count of base station,
be in whole industry wireless sensor network system sensor node to the higher limit of base station maximum hop count, sensor node exceedes higher limit to the jumping figure between base station cannot be transferred to control system in time by the data that cause transducer collection, thereby make the real-time of system cannot meet industrial requirement
set point should determine according to actual industrial system real time demand and real sensor network hops required time, according to current industrial application testing situation, S
maxshould be less than or equal to 6;
If do not meet formula (11), calculate the separated amount of restraint of its real-time to running counter to the infeasible solution of constraint, it is as follows that real-time is disobeyed amount of restraint computing formula:
(9-5), judge in disaggregation X, whether each solution meets system maintainability and extensibility constraint, and maintainability is as follows with extensibility constraint formulations:
(13)
Wherein,
it is the load number of j leader cluster node, MCL is the accessible maximum load nodes of leader cluster node, consider that industrial wireless sensor network must have the demand of test, maintenance and upgrading in structure and normal course of operation, each leader cluster node should at least be reserved 1 load access point, therefore in order to guarantee that the extensibility of system requires the actual maximum communication load of each leader cluster node to be less than MCL-1;
If do not meet formula (13), calculate its maintainable and separated amount of restraint of extensibility for the infeasible solution of running counter to constraint, the computing formula that maintainable and extensibility is disobeyed amount of restraint is as follows:
Wherein,
disobey amount of restraint for industrial wireless sensor network is maintainable with extensibility;
(10), infeasible solution in step (9) is calculated to total amount of restraint of disobeying
and punish, after infeasible solution punishment, fitness function is:
Wherein, P is an abundant large positive number, by adding
make each target function value of infeasible solution all be greater than target function fitness value corresponding to any feasible solution,
for total amount of restraint of disobeying,
(i=1,2 ..., 5) and be the separated amount of restraint of i constraint of infeasible solution.
(11), according to the fitness value of the construction cost index of each solution in disaggregation X, maintenance cost index, system real time index and always disobey amount of restraint disaggregation X is carried out to non-bad sequence, before getting rank according to the Pareto grade of each solution and the distance value that blocks up after sequence, ss solution, as initial social knowledge storehouse SKD disaggregation, deposits all Pareto optimal solutions in optimal solution set Xo in.
Industrial wireless sensor network node deployment problem is multi-objective optimization question, the improvement of a target capabilities may cause the reduction of another target capabilities, for example reducing leader cluster node quantity can Cost optimization, but may cause a bunch load balancing variance to become large simultaneously, therefore not have the absolute optimal solution that makes all target functions simultaneously reach optimal value.In traditional method, multiple optimization aim are become to a single-goal function by weighted accumulation, the Pareto multiobjective optimization dispositions method that the present invention proposes is based on the optimum concept of Pareto, the set that the optimal solution that method finally obtains is made up of a series of noninferior solutions, is called " the optimum collection of Pareto ".In Pareto Multipurpose Optimal Method, " being dominant " is defined as follows:
Supposing to separate x reconciliation y is two feasible solutions, and claiming to separate x is that Pareto is dominant than separating y, and and if only if meets:
Wherein m is the optimization aim number in multi-objective optimization question.Specifically can be described as all
, make
, and at least have one
, make
, now title is separated x with respect to separating y for being dominant, or is called solution x domination solution y, is designated as
.The pass that is dominant of for example separate A in Fig. 5, separate B, solution C conciliating tetra-solutions of D is:
,
,
, B is inferior solution,
with
noninferior solution each other,
with
form Pareto optimal solution set.
Industrial wireless sensor network multiple target node deployment optimization problem is the optimization problem of a belt restraining simultaneously, punishes to make this infeasible solution to be inferior to feasible solution for the infeasible solution needs that do not meet constraints, for this reason further definition:
If,
(a), separating x is feasible solution and to separate y be infeasible solution;
(b), separate x conciliate y be all infeasible solution but total promise breaking bundle value of separating x is less than separates y;
(c), separate x conciliate y be all feasible solution but separate x be dominant separate y;
These three kinds of situations are all called separates x and is dominant and separates y, separates the Pareto grade of x and is better than separating y.Non-bad sequence flow process is: first all noninferior solutions in disaggregation X are elected, formed the solution set that Pareto grade is 1
, the solution that is then 1 by these Pareto grades shifts out from disaggregation X, then from the remaining solution of disaggregation X, selects all noninferior solutions again, forms the solution set that Pareto grade is 2
, the rest may be inferred, and in disaggregation X, all solutions are divided into the most at last
individual Pareto grade.Meanwhile, the Pareto multiple target mankind learn optimized algorithm and adopt the quality of blocking up apart from assessing the noninferior solution in same Pareto grade.The distance of blocking up is calculated and is carried out take same Pareto grade Noninferior Solution Set as unit, the solution that to calculate Pareto grade be 1 block up apart from time, only consider
in solution; The solution that to calculate Pareto grade be 2 block up apart from time, only consider
in solution; By that analogy.
Calculating block up apart from time, first solutions all in same Pareto grade is arranged from small to large according to each target function value respectively, then calculate successively under each target function the corresponding distance of blocking up, the block up cumulative sum of distance of all target function correspondences of last each solution is the actual distance of blocking up of this solution.Calculate each target function corresponding block up apart from time, take j target function as example, after arranging according to all solutions of the current Pareto grade of large young pathbreaker of j target function value, the distance of blocking up of order sequence first and the last solution that sorts is infinity, in practical operation, can assignment be an abundant large positive number; Of all the other solutions
what individual target function was corresponding block up, and distance is: after sequence before and after this solution the of adjacent two solutions
the absolute value of the difference of individual target function value and sequence first and sequence be last separate the
the ratio of the absolute value of the difference of individual target function value.Without loss of generality, Fig. 6 is the schematic diagram that calculates the distance of blocking up of two target function solutions, and in figure, solid dot representative is separated, and solution sorts according to target function value.In the direction of transverse axis target function f1, this N solution is arranged in order according to the size of f1 target function value; For target function f1, X
1and X
nbe respectively sequence first and the last solution of sequence, its correspondence is blocked up apart from being made as an abundant large positive CV; Separate for other, separate X with i
ifor example, what its target function f1 was corresponding blocks up distance for separating X
i-1conciliate X
i+1the absolute value of difference of target function f1 value (be f1 (X in Fig. 6
i) distance that represents) and X
1and X
nthe absolute value of difference of target function f1 value (be f1 in Fig. 6
maxrepresent distance) ratio,
; For longitudinal axis target function f2, after arranging from small to large according to the value of target function f2, separate X
nconciliate X
1be respectively first and separate with last, in like manner separate corresponding the blocking up apart from being all made as an abundant large positive CV of target function f2 for these two; And solution X
itarget function f2 correspondence block up distance for f2 (X in Fig. 6
i) representative distance divided by f2
maxthe distance representing,
.Therefore in the example shown in Fig. 6, X
1and X
nthe corresponding actual distance C d=2 that blocks up
cV; Separate X
ithe corresponding actual distance C d=that blocks up
; Other solutions can be separated X according to calculating
ithe account form of the distance of blocking up calculates the corresponding distance of blocking up; The value of CV should be far longer than 1.
(12), carry out the study operator that the Pareto multiple target mankind learn optimized algorithm and generate disaggregation of new generation
, the Pareto multiple target mankind learn optimized algorithm and produce new explanation, formula specific as follows with specific probability by implementing individual study, social learning and random search study:
Wherein, rand () represents the random number between 0 to 1, pr is the probability of carrying out random search study operator, Rand (0,1) represent the random binary bits 0 or 1 that produces, p is [1, si] between random integers, q is [1, ss] between random integers, p is for determining the solution of individual knowledge base IKD for new explanation study, q is for determining the individuality of the storehouse SKD of social knowledge for new explanation study, pi is the probability of carrying out individual study operator, (1-pr-pi) is the probability of implementing social learning.
When concrete enforcement, for j bit in i solution
, first produce a random number rand (), if
carry out random search study operator, generate at random a binary bits " 0 " or " 1 " and be assigned to
; If
, generate at random a random integers p between [1, si], make p solution in individual knowledge base IKD
selected, then order
equal
the value of j position bit; If
, generate a random integers q between [1, ss], make q solution in the storehouse SKD of social knowledge
for new explanation study, then order
equal
the value of j position bit; So repeat until generate disaggregation of new generation
.
(13), calculate disaggregation of new generation according to the fitness function providing in step (8)-step (10) and constraints
in fitness function value corresponding to each solution and total amount of restraint of disobeying.
(14), the renewal multiple target mankind learn the individual knowledge base of optimized algorithm
, social knowledge storehouse
and optimal solution set Xo, concrete steps are as follows:
(14-1), by disaggregation of new generation
merge with the storehouse SKD of social knowledge and optimal solution set Xo, then the disaggregation obtaining after being combined is carried out non-bad sequence according to fitness function value and total amount of restraint of disobeying, before getting rank according to the Pareto grade of separating after sequence and the distance value that blocks up, ss solution is as the new storehouse SKD of social knowledge, and all Pareto optimal solutions deposit Xo in.
(14-2), upgrade individual knowledge base
time, first check newly-generated individuality solution
(i=1,2 ..., NP) whether be present in the storehouse SKD of social knowledge after renewal, if existed, use newly-generated individuality solution
replace the poorest individual solution of rank in corresponding individual knowledge base IKD, if there is no in the storehouse SKD of social knowledge, new explanation
after mixing with individuality solutions all in individual knowledge base, carry out non-bad sequence, get before rank si according to the Pareto grade of separating after sequence and the distance value that blocks up and separate as the individual knowledge base IKD after renewal.
(15), judge whether to reach the default Pareto multiple target mankind and learnt optimized algorithm maximum iteration time, stop iterative search if reached, all Pareto optimal solutions in output optimal solution set Xo, in Xo, select according to the actual requirements an optimal solution, if do not reach maximum iteration time, jump procedure (12) continues iterative search.
A kind of industrial wireless sensor network Pareto multiple target of the present invention is disposed optimization method for the optimum deployment issue of the conventional double-layer separate clustering architecture industrial wireless sensor network of industrial system, provide a kind of Pareto multiple target to dispose optimization method, adopting the Pareto mankind to learn optimized algorithm is optimized leader cluster node deployed position, by being sensor node, mobile node in leader cluster node and network provides redundancy bunch head to guarantee the reliability of industrial wireless sensor network communication, by optimizing, in restriction industrial wireless sensor network, sensor node is optimized to the multi-hop step number of base station, guarantee the real-time of network, bunch quantity that optimizing minimizing needs realizes the optimization of system constructing cost, the harmony of a balanced bunch traffic load improves the economy of system operation maintenance to reduce maintenance, reservation networks mobile node and system upgrade safeguard that required interface guarantees the follow-up maintenance of industry wireless sensor network system, upgrade requirement, the impacts of barrier on network service such as on-the-spot large-scale hardware and building are considered, and practicality and the robustness of Optimization Design a kind of new deployment model have been proposed are further improved, learn optimized algorithm by the Pareto mankind and can effectively solve one group of Pareto optimization design scheme of acquisition, thereby make in practical application engineer can according to application demand, in conjunction with known each optimization aim functional value effectively Tactic selection go out optimum deployment scheme, without the weighted value that presets each optimization aim when the design, for the application of industrial wireless sensor network in large-scale process industrial provides an effective Optimization Design.
Claims (1)
1. industrial wireless sensor network Pareto multiple target is disposed an optimization method, it is characterized in that the method comprises the following steps:
(1), first obtain industry spot real space, barrier size and location, wireless senser power that industrial wireless sensor network is disposed, determine grid units precision, monitored area is divided into
three-dimensional grid, X, Y, Z divide according to grid units precision the grid number obtaining in corresponding horizontal, vertical, perpendicular space coordinates respectively;
(2), record the sensor node number that need to dispose according to process requirements in each grid, generate sensor node array A, sensor node array A size is
, grid value is the sensor node number in each grid;
(3), according to industry spot, wireless signal noisy barrier is being divided
position dyspoiesis thing array B in three-dimensional network coordinate, barrier array B size is
, grid value has barrier for " 1 " represents on the corresponding field position of this grid, and grid value represents clear on the corresponding field position of this grid for " 0 ";
(4), whether there will be according to mobile node
on field position corresponding to three-dimensional grid, generate positions of mobile nodes array C, positions of mobile nodes array C size is
, grid value there will be mobile node for " 1 " represents on field position that this grid is corresponding, if grid value is " 0 ", represents to there will not be mobile node on field position that this grid is corresponding;
(5), the communication radius of setting sensor node is
, the communication radius of leader cluster node is
the position of sensor node and leader cluster node can be the optional position in cube grid, and the condition that sensor node can be communicated by letter with leader cluster node is the communication radius that the ultimate range between the residing cube grid of sensor node and the residing cube grid of leader cluster node is less than sensor node
, and between both communication links, there is not barrier, otherwise think sensor node can not with leader cluster node communication; The condition that leader cluster node can be communicated by letter with leader cluster node is the communication radius that the residing cube grid of leader cluster node and the ultimate range between the residing cube grid of the leader cluster node that will communicate by letter are less than leader cluster node
, and do not have barrier between communication link between two leader cluster nodes, otherwise think between two leader cluster nodes can not communication;
(6), the solution of problem is expressed as vector
,
,
,
, n is the clear grid sum that can be used to dispose leader cluster node,
represent that j clear grid is vacant,
be illustrated in j clear grid and dispose leader cluster node, No is the grid sum that has barrier;
(7), the setting Pareto multiple target mankind learn the parameter of optimized algorithm, comprise disaggregation scale NP, individual knowledge base size si, social knowledge storehouse size ss, incidental learning Probability p r, individual learning probability pi, the initial disaggregation X of random generation, deposits in individual knowledge base IKD the solution in initial disaggregation X as initial IKD disaggregation in;
(8), calculate sensor network construction cost index in the fitness function F that in disaggregation X, each individual corresponding industrial wireless sensor network node multiple-objection optimization is disposed
, maintenance cost index
, the real-time index of system
fitness value:
(3)
Wherein,
for the leader cluster node sum of disposing,
the traffic load variance of leader cluster node,
the load number of j leader cluster node, L
cHall leader cluster node load average numbers,
for sensor node is to the maximum hop count of base station,
the multi-hop jumping figure that k sensor node signal is transferred to base station,
it is the number of sensor node in industrial wireless sensor network;
(9), judge in disaggregation X, whether each solution meets system reliability constraint, real-time constraint, maintainable and extensibility constraint, if do not meet constraints, calculate its separated amount of restraint to running counter to the infeasible solution of constraint, continue to optimize otherwise jump to step (11), concrete steps are as follows:
(9-1), judge in disaggregation X, whether each solution meets the reliability constraint of system sensor node, and sensor node reliability constraint formula is as follows:
Wherein,
can the communicate by letter minimum value of leader cluster node number of all the sensors node in industrial wireless sensor network,
it is the I communication leader cluster node number of sensor node in order to guarantee system communication reliability defined;
If do not meet formula (5), calculate the separated amount of restraint of its sensor node reliability to running counter to the infeasible solution of constraint, it is as follows that sensor node reliability is disobeyed amount of restraint computing formula:
(6)
Wherein,
for sensor node reliability is disobeyed amount of restraint,
(k=1,2 ..., N
s) be the leader cluster node number that k sensor node can be communicated by letter;
(9-2), judge in disaggregation X, whether each solution meets the reliability constraint of system leader cluster node, and leader cluster node reliability constraint formula is as follows:
Wherein,
the minimum value of communicated by letter with the leader cluster node number of all leader cluster nodes in industrial wireless sensor network,
it is the I communication leader cluster node number of leader cluster node in order to guarantee system communication reliability defined;
If do not meet formula (7), calculate the separated amount of restraint of its leader cluster node reliability to running counter to the infeasible solution of constraint, it is as follows that leader cluster node reliability is disobeyed amount of restraint computing formula:
Wherein,
for leader cluster node reliability is disobeyed amount of restraint,
(j=1,2 ..., N
cH) be the leader cluster node number that j leader cluster node can be communicated by letter;
(9-3), judge in disaggregation X, whether each solution meets the reliability constraint of system mobile node, system mobile node reliability constraint formula is as follows:
Wherein,
the minimum value of communicated by letter with the leader cluster node number of all mobile nodes in industrial wireless sensor network,
it is the I communication leader cluster node number of mobile node in order to guarantee system communication reliability defined;
If do not meet formula (9), calculate the separated amount of restraint of its mobile node reliability to running counter to the infeasible solution of constraint, it is as follows that mobile node reliability is disobeyed amount of restraint computing formula:
Wherein,
for industrial wireless sensor network mobile node reliability is disobeyed amount of restraint,
(t=1,2 ..., N
m) be the leader cluster node number that in the grid there will be at t mobile node, mobile node can be communicated by letter, N
mthe sum of the grid there will be for mobile node, N
mcan calculate by positions of mobile nodes array C;
(9-4), judge in disaggregation X, whether each solution meets system real time constraint, and real-time constraint formulations is as follows:
(11)
Wherein,
for sensor node is to the maximum hop count of base station,
be in whole industry wireless sensor network system sensor node to the higher limit of base station maximum hop count, sensor node exceedes higher limit to the jumping figure between base station cannot be transferred to control system in time by the data that cause transducer collection, thereby makes the real-time of system cannot meet industrial requirement;
If do not meet formula (11), calculate the separated amount of restraint of its real-time to running counter to the infeasible solution of constraint, it is as follows that real-time is disobeyed amount of restraint computing formula:
Wherein,
for industrial wireless sensor network real-time is disobeyed amount of restraint,
for sensor node is to the maximum hop count of base station,
be in whole industry wireless sensor network system sensor node to the higher limit of base station maximum hop count;
(9-5), judge in disaggregation X, whether each solution meets system maintainability and extensibility constraint, and maintainability is as follows with extensibility constraint formulations:
Wherein,
it is the load number of j leader cluster node, MCL is the accessible maximum load nodes of leader cluster node, consider that industrial wireless sensor network must have the demand of test, maintenance and upgrading in structure and normal course of operation, each leader cluster node should at least be reserved 1 load access point, therefore in order to guarantee that the extensibility of system requires the actual maximum communication load of each leader cluster node to be less than MCL-1;
If do not meet formula (13), calculate its maintainable and separated amount of restraint of extensibility for the infeasible solution of running counter to constraint, the computing formula that maintainable and extensibility is disobeyed amount of restraint is as follows:
Wherein,
disobey amount of restraint for industrial wireless sensor network is maintainable with extensibility;
(10), infeasible solution in step (9) is punished, the fitness value after infeasible solution punishment is:
Wherein P is an abundant large positive number,
for total separated amount of restraint,
(i=1,2 ..., 5) and be the separated amount of restraint of i constraint of infeasible solution;
(11), according to the fitness value of the construction cost index of each solution in disaggregation X, maintenance cost index, system real time index and always disobey amount of restraint disaggregation X is carried out to non-bad sequence, before getting rank according to the Pareto grade of separating and the distance value that blocks up after sequence, ss solution, as initial social knowledge storehouse SKD disaggregation, deposits all Pareto optimal solutions in optimal solution set Xo in; Wherein, in non-bad sequence, two solutions are dominant and are defined as follows:
(a) if separating x is feasible solution, separating y is infeasible solution, thinks that separating x is dominant and separates y, separates the Pareto grade of x and is better than separating y;
(b) if separating x is infeasible solution, separating y is infeasible solution, separates y but total promise breaking bundle value of separating x is less than, and thinks that separating x is dominant and separates y, and the Pareto grade of separating x is better than separating y;
(c) if separating x is feasible solution, separating y is feasible solution, is dominant and separates y but separate x, thinks that separating x is dominant and separates y, separates the Pareto grade of x and is better than separating y;
(12), carry out the study operator that the Pareto multiple target mankind learn optimized algorithm and generate disaggregation of new generation
, the Pareto multiple target mankind learn optimized algorithm and produce new explanation, formula specific as follows with specific probability by implementing individual study, social learning and random search study:
Wherein, rand () represents the random number between 0 to 1, pr is the probability of carrying out random search study operator, Rand (0,1) represent the random binary bits 0 or 1 that produces, p is [1, si] between random integers, q is [1, ss] between random integers, p is for determining the individuality of individual knowledge base IKD for new explanation study, q is for determining the individuality of the storehouse SKD of social knowledge for new explanation study, pi is the probability of carrying out individual study operator, (1-pr-pi) is the probability of implementing social learning;
(13), calculate disaggregation of new generation according to the fitness function providing in step (8)-step (10) and constraints
in fitness function value corresponding to each solution and total amount of restraint of disobeying;
(14), the renewal multiple target mankind learn the individual knowledge base of optimized algorithm
, social knowledge storehouse
and optimal solution set Xo, concrete steps are as follows:
(14-1), by disaggregation of new generation
, the storehouse SKD of social knowledge and optimal solution set Xo merge, then the disaggregation obtaining after being combined is carried out non-bad sequence according to fitness function value and total amount of restraint of disobeying, before getting rank according to the Pareto grade of separating after sequence and the distance value that blocks up, ss solution is as the new storehouse SKD of social knowledge, and all Pareto optimal solutions deposit Xo in;
(14-2), upgrade individual knowledge base
time, first check newly-generated individuality solution
(i=1,2 ..., NP) whether be present in the storehouse SKD of social knowledge after renewal, if existed, use newly-generated individuality solution
replace the poorest individual solution of rank in corresponding individual knowledge base IKD, if there is no in the storehouse SKD of social knowledge, new explanation
after mixing with individuality solutions all in individual knowledge base, carry out non-bad sequence, get before rank si according to the Pareto grade of separating after sequence and the distance value that blocks up and separate as the individual knowledge base IKD after renewal;
(15), judge whether to reach the default Pareto multiple target mankind and learnt optimized algorithm maximum iteration time, stop iterative search if reached, all Pareto optimal solutions in output optimal solution set Xo, in Xo, select according to the actual requirements an optimal solution, if do not reach maximum iteration time, jump procedure (12) continues iterative search.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410068269.3A CN103856952A (en) | 2014-02-27 | 2014-02-27 | Method for optimizing Pareto multiple target deployment of industrial wireless sensor network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410068269.3A CN103856952A (en) | 2014-02-27 | 2014-02-27 | Method for optimizing Pareto multiple target deployment of industrial wireless sensor network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103856952A true CN103856952A (en) | 2014-06-11 |
Family
ID=50864067
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410068269.3A Pending CN103856952A (en) | 2014-02-27 | 2014-02-27 | Method for optimizing Pareto multiple target deployment of industrial wireless sensor network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103856952A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107995630A (en) * | 2016-10-26 | 2018-05-04 | 中国移动通信集团广东有限公司 | A kind of the signal data computational methods and device of wireless network covering |
CN110081930A (en) * | 2019-04-24 | 2019-08-02 | 中国科学院城市环境研究所 | A kind of land, sea and air integration ecological environmental monitoring system |
CN110533233A (en) * | 2019-08-20 | 2019-12-03 | 河海大学 | Wireless microwave based on fitness optimization surveys rain link planing method |
CN111163477A (en) * | 2020-02-29 | 2020-05-15 | 南开大学 | Automatic deployment method of integrated intelligent base station in wide-area three-dimensional environment |
CN111314873A (en) * | 2020-02-14 | 2020-06-19 | 西北工业大学 | Stable high-precision network cooperative detection method |
CN112262320A (en) * | 2018-04-16 | 2021-01-22 | 普罗泰克斯公司 | Integrated circuit profiling and anomaly detection |
CN112996009A (en) * | 2021-03-31 | 2021-06-18 | 建信金融科技有限责任公司 | Wireless device deployment method and device, electronic device and storage medium |
CN114845315A (en) * | 2022-04-21 | 2022-08-02 | 福州大学 | Method for mobile balancing of target coverage based on sensors of base stations with limited capacity |
CN115508511A (en) * | 2022-09-19 | 2022-12-23 | 中节能天融科技有限公司 | Sensor self-adaptive calibration method based on gridding equipment full-parameter feature analysis |
CN116405874A (en) * | 2023-06-06 | 2023-07-07 | 湖南省水务规划设计院有限公司 | Reservoir ecological environment real-time monitoring method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101448267A (en) * | 2008-12-31 | 2009-06-03 | 中山大学 | Wireless sensor network node coverage optimization method based on particle swarm algorithm |
CN102098687A (en) * | 2011-03-02 | 2011-06-15 | 上海大学 | Multi-object optimized deployment method for industrial wireless sensor network |
CN103228023A (en) * | 2013-04-25 | 2013-07-31 | 青岛农业大学 | Wireless sensor network clustering routing method based on particle swarm optimization and ant colony optimization |
-
2014
- 2014-02-27 CN CN201410068269.3A patent/CN103856952A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101448267A (en) * | 2008-12-31 | 2009-06-03 | 中山大学 | Wireless sensor network node coverage optimization method based on particle swarm algorithm |
CN102098687A (en) * | 2011-03-02 | 2011-06-15 | 上海大学 | Multi-object optimized deployment method for industrial wireless sensor network |
CN103228023A (en) * | 2013-04-25 | 2013-07-31 | 青岛农业大学 | Wireless sensor network clustering routing method based on particle swarm optimization and ant colony optimization |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107995630B (en) * | 2016-10-26 | 2021-06-11 | 中国移动通信集团广东有限公司 | Wireless network coverage signal data calculation method and device |
CN107995630A (en) * | 2016-10-26 | 2018-05-04 | 中国移动通信集团广东有限公司 | A kind of the signal data computational methods and device of wireless network covering |
CN112262320A (en) * | 2018-04-16 | 2021-01-22 | 普罗泰克斯公司 | Integrated circuit profiling and anomaly detection |
CN110081930A (en) * | 2019-04-24 | 2019-08-02 | 中国科学院城市环境研究所 | A kind of land, sea and air integration ecological environmental monitoring system |
CN110533233A (en) * | 2019-08-20 | 2019-12-03 | 河海大学 | Wireless microwave based on fitness optimization surveys rain link planing method |
CN110533233B (en) * | 2019-08-20 | 2020-07-31 | 河海大学 | Wireless microwave rain measuring link planning method based on fitness optimization |
CN111314873A (en) * | 2020-02-14 | 2020-06-19 | 西北工业大学 | Stable high-precision network cooperative detection method |
CN111314873B (en) * | 2020-02-14 | 2022-05-17 | 西北工业大学 | Stable high-precision network cooperative detection method |
CN111163477B (en) * | 2020-02-29 | 2022-06-14 | 天津灵机科技有限公司 | Automatic deployment method of integrated intelligent base station in wide-area three-dimensional environment |
CN111163477A (en) * | 2020-02-29 | 2020-05-15 | 南开大学 | Automatic deployment method of integrated intelligent base station in wide-area three-dimensional environment |
CN112996009A (en) * | 2021-03-31 | 2021-06-18 | 建信金融科技有限责任公司 | Wireless device deployment method and device, electronic device and storage medium |
CN112996009B (en) * | 2021-03-31 | 2022-11-15 | 建信金融科技有限责任公司 | Wireless device deployment method and device, electronic device and storage medium |
CN114845315A (en) * | 2022-04-21 | 2022-08-02 | 福州大学 | Method for mobile balancing of target coverage based on sensors of base stations with limited capacity |
CN114845315B (en) * | 2022-04-21 | 2024-03-26 | 福州大学 | Method for balancing movement of target coverage of sensor based on capacity-limited base station |
CN115508511A (en) * | 2022-09-19 | 2022-12-23 | 中节能天融科技有限公司 | Sensor self-adaptive calibration method based on gridding equipment full-parameter feature analysis |
CN116405874A (en) * | 2023-06-06 | 2023-07-07 | 湖南省水务规划设计院有限公司 | Reservoir ecological environment real-time monitoring method |
CN116405874B (en) * | 2023-06-06 | 2023-08-15 | 湖南省水务规划设计院有限公司 | Reservoir ecological environment real-time monitoring method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103856952A (en) | Method for optimizing Pareto multiple target deployment of industrial wireless sensor network | |
CN102014398B (en) | Optimal deployment method of large-scale industrial wireless sensor network based on differential evolution algorithm | |
CN102098687B (en) | Multi-object optimized deployment method for industrial wireless sensor network | |
Ye et al. | A survey of model predictive control methods for traffic signal control | |
CN109862612B (en) | Data collection and wireless charging method based on dual-function trolley moving path planning | |
CN107317706A (en) | A kind of power transmission line intelligent monitoring system | |
CN105898764B (en) | Multi-level energy heterogeneous wireless sensor network deployment method | |
CN105577452B (en) | A kind of wireless sensor network connectivity recovery method and device towards obstacle | |
CN103716194A (en) | Method for optimizing network topology of urban rail train | |
Saravanan et al. | Modeling an Energy Efficient Clustering Protocol with Spider Cat Swarm Optimization for WSN | |
CN113705085A (en) | Modeling and risk assessment method for multi-level structure of smart power grid | |
CN110212527A (en) | A kind of grid structure of power distribution network and power distribution automation collaborative planning method | |
Zhang et al. | Resilience-based restoration sequence optimization for metro networks: A case study in China | |
CN112996076B (en) | Mobile charging and data collection method in wireless sensor network | |
CN111225398B (en) | Micro-grid wireless sensor network energy consumption optimization method based on cooperative coverage | |
Farahani | Energy Consumption Reduction in Wireless Sensor Network Based on Clustering | |
CN105139081A (en) | Highly reliable planning method for contact points in power distribution system | |
CN114358447B (en) | Flexible interconnection device configuration method and device, electronic equipment and storage medium | |
CN102970772B (en) | A kind of method building industry wireless network topological diagram based on improved adaptive GA-IAGA | |
Lee et al. | Relocation Framework to Improve the Usage Efficiency of Bicycle-sharing Systems | |
CN104599197B (en) | Intelligent substation protection system reliability layering equivalence method | |
Khamayseh et al. | On the Optimality of Route Selection in Grid Wireless Sensor Networks: Theory and Applications. | |
Dhanalakshmi et al. | Shortest path routing in solar powered WSNs using soft computing techniques | |
CN103166317A (en) | Reliable detection method and device containing distinguished encoding rule (DER) power distribution communication manner | |
CN105246121A (en) | Method for constructing variable dimension particle swarm in mobile sink information collection path |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140611 |