CN112836845A - Method for solving shortest path of multiple targets in time-varying environment based on neural network - Google Patents

Method for solving shortest path of multiple targets in time-varying environment based on neural network Download PDF

Info

Publication number
CN112836845A
CN112836845A CN202011275087.5A CN202011275087A CN112836845A CN 112836845 A CN112836845 A CN 112836845A CN 202011275087 A CN202011275087 A CN 202011275087A CN 112836845 A CN112836845 A CN 112836845A
Authority
CN
China
Prior art keywords
wave
time
function
neural network
network
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
Application number
CN202011275087.5A
Other languages
Chinese (zh)
Inventor
黄玮
刘晋
王劲松
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University of Technology
Original Assignee
Tianjin University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tianjin University of Technology filed Critical Tianjin University of Technology
Priority to CN202011275087.5A priority Critical patent/CN112836845A/en
Publication of CN112836845A publication Critical patent/CN112836845A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Molecular Biology (AREA)
  • Neurology (AREA)
  • Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Evolutionary Computation (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Feedback Control In General (AREA)

Abstract

A method for solving the shortest path of multiple targets in a time-varying environment based on a neural network comprises the following steps: monitoring and acquiring data of the urban traffic network in real time by using map software, acquiring cost functions of all sides in the network, and constructing a traffic network model; combining the topological structure of the network, designing a brand-new neuron structure and constructing a neural network model. The topological information of the traffic network is loaded into a neural network model, the wave generator and the filter of the neuron are used for sending automatic waves to the following neuron, other neurons in the network are activated, and each neuron ignores the response to partial waves by using the filter. The method prunes the non-optimal sub-paths by designing the neural network, greatly reduces the computational complexity, and realizes the problem of accurately solving the shortest path of multiple targets in the time-varying environment, which is the network optimization field. In addition, the method applies the neural network parallel computation to accelerate the problem solving speed.

Description

Method for solving shortest path of multiple targets in time-varying environment based on neural network
Technical Field
The invention belongs to the combination of the technical field of network optimization and the technical field of artificial intelligence, and particularly relates to a method for solving a shortest path of multiple targets in a time-varying environment based on a neural network.
Background
The classical multi-objective shortest path problem is a routing problem for optimizing multiple objectives, for example, in the process of transportation, if a path with short distance and good road condition is desired to be found, two objective functions, the shortest travel time and distance, need to be optimized simultaneously. The value of the weight function of each edge in the network is invariant over time. However, for example, in the transportation process, the transportation time varies with the congestion condition of the road condition, that is, the edge weight function is time-dependent, which expands the problem of solving the shortest path of multiple targets in a time-varying environment.
In recent years, researchers at home and abroad successively develop the research on the shortest path of multiple targets in a time-varying environment, and the technical difficulties mainly focus on a solving method, solving speed and solving precision.
In the aspect of the solving method, from the research results at home and abroad, the calculation complexity for solving the multi-target problem in the time-varying environment is high, the traditional graph theory solving methods such as dynamic programming and a label setting algorithm are difficult to break through, and the intelligent algorithms such as a genetic algorithm and an ant colony algorithm cannot be used for solving the accurate solution.
In the aspect of solving speed, the method for solving the accurate solution by the research papers at home and abroad is based on the traditional algorithm such as dynamic programming, label setting algorithm and the like, and cannot meet the actual requirement on popular PCs due to large calculated amount, low algorithm efficiency and dependence of the operation environment on the support of high-performance computers.
In the aspect of solving the precision, the traditional routing algorithm adopted by the general system ignores the time-varying environment to find the shortest path of multiple targets, in other words, the network is regarded as static, and the precise solution is difficult to solve.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a method for solving the shortest path of multiple targets in a time-varying environment based on a neural network.
The invention relates to a method for solving the shortest path of multiple targets in a time-varying environment based on a neural network, which comprises the following specific steps:
step 1, utilizing map software to collect network topology information in real time to construct a traffic network model, wherein the collected information comprises predecessors and successors of all nodes and weighting functions of all edges, such as a routing time function Bij(ti) And a routing cost function Rij(ti)；
Step 2, loading the acquired network topology information into a neural network model, wherein the topological structure of the designed and constructed neural network model is the same as that of a traffic network, and training of a neural network by a previous data set is not needed;
step 3, initializing a neural network, comprising 5 parts of a receiver, a filter, a memory, a wave generator and a wave transmitter of each neuron, setting a decoding function F in the receiver, setting a filtering function D in the filter, opening up a storage space S in the memory, initializing an energy function E and a threshold function theta in the wave generator, setting an encoding function J in the wave generator and setting the initial time t of the network to be 0;
and step 4, updating the neural network, which is divided into 3 sub-steps: receiving and filtering the automatic waves, storing and generating the automatic waves and activating and forwarding the automatic waves; updating the state of each neuron at the moment t of the network by adopting a neuron activation technology and an automatic wave filtering technology;
step 5, updating the network time T to be T +1, and repeating the step 4 until T to be T, wherein T is the time upper limit of the network;
and 6, traversing the waves received by the target neuron and finding out all multi-target shortest paths by a backtracking method.
The weighting function described in step 1 is a probability distribution function for calculating variables by using statistical software according to the recorded topological data, and is generally a piecewise function or a linear continuous function.
And (3) the topological structure of the neural network model in the step 2 is the same as that of the traffic network, and the neural network is not trained by a previous data set.
The initialized neural network in the step 3 comprises 5 parts;
step 3.1, initialization in the receiver is
Figure BSA0000224726600000021
And a decoding function F;
step 3.2, initializing a filter function D in the filter, and screening to follow a Pareto optimal non-dominance principle;
step 3.3, opening up a storage space S in the wave storage devicei[|M|]M is a predefined positive integer constant;
step 3.4, initialize threshold function in wave generator
Figure BSA0000224726600000022
Wherein B isij(t) is the route time from point i to point j when the traveler passes through arc (i, j), t is the network time, h is the time interval of t;
step 3.5, initialization in the wave generator
Figure BSA0000224726600000023
And an encoding function J.
The neural network update described in step 4 consists of the following three substeps: the method comprises the following steps of receiving and filtering the automatic wave, storing and generating the automatic wave and activating and forwarding the automatic wave, wherein the following steps are respectively detailed:
step 4.1, automatic wave receiving and filtering;
step 4.1.1, detecting the receiver WxiAnd if the wave sent by the precursor neuron exists, decoding a path weight vector carried by each wave by using a decoding function F, wherein the path weight vector comprises a path time txiAnd a routing cost rxiThe formula is as follows: t is txi＝Fb(Wxi)，rxi＝Fr(Wxi)；
Step 4.1.2, according to the Pareto optimal non-dominance principle, the filter filters out a part of automatic waves by using a filter function D, for example, the weight vector of the wave No. 1 is (3, 5), the weight vector of the wave No. 2 is (4, 6), the two wave weight vectors are respectively compared with 3 < 4 and 5 < 6, the wave No. 1 is superior to the wave No. 2, so the wave No. 2 is filtered out by a neuron;
step 4.2, automatic wave storage and generation;
step 4.2.1, filtering each wave weight vector (t, r) and the cost vector (B) of the subsequent edgeij(t)，Rij(t)), calculating a wave correlation parameter tij＝t+Bij(t) and rij＝r+Rij(t)；
Step 4.2.2, correlating the parameter tijAnd rijStored in wave storage device R [ k ]i]＝(tij，rij) Energy function E of the initialization waveij(t，ki)，kiIs the subscript value of the wave stored in the wave storage device.
Step 4.3, activating and forwarding the automatic waves;
step 4.3.1, energy function E initialized for step 4.2.2ij(t，ki) Performing energy accumulation Eij(t，ki)＝Eij(t，ki) + Δ t, when
Figure BSA0000224726600000031
The automatic wave is activated, where at is the increment of the energy function per second,
Figure BSA0000224726600000032
is a threshold function defining the corresponding wave on edge (i, j);
step 4.3.2, the activated automatic wave is according to its energy function Eij(t，ki) The parameter information recorded in the memory S is loaded into the automatic wave by the coding function J of the wave transmitter and is transmitted to the subsequent neuron, and the formula is Wij＝J(S[ki])，j＝g，...，p。
And (5) updating the network time T, wherein each time T is T +1 after each updating iteration, all neurons in the neural network update the states of the neurons, namely, the step 4 is executed, and when the time T reaches the upper time limit T, the neural network stops updating.
And 6, traversing each automatic wave reaching the target neuron, backtracking the paths through each neuron wave storage device, and outputting all multi-target shortest paths, including the starting time of each path, the nodes passed by the path, the time of leaving each node, the cost vector of each arc, the total time and the total cost of the paths.
Advantages and advantageous effects of the invention
The method can accurately solve the shortest path of multiple targets in a time-varying environment, and compared with the traditional algorithm, the designed and constructed neural network improves the computing capability, simplifies the computing complexity and further improves the computing efficiency. In addition, the designed and constructed neural network can solve the complex network optimization problem through further improvement.
Drawings
FIG. 1 is a general flow chart of the present invention for solving the problem of shortest path of multiple targets in a time-varying environment;
FIG. 2 is a diagram of a neuron architecture;
fig. 3 is a time-varying network topology.
Detailed Description
The invention is described in detail below with reference to the figures and examples.
A method for solving the shortest path of multiple targets in a time-varying environment based on a neural network provides all Pareto optimal paths for decision makers. As shown in fig. 1, the specific embodiment includes the following steps:
TABLE 1 topology information of a network
Figure BSA0000224726600000041
Step 1, utilizing map software to collect network topology information in real time to construct a traffic network model, wherein the collected information comprises predecessors and successors of all nodes and time-varying weight functions of all edges, such as a routing time function Bij(ti) And a routing cost function Rij(ti) (ii) a The topology of this example is shown in FIG. 3, and the topology information is shown in Table 1, such as the time interval [0, 1 ] for an arc (2, 3) from node 2 to node 3]The routing time is 1 and the routing cost is 2.
Step 2, loading the acquired network topology information into a neural network model to construct 4 neurons in total, wherein the neuron structure is shown in fig. 2, and the constructed neural network topology structure is the same as that of a traffic network; and setting the time upper limit T of the source point neuron, the target neuron and the network to be 5, wherein the constructed neural network is not trained by a previous data set.
Step 3, initializing a neural network, comprising 5 parts of a receiver, a filter, a memory, a wave generator and a wave transmitter of each neuron;
step 3.1, the receiver initializes and decodes the function F,
Figure BSA0000224726600000042
step 3.2, initializing a filter function D in the filter, and screening to follow a Pareto optimal non-dominance principle;
step 3.3, opening up a storage space S in the wave storage device0[|10|]，S1[|10|]，S2[|10|]，S3[|10|]Is a predefined positive integer constant;
step 3.4, initialize threshold function in wave generator
Figure BSA0000224726600000043
Wherein B isij(t) is the route time from point i to point j when the traveler passes through arc (i, j), and t is the network time;
step 3.5, initialization in the wave generator
Figure BSA0000224726600000044
And an encoding function J.
And 4, updating the state of each neuron of the neural network at the moment t, and dividing the updating into 3 sub-steps: the method comprises the following steps of receiving and filtering the automatic wave, storing and generating the automatic wave, and activating and forwarding the automatic wave:
step 4.1, automatic wave receiving and filtering;
step 4.1.1, neurons 1, 2, 3, 4 detect receiver W, respectivelyxiIf m has a wave from a precursor neuron, a decoding function F is usedb，FrDecoding the path weight vector carried by each wave, including path time bxiAnd a routing cost rxi(ii) a Neuron 2 decodes the t ═ 1 received wave: t ═ Fb(W12)，r＝Fr(W12)；
Step 4.1.2, the routing time and the routing cost are taken as two optimization targets, and two conditions are divided according to the Pareto optimal non-dominance principle: in the first case, when multiple waves arrive at the neuron at the same time, the waves with the same routing time and higher routing cost are filtered by the filter; in the second case, multiple waves arrive at the neuron in sequence, and the later arriving waves are routed for a long time, and then are filtered by the filter if the routing cost is greater than the cost of the first arriving waves;
step 4.2, automatic wave storage and generation;
step 4.2.1, filtering each wave weight vector (t, r) and the cost vector (B) of the subsequent edgeij(t)，Rij(t)), calculating a wave correlation parameter tij＝t+Bij(t) and rij＝r+Rij(t)；
Step 4.2.2, correlating the parameter tijAnd rijStored in wave storage device R [ k ]i]＝(tij，rij) And initializing the wave energy function Eij(t，ki) 0, where t is the network current time, kiThe weight parameter of the wave is recorded in the wave storage device SiThe subscript of (1).
Step 4.3, activating and forwarding the automatic waves;
step 4.3.1, energy function E initialized for step 4.2ij(t，ki) Performing an energy accumulation Eij(t，ki)＝Eij(t，ki) + Δ t, when
Figure BSA0000224726600000051
The automatic wave is activated, where at is the increment of the energy function per second,
Figure BSA0000224726600000052
is a threshold function defining the corresponding wave on the edge (i, j), h being the time interval in which time t is located;
step 4.3.2, the activated automatic wave is according to its energy function Eij(t，ki) The parameter information recorded in the memory S is loaded into the automatic wave by the coding function J of the wave transmitter and is transmitted to the subsequent neuron, and the formula is Wij＝J(S[ki])，j＝g，...，p。
And step 5, updating the network time, setting the time t to be t +1, repeating the step 4 until the time t to be 5, wherein 5 is the time upper limit of the network, and each neuron stops wave generation.
And 6, receiving two waves at t 2 and t 3 by the No. 3 target neuron according to a method of backtracking according to each neuron wave storage device SiFinding the track from the source point neuron to the target neuron of the two waves, and outputting two multi-target shortest paths, wherein the path 1: 0- > 1- > 2- > 3, arrival time is 3, and routing cost is 3; route 2: 0- > 2- > 3, arrival time is 2, and routing cost is 4.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be made in the present invention; any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims (7)

1. A method for solving the shortest path of multiple targets in a time-varying environment based on a neural network is characterized by comprising the following steps:
step 1, utilizing map software to collect network topology information in real time to construct a traffic network model, wherein the collected information comprises predecessors and successors of all nodes and weighting functions of all edges, such as a routing time function Bij(ti) And a routing cost function Rij(ti)；
Step 2, loading the acquired network topology information into a neural network model, wherein the topological structure of the designed and constructed neural network model is the same as that of a traffic network, and training of a neural network by a previous data set is not needed;
step 3, initializing a neural network, comprising 5 parts of a receiver, a filter, a memory, a wave generator and a wave transmitter of each neuron, setting a decoding function F in the receiver, setting a filtering function D in the filter, opening up a storage space S in the memory, initializing an energy function E and a threshold function theta in the wave generator, setting an encoding function J in the wave generator and setting the initial time t of the network to be 0;
and step 4, updating the neural network, which is divided into 3 sub-steps: receiving and filtering the automatic waves, storing and generating the automatic waves and activating and forwarding the automatic waves; updating the state of each neuron at the moment t of the network by adopting a neuron activation technology and an automatic wave filtering technology;
step 5, updating the network time T to be T +1, and repeating the step 4 until T to be T, wherein T is the time upper limit of the network;
and 6, traversing the waves received by the target neuron and finding out all multi-target shortest paths by a backtracking method.
2. The method for solving multiple-objective shortest path according to claim 1, wherein: the weighting function described in step 1 is a probability distribution function for calculating variables by using statistical software according to the recorded topological data, and is generally a piecewise function or a linear continuous function.
3. The method for solving multiple-objective shortest path according to claim 1, wherein: and (3) the topological structure of the neural network model in the step 2 is the same as that of the traffic network, and the neural network is not trained by a previous data set.
4. The method for solving multiple-objective shortest path according to claim 1, wherein: the initialized neural network in the step 3 comprises 5 parts;
step 3.1, initialization in the receiver is
Figure RE-FSB0000192446930000011
And a decoding function F;
step 3.2, initializing a filter function D in the filter, and screening to follow a Pareto optimal non-dominance principle;
step 3.3, opening up a storage space S in the wave storage devicei[|M|]M is a predefined positive integer constant;
step 3.4, initialize threshold function in wave generator
Figure RE-FSB0000192446930000012
Wherein B isij(t) is the route time from point i to point j when the traveler passes through arc (i, j), t is the network time, h is the time interval of t;
step 3.5, initialization in the wave generator
Figure RE-FSB0000192446930000013
And an encoding function J.
5. The method for solving multiple-objective shortest path according to claim 1, wherein: the neural network update described in step 4 consists of the following three substeps: the method comprises the following steps of receiving and filtering the automatic wave, storing and generating the automatic wave and activating and forwarding the automatic wave, wherein the following steps are respectively detailed:
step 4.1, automatic wave receiving and filtering;
step 4.1.1, detecting the receiver WxiAnd if the wave sent by the precursor neuron exists, decoding a path weight vector carried by each wave by using a decoding function F, wherein the path weight vector comprises a path time txiAnd a routing cost rxiThe formula is as follows: t is txi＝Fb(Wxi)，rxi＝Fr(Wxi)；
Step 4.1.2, according to the Pareto optimal non-dominance principle, the filter filters out a part of automatic waves by using a filter function D, for example, the weight vector of the wave No. 1 is (3, 5), the weight vector of the wave No. 2 is (4, 6), the two wave weight vectors are respectively compared with 3 < 4 and 5 < 6, the wave No. 1 is superior to the wave No. 2, so the wave No. 2 is filtered out by a neuron;
step 4.2, automatic wave storage and generation;
step 4.2.1, filtering each wave weight vector (t, r) and the cost vector (B) of the subsequent edgeij(t)，Rij(t)), calculating a wave correlation parameter tij＝t+Bij(t) and rij＝r+Rij(t)；
Step 4.2.2, correlating the parameter tijAnd rijStored in wave storage device R [ k ]i]＝(tij，rij) Energy function E of the initialization waveij(t，ki)，kiIs the subscript value of the wave stored in the wave storage device.
Step 4.3, activating and forwarding the automatic waves;
step 4.3.1, energy function E initialized for step 4.2.2ij(t，ki) Performing energy accumulation Eij(t，ki)＝Eij(t，ki) + Δ t, when
Figure RE-FSB0000192446930000021
The automatic wave is activated, where at is the increment of the energy function per second,
Figure RE-FSB0000192446930000022
is a threshold function defining the corresponding wave on edge (i, j);
step 4.3.2, the activated automatic wave is according to its energy function Eij(t，ki) The parameter information recorded in the memory S is loaded into the automatic wave by the coding function J of the wave transmitter and is transmitted to the subsequent neuron, and the formula is Wij＝J(S[ki])，j＝g，...，p。
6. The method for solving multiple-objective shortest path according to claim 1, wherein: the method for solving multiple-objective shortest path according to claim 1, wherein: and (5) updating the network time T, wherein each time T is T +1 after each updating iteration, all neurons in the neural network update the states of the neurons, namely, the step 4 is executed, and when the time T reaches the upper time limit T, the neural network stops updating.
7. The method for solving multiple-objective shortest path according to claim 1, wherein: the method for solving multiple-objective shortest path according to claim 1, wherein: and 6, traversing each automatic wave reaching the target neuron, backtracking the paths through each neuron wave storage device, and outputting all multi-target shortest paths, including the starting time of each path, the nodes passed by the path, the time of leaving each node, the cost vector of each arc, the total time and the total cost of the paths.
CN202011275087.5A 2020-11-11 2020-11-11 Method for solving shortest path of multiple targets in time-varying environment based on neural network Pending CN112836845A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011275087.5A CN112836845A (en) 2020-11-11 2020-11-11 Method for solving shortest path of multiple targets in time-varying environment based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011275087.5A CN112836845A (en) 2020-11-11 2020-11-11 Method for solving shortest path of multiple targets in time-varying environment based on neural network

Publications (1)

Publication Number Publication Date
CN112836845A true CN112836845A (en) 2021-05-25

Family

ID=75923068

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011275087.5A Pending CN112836845A (en) 2020-11-11 2020-11-11 Method for solving shortest path of multiple targets in time-varying environment based on neural network

Country Status (1)

Country Link
CN (1) CN112836845A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113660677A (en) * 2021-07-29 2021-11-16 西安电子科技大学 Maximum error independent path calculation method of weighted time-varying network under consumption limit

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113660677A (en) * 2021-07-29 2021-11-16 西安电子科技大学 Maximum error independent path calculation method of weighted time-varying network under consumption limit
CN113660677B (en) * 2021-07-29 2022-08-19 西安电子科技大学 Maximum error independent path calculation method of weighted time-varying network under consumption limit

Similar Documents

Publication Publication Date Title
Chouikhi et al. PSO-based analysis of Echo State Network parameters for time series forecasting
Kanan et al. Feature selection using ant colony optimization (ACO): a new method and comparative study in the application of face recognition system
Luo et al. Species-based particle swarm optimizer enhanced by memory for dynamic optimization
Helbig et al. Population-based metaheuristics for continuous boundary-constrained dynamic multi-objective optimisation problems
CN109102124B (en) Dynamic multi-target multi-path induction method and system based on decomposition and storage medium
CN113554875B (en) Variable speed-limiting control method for heterogeneous traffic flow of expressway based on edge calculation
Omidshafiei et al. Graph-based cross entropy method for solving multi-robot decentralized POMDPs
Nasiri et al. History-driven firefly algorithm for optimisation in dynamic and uncertain environments
Cao et al. A collaboration-based particle swarm optimizer with history-guided estimation for optimization in dynamic environments
CN115862319A (en) Traffic flow prediction method for space-time diagram self-encoder
CN112836845A (en) Method for solving shortest path of multiple targets in time-varying environment based on neural network
Arora et al. A survey of comparison between various metaheuristic techniques for path planning problem
Sulakshana et al. Data acquisition through mobile sink for wsns with obstacles using support vector machine
Gupta et al. Optimization of stacking ensemble configuration based on various metahueristic algorithms
He et al. Particle swarm optimization RBF neural network model for internet traffic prediction
CN112770256A (en) Node track prediction method in unmanned aerial vehicle self-organizing network
CN116886176A (en) Predictable inter-satellite routing method based on link utility function
Wang et al. A hybrid optimization algorithm based on ant colony and immune principles.
CN112073983B (en) Wireless data center network topology optimization method and system based on flow prediction
You et al. On parallel immune quantum evolutionary algorithm based on learning mechanism and its convergence
CN115202357A (en) Autonomous mapping method based on impulse neural network
Ojha et al. An analysis of artificial immune system and genetic algorithm in urban path planning
Zhao et al. Learning multi-agent communication with policy fingerprints for adaptive traffic signal control
Zhang et al. Selecting the best routing traffic for packets in LAN via machine learning to achieve the best strategy
Boryczka et al. Efficient DPSO neighbourhood for dynamic traveling salesman problem

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210525