粒子群蚁群算法路径优化算法==要点详解.ppt

This paper introduces a mutation operator to basic PSO, something like GA. If the historic optimal position Pl keeps continuously unchanged or has changed extremely a little, in other word, the particle swarm aggregates heavily, we will keep Pl and re-random several dimension of particles according to certain probability to improve global searching ability without decreasing the convergent speed and search accuracy. While kMaxIterations ActualErrorAcceptableError If LogjamStep=MaxStep If SwarmRadiusBorderRadius Re-random the position and velocity of several data of each particles according to mutation probability ρ; LogjamStep=0; Else LogjamStep=LogjamStep+1; End End Update the position and velocity according to Eq. (1) and Eq. (2) End TABLE 1 RESULTS OF LAYOUT SCHEME FOR TEST SUIT 1 No Initial layout scheme Layout scheme by HCIGA Layout scheme By PSO r(mm) m(g) x(mm) y(mm) x(mm) y(mm) 1 10.0 100.00 -12.883 17.02 14.367 16.453 2 11.0 121.00 8.847 19.773 -18.521 -9.560 3 12.0 144.00 20.662 0.000 2.113 -19.730 4 11.5 132.00 -8.379 -19.43 19.874 -4.340 5 9.5 90.25 -1.743 0.503 -19.271 11.241 6 8.5 72.25 12.368 -18.989 -3.940 22.157 7 10.5 110.25 -21.639 -1.799 -0.946 2.824 In the test suite, the radius of the round container is R, which equals to 50mm, and the number of the round to-be-laid objects is 7. Suppose the acceptable value of the static equilibrium error, J, is [δJ=3.4 g·mm], and the other data of the to-be-laid objects and the result of the layout is shown in TABLE 1. (a) Layout Pattern by HCIGA (b) Layout Pattern by PSO TABLE 2 COMPARISON OF LAYOUT SCHEMES FOR TEST SUIT 1 Algorithm GA Ref [11] HCIGA PSO Radius of the out warp circle (mm) 32.837 32.662 31.843 Static equilibrium error (g.mm) 0.102 0.029 0.0022 Interference (mm) 0 0 0 Computation time (s) 1735 1002 1874 Ps: The computation time is converted into the time of the computer with 166M main frequency, by usin