In particle swarm optimization (PSO), a particle’s best position is the strongest candidate solution it has found so far according to the optimization objective or fitness value. This personal best, or pbest, records individual experience while the swarm also tracks neighborhood and global best positions. section 9.2
PSO uses the best position as a continuing memory during each iteration:
- Initialize each particle’s personal best to its starting position, because that is its best-known solution before searching. section 9.2.3
- Evaluate the particle’s current fitness after it moves through the search space. section 9.2
- Replace pbest when the current position has superior fitness; otherwise, retain the previous best position. section 9.2.2
- Use pbest in the cognitive component of the velocity update, attracting the particle toward areas supported by its own past experience. section 9.2.1
A particle’s remembered solution
A particle begins with its initial position stored as pbest. After an update, its new position is evaluated. If that position produces better fitness, pbest moves to the new position; if not, the earlier position remains its personal reference for later velocity updates. section 9.2.2 section 9.2.3
Figure 9.2 The PSO algorithm
The PSO flowchart shows where fitness evaluation, personal-best updates, global-best identification, and position and velocity updates fit into the iterative search. section 9.2
