Intelligent scheduling of FMSs with inductive learning capability using neural networks

Abstract

Abstract. With the growing uncertainty and complexity in the manul~acturing environment, most scheduling problems have been proven to be NP-complete and this can degrade the performance of conventional operations research (OR) techniques~ This article presents a system-attribute-oriented knowledge-based scheduling system (SAOSS) with inductive learning capability. With the rich heritage from artificial intelligence (AI), SAOSS takes a multialgorithm paradigm which makes it more intelligent, flexible, and suitable than others for tackling complicated, dynamic scheduling problems.SAOSS employs an efficient and effective inductive learning method, a continuous iterative dichotomister 3 (CID3) algorithm, to induce decision rules for scheduling by converting corresponding decision frees into hidden layers of a self-ger~rated neural network. Connection weights between hidden units imply the schethling heuristics, which are then formulated into scheduling rules. An FMS scheduling problem is also given for illustration. The scheduling results show that the system-attribute-oriented knowledge-based approach is capable of addressing dynamic scheduling problems.

Dynamic scheduling of flexible manufacturing system using heuristic approach

“…Some seconds are sufficient for GA to converge to HLB in layouts with different values of mi (eg. L (15,6), L(8.5.3), etc.). Instead, in a symmetrical layout (eg.…”

“…with respect to a given set of jobs. This strategy has also been implemented through AI techniques such as neural networks [15], discrete event simulation [22], fuzzy logic [14], knowledge-based systems (O' Kane et al 1994) and hybrid systems [18]. The second group includes 'job-oriented' methods which generate the schedule through the analysis of the most efficient alternatives in order to select the optimal (or the near-optimal) solution.…”

An evolutionary approach to complex job-shop and flexible manufacturing system scheduling

“…This problem is NP-hard [7], and thus a generalization of JSSP maintains this feature. Strategies e ciently to solve JSSP have been implemented by means of continuous ow-based relaxed models [8], neural networks [9,10], discrete event simulation [11], fuzzy logic [12] and tabu search [13]. An overview of the major approaches to deterministic job-shop scheduling, aimed at minimizing the makespan, and the results obtained on the available benchmarks are reported in reference [14].…”

Knowledge-based workcell attribute oriented dynamic schedulers for flexible manufacturing systems

“…For n jobs and m machines in the general case there will be (n!) The techniques included in approximation algorithms are: branch-bound [9,10]; Lagrangian relaxation decomposition [11]; simulated annealing [12]; Tabu-search [13]; genetic programs [14][15][16][17][18][19]; artificial intelligence (AI) techniques [20][21][22][23][24][25][26]; priority rule based [27]; heuristics [28][29][30][31][32][33]. Scheduling problems are complex even for simple formulations and NP hard in many cases.…”

“…Wang et al [25] presented a system-attribute oriented knowledge-based scheduling system which uses a maximise resource utilisation (MRU) algorithm for the generation of training examples and an inductive learning method (continuous iterative dichotomister 3 "CID3" algorithm) for scheduling knowledge acquisition and rule inferencing. At various points during a simulation run, the system attributes of both the parts and the shop status, which describe the instantaneous characteristics and the best MRU dispatching rule for that instant, have been established and used as training examples.…”