Implementation Of Neural Network In Process Problem

Neural networks are widely utilized in process planning problems. The process planner learns mappings among input patterns, consisting of the features and attributes of a past, and output patterns, consisting of sequences of machining operations to apply to these parts. Hence neural networks offer a promising solution for automating the learning of process knowledge.

Problem Representation

The method or process planning tasks may be presented by the transformation

                                                                     F A → C,......................Eqn(23)

Whereas

-          F is a set of part feature,

-          A is a set of feature attributes,

-          C as a set of feasible operation sequences,

-          And → indicates a mapping function.

Process planners are interested in those features such are generated by several sequence of machining process. Typical feature involves holes or threaded and unthreaded, external cylinders or threaded and unthreaded, slots, faces, keyways, and gears. All feature is associated along with a set of attributes that define this from a manufacturing standpoint. These involve dimensions, tolerances, and surface finish needs.

Based upon the particular values of feature attributes, the process planner can identify the sequence of operations essential to produce the feature. All sequence corresponds to a exact classification of the input pattern. In the neural network the transformation function is embedded in the connection of networks weights through training. Following diagram demonstrates how inputs are physically presented to the network. All known feature is associated along with an input unit, which is highly active (+ 1) if the feature is represent, and inactive (0) or else. All known attribute is associated also with an input unit. The input unit takes on the value of the attribute, normalized to lie between 1 and 0.

The features composing the part being planned are represented to the network one at a time, with their corresponding attributes. The networks response to the feature pattern presents selection of machining operation. If the activation of the output unit is positive, this is interpreted as the selection of the corresponding machining operation is supported. A threshold mechanism chooses the operation whose output unit has the highest positive activation above any threshold value as presented in Eq. 6. This mechanism can be implemented in a neural network though lateral inhabitation in the output layer.

                                Diagram of: A Neural Network for Process Planning Knowledge Acquisition

Note that output units are located to single operations rather than sequences. To learn sequencing constraints, this is necessary to give encoding of operation position inside a sequence. Encoding may be accomplished directly by assigning one output unit to every possible sequence or, for every operation, to every possible position of that operation in a sequence. Neither of these encodings exactly reflects the approach of the expert process planner, though, and both require a large number of output units, many of that are infrequently utilized. An expert process planner builds a sequence of operations for every feature, with every operation being added depending on an exact, but not essentially identical set of attributes. In order to select operations individually, yet keep correct sequencing, the outputs of the network may be fed back as inputs, hence serving as a context for a decision on the next operation in the sequence. Reoccurrence ends while a null output is obtained; which is, all output units have activations below the threshold value, signaling the sequence and by utilizing recurrence, output units are efficiently utilized, along with little sacrifice in the final performance speed of the network.