| CPC G06N 3/086 (2013.01) [G06N 3/045 (2023.01)] | 9 Claims |

|
1. A method for dynamic evolution of a composite service, comprising the following steps:
(1) when a new service request occurs, firstly importing an initial composite service CS, and initializing an evolution threshold TH1 and an evolution success threshold TH2 at the same time;
(2) taking the composite service CS as an input, calculating an evolution probability EP through a neural network NN1;
(3) comparing the evolution probability EP with the evolution threshold TH1, and if EP>TH1, then performing step (4), otherwise, performing step (11);
(4) performing distributed service evolution;
(5) selectively performing multi-layer process evolution;
(6) responding to the request through an evolved composite service CS′;
(7) recording the performance CS′.PF of the evolved composite service CS′, and adding a record into a set SPF for the performance of the composite service;
(8) taking the composite service CS and the evolved composite service CS' as inputs, calculating an evolution success probability SP through a neural network NN2;
(9) back-propagating, and updating parameters of the neural networks NN2, NN1;
(10) comparing the evolution success probability SP with the evolution success threshold TH2, and if SP>TH2, then performing step (14), otherwise, performing step (13);
(11) responding to the request through the composite service CS;
(12) recording the performance CS.PF of the composite service CS, and adding a record into the set SPF for the performance of the composite service;
(13) retaining the composite service CS as an initial service for a next run, and ending;
(14) making the evolved composite service CS' as an initial service for a next run, and ending.
|