Calculation Method Based On Genetic Neural Network Working Hours Fixed

The management of man-hour quota is an important foundation management in enterprise, and it plays a great role in enterprise planning management, economic accounting, manufacturing schedule regulation and control. Searching standard man-hour quota table is a popular, important method to calculate man-hour quota, but when the method is used to calculate non-standard nodes in man-hour quota table, it has large errors and is hard to compute accurate man-hour quota. Artificial neural network and genetic algorithm, which have gradually sophisticated applications in recent years, are widely used in function approximation, data fitting, structural optimization, etc, and have high computation accuracy. Therefore, in order to increase computation accuracy of man-hour quota, the method of calculating man-hour quota based on artificial intelligence is studied, which has great significance in increasing production control, optimizing equipment utilization and designing production cycle scientifically.Artificial neural network and genetic algorithm are used to train and calculate man-hour quota, and several key technologies are deeply studied, including neural network design, genetic algorithm design and genetic neural network design. Through the exploitation and application of prototype system, the accuracy and superiority of man-hour quota computation based on genetic neural network are proved. The study makes useful explorations in the man-hour quota computation based on genetic neural network. The main research results in theory and practice are as follows:(1) Based on the characteristics of standard man-hour quota table, the method of hide layer node calculation, training algorithms and data normalization are analyzed, and the factors that affect generalization error of neural network are presented, such as data set, neural network architectures and convergence error. In the end, training module and computation module based on neural network are established in MATLAB. The application results show that training man-hour quota with neural network is superior.(2) In view of disadvantages of BP neural network, such as getting into local minimum easily and weak global search ability, genetic algorithm (GA) is adopted to optimize the parameters of neural network. The real coding method is proposed to implement genetic neural network, and selection operation, cross operation, mutation operation in genetic algorithm are analyzed and improved. In the end, the application results show that genetic neural network is valid in man-hour quota computation.(3) Via the admixture programming technology between MATLAB 6.5 and Visual C++ 6.0, a man-hour quota calculation prototype system based on genetic neural network is designed and developed, the function modules, including data preprocessing, neural network training, genetic algorithm optimization, neural network model management, man-hour quota calculation and fast training and calculation are presented in details. The practical application shows that the system has high computation accuracy and universality, and can run without the support of MATLAB.