Study On The Application Of Neural Network Model Of Bidding Decision For Highway Engineering

On the foundation of analysis for before related theoretical and model, combining to recruit enter a bid working characteristic, because it is a complex decision course that fills with uncertainty to comment to mark and enter a bid, to influence by various factor, for such problem, it is very difficult to establish accurate mathematics model, is even been impossible, to solve this difficulty, this paper has put forward, comment based on vague nerve network to mark and enter a bid decision model, and mark and enter a bid with vague nerve network comment to make policy model, have put forward further to Use Judge theoretical and computer technology decision theory and vague nerve network, have developed highway project comment mark and enter a bid to make policy support system.According to actual condition of highway project, appraisement index is definite and make up each appraisement index to classify 11 indexes, establishment comment to mark decision vague nerve network model, since commenting to mark systematic complex, appraisement parameter, if contain the parameter owned with a network, can make entire network complex and huge, is hard to train, so adopt the form of sub-network and combination network. Comment to mark decision vague nerve network model from 5-sub network and a combination network composition. The structure of combination network and sub-network is to adopt standard vague nerve network. The input of sub network is the unitized value of related index, is exported for the appraisement result of sub-network, the input of combination network is the output of each sub-network, is export to be entire to invest in the comment of project to mark decision result.With the related data information of actual project inspect to enter a bid the adaptability of decision model. So to comment to mark decision, offer the correctness and scientific ness that can carry out comparison and the basis of differentiating and raises decision. According to the principle of fuzzy mathematics have established to build project differentiate model, have put forward a kind of practical effective method for project practice. In which, subjective judgment lies in weight distribution and the row order of single factor item, these standards show have accumulated enough experience, and usually reflect in the comment of the file of public bidding to mark program in. Have just transparency, other judge operation complete according to the logicality of fuzzy mathematics go on, thus the comprehensive index that gets finally is objective and scientific.This paper still adopt 12 former enter a bid example it and the decision result of training and verifying vague nerve network from explanation function, them have been divided into two groups of 10 example data is used to train vague nerve network, 2 example data are used to verify vague nerve network, these data of adopting pass the pretreatment that adopts vague principle, have carried out the influencing factor in reality quantitative analysis, after being handled, normalization forms.From inspection result, expectation output is consistent with actual export basic last maintenance, therefore this model is effective, can be used to enter a bid to make policy. But because of original train data lose comparatively, still can existent greater error, in reality is utilized, it does not be very accurate result to also have possibility to reach, but can know to count the related data that accumulates the reality of project continuously according to storehouse, after having enough source data, the decision effect of model will be better.