The Method Based On Bayesian Network In Assessment Of Cleaner Production In ECO-Industrial Park

Ecology industrial park (EIP) provides a basis for applying the concepts ofIndustrial Ecology and Circulation Economy and serves as an ultimately necessaryapproach to achieve sustainable development. Being an independent andcutting-edge research subject, EIP is alsoattracting worldwide attention fromgovernments, large enterprises and research institutions. Relative theories andindustrialpracticeshavebeenshowingastrongflourishingtrend.Efficient Assessment of the construction and development level of EIP willpromote complete comprehension towards the impacts of various factors onconstruction of EIP andwill provide advice for management and policies. ThusscientificandcomprehensiveevaluationofEIPisofmuchimportanceforpromotingtheameliorationanddevelopmentofnativeEIP.The aim of our research is to discuss and then design approaches and methods forassessing the development level of EIP scientifically. In the construction process ofthe ecology industrial park, the cleaner production is construction premise in thepark and the necessary tache in the ecology industrial system constructing, namelyin system each link must carry on the source reducing, achieves the cleanerproduction. Carring on the cleaner production for EIPvarious enterprises, accordingto certain order, namelyneeds to determine which enterprises will be carried out theshort-term plan of cleaner production and which is the long-dated plan.Eco-industrial park took the main practice form of the ecology industry idea, indomestic and foreign all obtained has widely taken with the promotion. Therealization of EIP is also so varied EIP may divide into the brand-new planapproximately, the virtual network and the existing transformation. In thetransformation ecology industrial park construction, through implementing cleanerproduction, It is advantageous for the construction of ecology chain furthest toreduce the existing waste capacity in the enterprise, roundly achieving cleanerproduction, raising the resources use factor in the park, reducing the pollutant to produce.At present domestic and foreign only limited to the gray relational analysismethod, fuzzy comprehensive evaluation, analytic hierarchy process,the artificialneuralnetworksontheresearchofassessmentmethodoftheecologyindustrialpark.In this foundation, A new kind of assessment method based on the Bayesiannetworks is provided in the paper and applied in the ecology industrial existingenterprise to carry on the promoting cleaner production, through establishingappraisaltargets andtheevaluationcriteriaandthenapplyingtheBayesiannetworksfor each kind of appraisal target modeling, carrying on the cleaner productionassessment by this to the ecology industrial existing enterprise, then makingenterprise to carry on the order of the cleaner production.Since The Bayesiannetwork has proposed, along with the fundamental research being thorough and thelevel of application unceasingly enhancing , It already has been obtained theacademic circles widely approved and regarded,for the decision-making of thecomplex problems, the uncertainty knowledge inference, domains of intelligentdiagnosis and reliable appraisal ,and so on, the Bayesian network even shows itsobvious superiorityand charm. Comparingwith the grayrelational analysis method,fuzzy comprehensive evaluation, analytic hierarchy process, the artificial neuralnetworks , The main superiority lies in the solid theoretics, effective processingincomplete data, unifying with other technologies can carry on the cause and effectanalysis and cause the apriori knowledge and the data organic union. The Bayesiannetwork can unify the apriori information and the sample data organically, unifyingthe subjectivity and objectivity organically, reflecting the data object intrinsicrelation and the essence comprehensively. Not only may avoid because thesubjective factor possibly creates prejudice, but also the noise problem which thesample data brings, the Bayesian network is to the data object is highlyabstract andbroad, the latent relatedness can be expressed among the variables with the succinctgraphic model, clear semantics, natural and direct-viewing. As a result of theBayesian network used the strict mathematics inference method, the inference obtainedtheresultsandconclusioniscredible,advantageousfortheexplanationandeasy to understand. Therefore, the appraisal target which reflect various enterprisescleaner production degree carries on modeling by the method of Bayesian network,through the Bayesian network which obtained may forecast the prior degree of eachenterprise cleaner production, it mayachieve the good effect. The Bayesian networkaccording to many enterprise's actual situation in the region, the most superiorcriterion of cleaner production degree is constructed in EIP, suitable to appraise thelevel of cleaner production in many enterprise. This research makes use of theBayesian network to calculate the probability of each enterprise, compared withthem betweentheirdegreeofcleanerproductionandfindthefirst enterprisetocarryon cleaner production, defining next step to carry on the key enterprise of cleanerproduction and advancement of cleaner production forward plan. Providing thetechnical route and the scientific basis for the similar ecology industrial park onpromotingcleanerproductionandthemanagementdecision-makingofEIP.The Bayesian network was applied in the ecology industrial existing enterprisepromoting cleaner production, the concrete step is: First using the equilibratedmethod to the data to carry on the zero dimension for the primary data, thendiscretize each target value, after the discretization through a kind of improvementbasedontheincreasingstudymethodoftheimmunitygeneticalgorithm,studiedtheBayesian networks, then by the Bayesian networks inferring, the order of cleanerproduction in the ecology industrial enterprise was carried through thedecision-making.The immunity genetic algorithm gave the adaptive function through thedescription length function transmutating of the Bayesian network model to judgethe network architecture fit or unfit, controlled the direction of the evolution usingthe boundedness of the immunity principle, then designed a kind of the immunityevolution algorithm based on studying the Bayesian network architecture. But thisalgorithm belongs to the batch learning algorithm, the efficiency is much lowerunder the increasing environment, but the learning question in the article make suitably the increasing study, therefore this article has made improvements to theimmunity genetic algorithm, and formed a Bayesian network increasing learningalgorithm based on the improved immunity genetic algorithm, this algorithmimproved and increased some immune operator in the original algorithm, caused theindividual through the evolution process in sufficiency function to adapt the recentdata, through the vaccinating vaccine, make the individual and the legacy data alsomatchwell,thestudyalgorithm is moresuitableforthedecision-makingquestionofcleanerproductionwhichisfitforincreasedlearningenvironment.The experiments were carried on in longmen ecology industrial coking enterpriseas the example, the result is consistent with the gray connection analysis result,Which explained that the appraisal method of the Bayesian network can unifieseffectively the qualitative analysis and the quantificational target, and realized theoverall quantification appraisal, i The method is a veryuseful one for assessment ofguiding cleaner production promotion in eco-industrial park with the advantages ofreliable results. The application of the Bayesian network in the industrial ecologydomain is a new attempt, how to apply the bayesian network in more widespreadecologyindustrialvariousdomainsneedfurtherstudy.