江西水利科技
江西水利科技
강서수리과기
JIANGXI HYDRAULIC SCIENCE & TECHNOLOGY
2014年
3期
219-223
,共5页
混凝土强度%高斯过程%机器学习%预测
混凝土彊度%高斯過程%機器學習%預測
혼응토강도%고사과정%궤기학습%예측
Concrete intensity%Gaussian Process%Machine learning%Prediction
针对混凝土强度与其各种影响因素之间存在着复杂的高度非线性关系,传统方法难以准确地预测出混凝土的强度值,本文将具有参数自适应获得,对处理高维数非线性复杂问题具有良好适应性的高斯过程机器学习方法引入到混凝土强度预测领域,并提出相应的预测模型。通过实验数据验证表明,利用高斯过程机器学习模型预测混凝土强度是科学可行的,且预测精度高。利用该模型可以早期预测出混凝土28d的抗压强度,对提高和控制混凝土施工质量具有重要的实际意义。
針對混凝土彊度與其各種影響因素之間存在著複雜的高度非線性關繫,傳統方法難以準確地預測齣混凝土的彊度值,本文將具有參數自適應穫得,對處理高維數非線性複雜問題具有良好適應性的高斯過程機器學習方法引入到混凝土彊度預測領域,併提齣相應的預測模型。通過實驗數據驗證錶明,利用高斯過程機器學習模型預測混凝土彊度是科學可行的,且預測精度高。利用該模型可以早期預測齣混凝土28d的抗壓彊度,對提高和控製混凝土施工質量具有重要的實際意義。
침대혼응토강도여기각충영향인소지간존재착복잡적고도비선성관계,전통방법난이준학지예측출혼응토적강도치,본문장구유삼수자괄응획득,대처리고유수비선성복잡문제구유량호괄응성적고사과정궤기학습방법인입도혼응토강도예측영역,병제출상응적예측모형。통과실험수거험증표명,이용고사과정궤기학습모형예측혼응토강도시과학가행적,차예측정도고。이용해모형가이조기예측출혼응토28d적항압강도,대제고화공제혼응토시공질량구유중요적실제의의。
Bacause the traditional method is difficult to predict accurately the intensity values of concrete and the relationship be-tween concrete strength and its influencing factors is highly complex nonlinear ,the model based on Gaussian Process (GP)ma-chine learning method was proposed in the prediction of the concrete intensity,which access to adaptive parameters,and was good in treating the high-dimensional nonlinear problems. The results of the experimental data study show that GP model is feasi-ble and can forecast the concrete strength precisely and simply. GP machine learning has a good capacity for predicting the 28d compressive strength of concrete. This has an important practical significane to improve and control the quality of concrete con-struction.