计算机与应用化学
計算機與應用化學
계산궤여응용화학
COMPUTERS AND APPLIED CHEMISTRY
2013年
9期
1055-1058
,共4页
杨光%刘俏%代蕊%马蓬勃%刘海霞
楊光%劉俏%代蕊%馬蓬勃%劉海霞
양광%류초%대예%마봉발%류해하
Bacillus amyloliquefaciens Q-426%摇瓶培养%抑菌活性%BP神经网络%预测
Bacillus amyloliquefaciens Q-426%搖瓶培養%抑菌活性%BP神經網絡%預測
Bacillus amyloliquefaciens Q-426%요병배양%억균활성%BP신경망락%예측
Bacillus amyloliquefaciens Q-426%shake flask%antibacterial activity%BP neural network%predictio
为了寻求工业培养基有效配比,以获得Bacillus amyloliquefaciens Q-426发酵培养较大活性,根据Bacillus amyloliquefaciens Q-426摇瓶培养实验数据,以甘蔗糖蜜、豆粕粉、氯化铵、氯化钙4个培养基成份作为影响抑菌圈直径的主要因素,利用BP(Back propagation)神经网络方法对其抑菌圈直径进行预测,获得抑菌圈直径的BP神经网络预测模型。结果表明:训练BP神经网络时,14组自变量数据中抑菌圈直径的BP神经网络拟合值与实测值的相对误差为-1.8519%~1.9231%,相对误差绝对值的平均值为1.1213%;测试BP神经网络时,3组自变量数据中抑菌圈直径的BP神经网络预测值与实测值的相对误差为-1.8519%~0.7692%,相对误差绝对值的平均值为1.1515%,说明建立的基于4个培养基成份的抑菌圈直径BP神经网络预测模型是可行的。
為瞭尋求工業培養基有效配比,以穫得Bacillus amyloliquefaciens Q-426髮酵培養較大活性,根據Bacillus amyloliquefaciens Q-426搖瓶培養實驗數據,以甘蔗糖蜜、豆粕粉、氯化銨、氯化鈣4箇培養基成份作為影響抑菌圈直徑的主要因素,利用BP(Back propagation)神經網絡方法對其抑菌圈直徑進行預測,穫得抑菌圈直徑的BP神經網絡預測模型。結果錶明:訓練BP神經網絡時,14組自變量數據中抑菌圈直徑的BP神經網絡擬閤值與實測值的相對誤差為-1.8519%~1.9231%,相對誤差絕對值的平均值為1.1213%;測試BP神經網絡時,3組自變量數據中抑菌圈直徑的BP神經網絡預測值與實測值的相對誤差為-1.8519%~0.7692%,相對誤差絕對值的平均值為1.1515%,說明建立的基于4箇培養基成份的抑菌圈直徑BP神經網絡預測模型是可行的。
위료심구공업배양기유효배비,이획득Bacillus amyloliquefaciens Q-426발효배양교대활성,근거Bacillus amyloliquefaciens Q-426요병배양실험수거,이감자당밀、두박분、록화안、록화개4개배양기성빈작위영향억균권직경적주요인소,이용BP(Back propagation)신경망락방법대기억균권직경진행예측,획득억균권직경적BP신경망락예측모형。결과표명:훈련BP신경망락시,14조자변량수거중억균권직경적BP신경망락의합치여실측치적상대오차위-1.8519%~1.9231%,상대오차절대치적평균치위1.1213%;측시BP신경망락시,3조자변량수거중억균권직경적BP신경망락예측치여실측치적상대오차위-1.8519%~0.7692%,상대오차절대치적평균치위1.1515%,설명건립적기우4개배양기성빈적억균권직경BP신경망락예측모형시가행적。
In order to acquire higher antimicrobial activities based on Bacillus amyloliquefaciens Q-426 fermentation, a prediction of size of inhibiting zones (SIZ) was carried out with BP neural network based on the experimental data of Bacillus amyloliquefaciens Q-426 in shake flask. Taking cane molasses, soybean meal, NH4Cl and CaCl2 as influence factors, the prediction model of SIZ based on BP neural network was obtained. The results show that the relative error of fitting value and predicting value of SIZ compared with the observed value for 14 groups and 3 groups of independent variables are from-1.8519%to 1.9231%and from-1.8519%to 0.7692%based on BP neural network model, respectively. The average value of absolute value of relative error are 1.1213%and 1.1515%, respectively. Therefore, the prediction model of SIZ with BP neural network based on 4 medium compositions is doable.