Fusion of Activation Functions: An Alternative to Improving Prediction Accuracy in Artificial Neural Networks

The purpose of this study was to address the challenges in predicting and classifying accuracy in modeling Container Dwell Time (CDT) using Artificial Neural Networks (ANN). This objective was driven by the suboptimal outcomes reported in previous studies and sought to apply an innovative approach to improve these results. To achieve this, the study applied the Fusion of Activation Functions (FAFs) to a substantial dataset. This dataset included 307,594 container records from the Port of Tema from 2014 to 2022, encompassing both import and transit containers. The RandomizedSearchCV algorithm from Python’s Scikit-learn library was utilized in the methodological approach to yield the optimal activation function for prediction accuracy. The results indicated that “ajaLT”, a fusion of the Logistic and Hyperbolic Tangent Activation Functions, provided the best prediction accuracy, reaching a high of 82%. Despite these encouraging findings, it’s crucial to recognize the study’s limitations. While Fusion of Activation Functions is a promising method, further evaluation is necessary across different container types and port operations to ascertain the broader applicability and generalizability of these findings. The original value of this study lies in its innovative application of FAFs to CDT. Unlike previous studies, this research evaluates the method based on prediction accuracy rather than training time. It opens new avenues for machine learning engineers and researchers in applying FAFs to enhance prediction accuracy in CDT modeling, contributing to a previously underexplored area.

Prediction Model of Wax Deposition Rate in Waxy Crude Oil Pipelines by Elman Neural Network Based on Improved Reptile Search Algorithm

A hard problem that hinders the movement of waxy crude oil is wax deposition in oil pipelines.To ensure the safe operation of crude oil pipelines,an accurate model must be developed to predict the rate of wax deposition in crude oil pipelines.Aiming at the shortcomings of the ENN prediction model,which easily falls into the local minimum value and weak generalization ability in the implementation process,an optimized ENN prediction model based on the IRSA is proposed.The validity of the new model was confirmed by the accurate prediction of two sets of experimental data on wax deposition in crude oil pipelines.The two groups of crude oil wax deposition rate case prediction results showed that the average absolute percentage errors of IRSA-ENN prediction models is 0.5476% and 0.7831%,respectively.Additionally,it shows a higher prediction accuracy compared to the ENN prediction model.In fact,the new model established by using the IRSA to optimize ENN can optimize the initial weights and thresholds in the prediction process,which can overcome the shortcomings of the ENN prediction model,such as weak generalization ability and tendency to fall into the local minimum value,so that it has the advantages of strong implementation and high prediction accuracy.

A Hybrid Neural Network Model for ENSO Prediction in Combination with Principal Oscillation Pattern Analyses

El Niño-Southern Oscillation(ENSO)can be currently predicted reasonably well six months and longer,but large biases and uncertainties remain in its real-time prediction.Various approaches have been taken to improve understanding of ENSO processes,and different models for ENSO predictions have been developed,including linear statistical models based on principal oscillation pattern(POP)analyses,convolutional neural networks(CNNs),and so on.Here,we develop a novel hybrid model,named as POP-Net,by combining the POP analysis procedure with CNN-long short-term memory(LSTM)algorithm to predict the Niño-3.4 sea surface temperature(SST)index.ENSO predictions are compared with each other from the corresponding three models:POP model,CNN-LSTM model,and POP-Net,respectively.The POP-based pre-processing acts to enhance ENSO-related signals of interest while filtering unrelated noise.Consequently,an improved prediction is achieved in the POP-Net relative to others.The POP-Net shows a high-correlation skill for 17-month lead time prediction(correlation coefficients exceeding 0.5)during the 1994-2020 validation period.The POP-Net also alleviates the spring predictability barrier(SPB).It is concluded that value-added artificial neural networks for improved ENSO predictions are possible by including the process-oriented analyses to enhance signal representations.

A higher prediction accuracy–based alpha–beta filter algorithm using the feedforward artificial neural network

The alpha–beta filter algorithm has been widely researched for various applications,for example,navigation and target tracking systems.To improve the dynamic performance of the alpha–beta filter algorithm,a new prediction learning model is proposed in this study.The proposed model has two main components:(1)the alpha–beta filter algorithm is the main prediction module,and(2)the learning module is a feedforward artificial neural network(FF‐ANN).Furthermore,the model uses two inputs,temperature sensor and humidity sensor data,and a prediction algorithm is used to predict actual sensor readings from noisy sensor readings.Using the novel proposed technique,prediction accuracy is significantly improved while adding the feed‐forward backpropagation neural network,and also reduces the root mean square error(RMSE)and mean absolute error(MAE).We carried out different experiments with different experimental setups.The proposed model performance was evaluated with the traditional alpha–beta filter algorithm and other algorithms such as the Kalman filter.A higher prediction accuracy was achieved,and the MAE and RMSE were 35.1%–38.2%respectively.The final proposed model results show increased performance when compared to traditional methods.

Prediction on the relative permittivity of energy storage composite dielectrics using convolutional neural networks:A fast and accurate alternative to finite-element method

The relative permittivity is one of the essential parameters determines the physical polarization behaviors of the nanocomposite dielectrics in many applications,particularly for capacitive energy storage.Predicting the relative permittivity of particle/polymer nanocomposites from the microstructure is of great significance.However,the classical effective medium theory and physics-based numerical calculation represented by finite element method are time-consuming and cumbersome for complex structures and nonlinear problem.The work explores a novel architecture combining the convolutional neural network(ConvNet)and finite element method(FEM)to predict the relative permittivity of nanocomposite dielectrics with incorporated barium titanite(BT)particles in polyvinylidene fluoride(PVDF)matrix.The ConvNet was trained and evaluated on big datasets with 14266 training data and 3514 testing data generated form a programmatic algorithm.Through numerical experiments,we demonstrate that the trained network can efficiently provide an accurate agreement between the ConvNet model and FEM by virtue of the significant evaluation metrics R2,which reaches as high as 0.9783 and 0.9375 on training and testing data,respectively.The strong universality of the presented method allows for an extension to fast and accurately predict other properties of the nanocomposite dielectrics.

Prediction of the Shearing Property of Worsted Fabrics Using BP Neural Network

In this paper, three layers of BP neural network were used to model the shearing properties of worsted fabrics. We train the neural network models with 27 kinds of fabrics, and then use 6 kinds of fabrics to validate the accuracy of the model. The result shows that the predicted accuracy of the models is about 85%.

Customized Convolutional Neural Network for Accurate Detection of Deep Fake Images in Video Collections

The motivation for this study is that the quality of deep fakes is constantly improving,which leads to the need to develop new methods for their detection.The proposed Customized Convolutional Neural Network method involves extracting structured data from video frames using facial landmark detection,which is then used as input to the CNN.The customized Convolutional Neural Network method is the date augmented-based CNN model to generate‘fake data’or‘fake images’.This study was carried out using Python and its libraries.We used 242 films from the dataset gathered by the Deep Fake Detection Challenge,of which 199 were made up and the remaining 53 were real.Ten seconds were allotted for each video.There were 318 videos used in all,199 of which were fake and 119 of which were real.Our proposedmethod achieved a testing accuracy of 91.47%,loss of 0.342,and AUC score of 0.92,outperforming two alternative approaches,CNN and MLP-CNN.Furthermore,our method succeeded in greater accuracy than contemporary models such as XceptionNet,Meso-4,EfficientNet-BO,MesoInception-4,VGG-16,and DST-Net.The novelty of this investigation is the development of a new Convolutional Neural Network(CNN)learning model that can accurately detect deep fake face photos.

ENSOMIM:一种新型ENSO时空预测模型

为了提高厄尔尼诺南方涛动(El Ni?o-southern oscillation,ENSO)预测的准确性,解决卷积核难以捕获ENSO的长距离前兆的问题,将ENSO预测视为一个时空序列预测问题,并提出一种基于注意力机制和循环神经网络的ENSO非稳态时空预测深度学习模型,称为ENSOMIM。该模型通过提出的新型注意力机制BGAM来局部和全局交互地学习空间特征,并使用高阶非线性时空网络对长期的时间序列特征进行编码。由于ENSO观测数据集样本数量少,为了更充分地训练模型,采用迁移学习的方法,使用历史模式模拟数据进行预训练再利用观测数据校正模型。实验结果表明,ENSOMIM更适合于大区域和长期的预测。在1984-2014年验证期间,ENSOMIM的Ni?o3.4指数的全季节相关性技巧比经典的卷积神经网络提高16%,均方误差降低17%,它可以为长达18个月的提前期提供有效预测,并且在23个月的提前期内相关技巧达到0.45。因此,ENSOMIM可以作为预测ENSO事件的有力工具。

Effect of Spatial Scale on Modeling and Predicting Mean Cavity Tree Density: A Comparison of Modeling Methods

Cavity trees are integral components of healthy forest ecosystems and provide habitat and shelter for a wide variety of wildlife species. Thus, monitoring and predicting cavity tree abundance is an important part of forest management and wildlife conservation. However, cavity trees are relatively rare and their abundance can vary dramatically among forest stands, even when the stands are similar in most other respects. This makes it difficult to model and predict cavity tree density. We utilized data from the Missouri Ozark Forest Ecosystem Project to show that it is virtually impossible to accurately predict cavity tree occurrence for individual trees or to predict mean cavity tree abundance for individual forest stands. However, we further show that it is possible to model and predict mean cavity tree density for larger spatial areas. We illustrate the prediction error monotonically decreases as the spatial scale of predictions in-creases. We successfully explored the utility of three classes of models for predicting cavity tree probability/density: logistic regression, neural network, and classification and regression tree (CART). The results provide valuable insights into methods for landscape-scale mapping of cavity trees for wildlife habitat management, and also on sample size determination for cavity tree surveys and monitoring.

基于预知维修的小麦播种机运行监控系统设计

为了减少播种机故障频率,提升小麦播种机的播种效率和播种质量,基于预知维修对小麦播种机的运行监控系统进行了设计。系统的主要组成包括主控单片机、检测系统、显示监控系统、报警系统及电源。为了对播种机进行预知维修,将灰色模型和神经网络模型结合,建立了动态灰色神经网络模型,并进行了算法设计。为了验证小麦播种机监控系统性能和预知维修算法的有效性,对其进行了监测精度和趋势预测试验,结果表明:监测系统的监测精度较高,播种机可有效对数据趋势进行预测。

基于CNN-LSTM的水泥熟料f-CaO预测模型

水泥熟料中游离氧化钙(f-CaO)含量的传统人工离线检测缺乏时效性,不利于生产指导。针对离线检测的滞后问题和软测量模型中f-CaO含量与辅助变量的时序匹配问题,提出了一种基于卷积神经网络(convolutional neural network,CNN)和长短时记忆(long short-term memory,LSTM)神经网络的f-CaO含量预测模型。首先,利用滑动窗口截取辅助变量的区间数据;然后,采用CNN提取区间数据的时序特征;之后,构建LSTM神经网络模型;最后,控制截取辅助变量的延迟时间和间隔时间,根据模型预测拟合度提取辅助变量的最优时序特征。仿真结果表明,所提模型提高了水泥熟料中f-CaO含量的预测精度。

基于遗传算法优化BP神经网络的生石膏超细磨预测效果研究

为提高BP神经网络对生石膏超细磨效果的预测准确性,采用Pearson相关系数对超细石膏粉体正交试验产品细度与影响因素的显著性进行分析,并利用遗传算法优化BP神经网络对超细石膏粉体试验产品的d_(50)和d_(90)进行预测,结果表明:超细石膏粉体制备过程中影响细度因素的显著性由大到小依次为排矿口宽度、矿浆质量分数和超细磨时间。利用排矿口宽度和矿浆质量分数两个主要影响因素,利用遗传算法对BP神经网络进行优化,与未优化的BP神经网络相比,经遗传算法优化的BP神经网络具有更高的精度,预测误差也更小,其d_(50)平均绝对误差为0.7575,均方误差为0.7977,均方误差根为0.8931,平均绝对百分比误差为4.4838%;d_(90)平均绝对误差为0.7870,均方误差为0.8294,均方误差根为0.9107,平均绝对百分比误差为1.6658%。研究成果可为超细粉体的制备提供参考。

MEPM模型:基于深度学习的多变量厄尔尼诺-南方涛动预测模型

厄尔尼诺-南方涛动(ENSO)是发生在热带太平洋年际时间尺度的海-气相互作用的异常现象,并由Nino3.4指数表征其发生情况;除此之外,ENSO与众多极端气候事件密切相关。因此,有效的ENSO预测对于预防极端气候事件和深入研究全球气候变化具有重要意义。然而,目前基于深度学习的ENSO预测大多数是预测一个指数或者单一变量,对于模拟多气候要素下的ENSO预测研究较少。通过提出一种利用多气候变量的ENSO预测模型——MEPM模型,其中包括多变量信息提取模块(MIEM)和时空融合模块(STFM),捕获不同气候变量在时空上的相互依赖性,进而提高ENSO预测的准确性。选取了纬向风应力异常(τ_(x))、经向风应力异常(τ_(y))、海表温度异常(SSTA)和海表下150 m温度异常(SSTA150)4个变量的距平值进行ENSO预测。结果表明:MEPM模型在提前11个月的Nino3.4指数相关技巧上分别比北美多模型集合中的动力预报系统CanCM4、CCSM3和GFDL-aer04高10%、20%和14%。此外,MEPM模型在中期Nino3.4指数相关技巧上显著优于其他深度学习模型,并可提供长达17个月的有效预测。

大数据环境下基于BIM与CNN的电力工程造价优化算法

针对大数据环境下电力工程造价在精准化、动态化等方面存在的不足,提出了一种基于BIM与CNN的电力工程造价优化算法。利用BIM技术的特点进行电力工程全生命周期的造价管理,实现了造价的动态化管控。并且采用Levenberg-Marquardt规则算法改进卷积神经网络,通过改进后的CNN网络对每个工程环节的造价完成预测,从而优化整个工程的施工方案。结合相关的电力工程造价数据,基于Matlab对所提算法进行实验测试。结果表明,当学习率为0.010时CNN网络的性能最佳,所提算法的预测准确率为94%,并且与造价的真实值最为接近。

基于EMD-LSTM模型的水轮机组实测摆度信号预测方法研究

水电机组的运行状态直接影响电站及电网的安全稳定,预测机组监测的振动信号有助于改善故障诊断的缺陷。为此,将经验模态分解(EMD)和神经网络模型相结合,提出一种基于EMD-LSTM的水轮机组摆度信号预测模型,将该模型应用于国内某水电站的机组摆度信号预测中,并与LSTM、GA-BP和EMD-GABP模型预测结果进行比较。结果表明,该模型在机组摆度信号的预测方面表现出较高的精度,且优于其他模型。

基于Sine-SSA-BP人工神经网络的腐蚀速率预测研究

海洋工程用钢广泛应用于海洋资源开发;然而,在海洋环境中,由于海洋环境复杂,钢的腐蚀速度大幅加快。为了评估其使用寿命,需要准确地预测钢的腐蚀速率。挂片实验法费时费力,经验模型预测虽然可以快速预测,但因海洋中影响腐蚀的因素较多,准确度较差。本文介绍了一种机器学习方法,即反向传播(BP)神经网络金属腐蚀速率预测模型。本研究创新性地将Sine混沌映射与麻雀搜索优化算法(SSA)引入腐蚀速率预测模型中,并利用2022年采集到的海洋环境要素和腐蚀速率数据导入模型进行训练预测。结果表明,SSA-BP和Sine-SSA-BP神经网络金属腐蚀速率预测模型的误差远低于BP神经网络腐蚀速率预测模型。经过充分的训练和学习,当预测样本数量由5至30逐渐增加时,Sine-SSA-BP预测模型的平均MAPE值为3.5002%,SSA-BP模型的平均MAPE值为6.0900%。

基于CNN-LSTM-AM的大坝变形预测

为提高大坝变形预测模型的预测精度,以长短期记忆(LSTM)作为基础模型预测大坝变形,在LSTM网络层前加入卷积神经网络(CNN)卷积层,以卷积层中卷积核刻画数据的局部模式实现数据特征的深度挖掘,提取大坝变形多因素序列时空特征;LSTM网络层后加入注意力机制层用于区分特征信息的重要程度并给予不同的关注度,进一步优化网络模型,构建了基于CNN-LSTM-AM的大坝预测模型。应用该大坝预测模型在工程实例中与LSTM、CNN-LSTM、LSTM-AM模型的预测结果和残差进行对比分析,CNN-LSTM-AM模型的预测结果和拟合度均更优;并以均方误差、均方根误差、平均绝对误差及决定系数R2作为精度评定指标对比各模型间预测性能,结果表明引入注意力机制能够提升模型预测性能,证实了基于CNN-LSTM-AM构建的大坝预测模型具有工程应用价值。

引力搜索算法优化ENN模型的天然气管道球阀冲蚀深度预测

天然气管道球阀在长期使用中易出现冲蚀现象,准确预测球阀的冲蚀深度对于管道的安全运行具有重要的实际意义。针对传统Elman神经网络(Elman Neural Network, ENN)模型的不足,提出了一种基于引力搜索算法(Gravitational Search Algorithm, GSA)的优化Elman神经网络模型并预测了不同影响因素下球阀的冲蚀深度,探讨了种群规模和隐含层节点个数对优化模型预测精度的影响。结果表明:传统模型预测所得的平均相对误差和均方误差分别为14.382%和0.042 5,优化模型预测所得的平均相对误差和均方误差分别为3.850%和0.003 9,因此,优化模型的预测精度大幅度高于传统模型;随着隐含层节点个数的增加,优化模型的预测精度先升高后降低;种群规模越大并不意味着优化模型的预测精度越高,合理的种群规模可使优化模型达到较好的预测精度;当种群规模和隐含层节点个数不同时,优化模型的预测精度始终高于传统模型,因此所提优化模型具有可靠性,可用于天然气管道球阀冲蚀深度的预测。

基于多层复杂网络的循环神经网络交通量预测模型

针对未安装车流量检测设备的高速公路路段进行短时交通量准确预测,是一个亟待解决的问题。为此,提出一种基于复杂网络的循环神经网络路段短时交通量预测模型。该模型以入口节点交通量为输入,输出路段动态预测交通量。模型由复杂网络、交通小区划分、循环神经网络三个模块组成。复杂网络由多层网络组成,是交通小区划分的基础;交通小区划分模块根据节点特征值,采用聚类方法将节点形成小区,使同小区内节点具有类似特征。最后,以交通小区为依据,将节点交通量合并为小区交通量,采用循环神经网络进行路段动态交通量的预测。通过模型示例并与其他模型预测结果进行对比分析,验证所提模型的准确性和可靠性。结果表明,该模型能够准确地预测不同时长的交通量,MAPE为9.275%,相比于其他方法,预测精度更高且性能稳定,具有重要的应用价值。