SOFT SENSING MODEL BASED ON SUPPORT VECTOR MACHINE AND ITS APPLICATION

Soft sensor is widely used in industrial process control. It plays animportant role to improve the quality of product and assure safety in production. The core of softsensor is to construct soft sensing model. A new soft sensing modeling method based on supportvector machine (SVM) is proposed. SVM is a new machine learning method based on statistical learningtheory and is powerful for the problem characterized by small sample, nonlinearity, high dimensionand local minima. The proposed methods are applied to the estimation of frozen point of light dieseloil in distillation column. The estimated outputs of soft sensing model based on SVM match the realvalues of frozen point and follow varying trend of frozen point very well. Experiment results showthat SVM provides a new effective method for soft sensing modeling and has promising application inindustrial process applications.

Local Partial Least Squares Based Online Soft Sensing Method for Multi-output Processes with Adaptive Process States Division

Local learning based soft sensing methods succeed in coping with time-varying characteristics of processes as well as nonlinearities in industrial plants. In this paper, a local partial least squares based soft sensing method for multi-output processes is proposed to accomplish process states division and local model adaptation,which are two key steps in development of local learning based soft sensors. An adaptive way of partitioning process states without redundancy is proposed based on F-test, where unique local time regions are extracted.Subsequently, a novel anti-over-fitting criterion is proposed for online local model adaptation which simultaneously considers the relationship between process variables and the information in labeled and unlabeled samples. Case study is carried out on two chemical processes and simulation results illustrate the superiorities of the proposed method from several aspects.

A subspace ensemble regression model based slow feature for soft sensing application

A novel adaptive subspace ensemble slow feature regression model was developed for soft sensing application.Compared to traditional single models and random subspace models,the proposed method is improved in three aspects.Firstly,sub-datasets are constructed through slow feature directions and variables in each subdatasets are selected according to the output related importance index.Then,an adaptive slow feature regression is presented for sub-models.Finally,a Bayesian inference strategy based on a slow feature analysis process that monitors statistics is developed for probabilistic combination.Two industrial examples were used to evaluate the proposed method.

Application in soft sensing modeling of chemical process based on K-OPLS method

Aiming at the problem of soft sensing modeling for chemical process with strong nonlinearity and complexity,a soft sensing modeling method based on kernel-based orthogonal projections to latent structures(K-OPLS)is proposed.Orthogonal projections to latent structures(O-PLS)is a general linear multi-variable data modeling method.It can eliminate systematic variations from descriptive variables(input)that are orthogonal to response variables(output).In the framework of O-PLS model,K-OPLS method maps descriptive variables to high-dimensional feature space by using“kernel technique”to calculate predictive components and response-orthogonal components in the model.Therefore,the K-OPLS method gives the non-linear relationship between the descriptor and the response variables,which improves the performance of the model and enhances the interpretability of the model to a certain extent.To verify the validity of K-OPLS method,it was applied to soft sensing modeling of component content of debutane tower base butane(C4),the quality index of the key product output for industrial fluidized catalytic cracking unit(FCCU)and H 2S and SO 2 concentration in sulfur recovery unit(SRU).Compared with support vector machines(SVM),least-squares support-vector machine(LS-SVM),support vector machine with principal component analysis(PCA-SVM),extreme learning machine(ELM),kernel based extreme learning machine(KELM)and kernel based extreme learning machine with principal component analysis(PCA-KELM)methods under the same conditions,the experimental results show that the K-OPLS method has superior modeling accuracy and good model generalization ability.

Soft-Sensing Method of Water Temperature Measurement for Controlled Cooling System

Aiming at the water temperature measuring problem for controlled cooling system of rolling plant,a new water temperature measuring method based on soft-sensing method with a water temperature model of on-line self correction parameter was built.A water temperature compensation factor model was also built to improve coiling temperature control precision.It was proved that the model meets production requirements.The soft-sensing technique has extensive applications in the field of metal forming.

Shadow Detection Method Based on HMRF with Soft Edges for High-Resolution Remote-Sensing Images

Shadow detection is a crucial task in high-resolution remote-sensing image processing. Various shadow detection methods have been explored during the last decades. These methods did improve the detection accuracy but are still not robust enough to get satisfactory results for failing to extract enough information from the original images. To take full advantage of various features of shadows, a new method combining edges information with the spectral and spatial information is proposed in this paper. As known, edge is one of the most important characteristics in the high-resolution remote-sensing images. Unfortunately, in shadow detection, it is a high-risk strategy to determine whether a pixel is the edge or not strictly because intensity values on shadow boundaries are always between those in shadow and non-shadow areas. Therefore, a soft edge description model is developed to describe the degree of each pixel belonging to the edges or not. Sequentially, the soft edge description is incorporating to a fuzzy clustering procedure based on HMRF (Hidden Markov Random Fields), in which more appropriate spatial contextual information can be used. More concretely, it consists of two components: the soft edge description model and an iterative shadow detection algorithm. Experiments on several remote sensing images have shown that the proposed method can obtain more accurate shadow detection results.

A new backtracking-based sparsity adaptive algorithm for distributed compressed sensing

A new iterative greedy algorithm based on the backtracking technique was proposed for distributed compressed sensing(DCS) problem. The algorithm applies two mechanisms for precise recovery soft thresholding and cutting. It can reconstruct several compressed signals simultaneously even without any prior information of the sparsity, which makes it a potential candidate for many practical applications, but the numbers of non-zero(significant) coefficients of signals are not available. Numerical experiments are conducted to demonstrate the validity and high performance of the proposed algorithm, as compared to other existing strong DCS algorithms.

面向无人机绝对定位的遥感影像快速检索方法

针对在复杂环境下无人机景象匹配导航中的视觉绝对定位问题,提出了一种聚合深度学习特征的实时影像快速检索方法。首先,引入可训练软分配深度学习框架—NetVLAD,结合VGG16网络提取并聚合生成影像稳定的全局特征表达向量;其次,在初始检索阶段,使用KD树结构对影像全局特征向量构建检索索引,在不损失检索精度的前提下提高检索速度;最后,使用皮尔逊积矩相关系数对初始检索结果进行快速预判断,自动过滤初始检索结果,对于需要重排序的影像则采用特征学习匹配算法——图神经网络SuperGlue进行匹配重排序。所提方法在公开的夏季和冬季遥感影像数据集分组进行实验,实验结果表明:未重排序条件下,初始检索结果第一张影像平均准确率达到了58.27%,部分特征较好地区准确率达到了85%,对不同时相遥感影像也有很好的适应性,平均检索一张影像耗时3.7 s,可为无人机景象匹配导航的初始定位提供参考。

软测量技术赋能燃煤电厂碳排放计量的研究进展

火力发电企业作为我国能源结构的重要组成部分,长期以来是我国碳排放的主要来源,在我国和全球加速推动低碳经济发展的宏观环境下,火电企业积极响应国家“能耗双控”向“碳排放双控”转变的战略部署。在此背景下,精确计量燃煤电厂的碳排放量变得至关重要。在燃煤电厂碳计量中,烟气流量影响燃煤发电中在线监测法的精度,而燃煤消耗量、燃煤元素碳含量以及飞灰碳含量共同决定核算法的可靠性。目前,大多数燃煤发电企业只对流量和燃煤消耗量进行实时监测,在现场恶劣的环境中对燃煤元素碳含量以及飞灰碳含量进行短周期、高频次的直接监测需要花费较大的人力以及物力,流量监测设备也易受烟道环境影响。而软测量技术以其高效和低成本的特点,可为传统碳排放计量过程中关键参数的监测提供一种替代方法。鉴于此,首先阐述了软测量模型的建立过程,包含数据预处理、辅助变量选择、软测量模型建立以及模型校正。数据预处理能够确保数据质量,提高建模效率;辅助变量选择是从大量潜在的变量中筛选出对目标变量的辅助变量,进一步提高建模效率;软测量模型建立主要是基于机理建模和数据驱动建模,是实现目标变量预测的核心;模型校正通过实际的离线或在线数据,对模型进行进一步优化,提高模型的预测精度。其次,针对碳计量相关参数,分析了烟气流量、燃煤消耗量、燃煤元素碳含量和飞灰碳含量监测存在的问题,论述了软测量技术在上述碳计量关键参数的国内外研究进展和应用,评估了机理建模和数据驱动建模技术的有效性、准确性和实用性。其中,机理分析建模主要基于电厂锅炉进出口的能量平衡以及烟风质量守恒等原理,有着确定的数学物理关系式,具有高度可解释性和稳定性,但是建模过程复杂,预测精度较低;数据驱动建模主要是利用各种机器学习方法,基于电厂分布式控制系统(Distributed control system,DCS)丰富的运行数据,对碳计量关键参数进行“黑箱建模”,克服了机理分析建模复杂的过程分析,精度相对较高,但是建模过程不明确,且模型对于不同机组的泛化能力较差。最后,对于软测量技术在碳排放计量领域的发展应用进行了总结与展望。对电厂各参数之间的时序结构、电厂自身计算能力的限制以及机理分析融合数据驱动方法的发展提出相关建议,并对国外二氧化碳预测性排放系统结合软测量技术在国内外燃煤电厂的应用进行展望。

WiCare:一种非接触式的老人如厕跌倒监测模型

老人在卫生间内的跌倒行为存在因救助及时性差而导致严重危害的风险,因此高效快捷的如厕跌倒监测研究具有重要意义。针对当前基于Wi-Fi感知的跌倒监测方法中存在的受噪声影响大而特征提取不充分、监测精度有限的问题,提出了一种基于多级离散小波变换和软阈值处理的信号降噪算法,及一种融合卷积神经网络、双向长短期记忆网络及自注意力机制的非接触式如厕跌倒监测模型WiCare。首先,从原始CSI数据中提取振幅作为基础数据;其次,使用多级离散小波变换和软阈值处理进行感知数据降噪;然后,将感知数据进行多维重构,以更准确地表征跌倒行为特征;最后,利用WiCare提取感知数据中的有效特征,进而实现卫生间如厕跌倒行为监测功能。实验结果表明,WiCare在居家卫生间环境下对跌倒行为监测的准确率为99.41%,与其他同类模型相比,WiCare的识别准确率高,模型复杂度低,且泛化能力更强。

一种面向多工序复杂制造过程的质量软测量方法

复杂制造过程关键质量变量的精准感知,是实现系统优化控制和保障系统安全稳定运行的必要前提。考虑复杂制造过程具有生产工序众多、回路互联耦合、工序质量遗传、数据时空分布等复杂特性,使得对过程质量的精准感知面临诸多困难。在此背景下,本文提出一种考虑过程时延的基于mRMR–GA–ResNet的多工序复杂制造过程质量软测量建模方法。首先,构建了一种考虑过程变量与质量变量间时延的基于最小冗余最大相关(mRMR)和遗传算法(GA)的多传感器过程变量筛选方法,以确定最优特征子集;其次,基于各工序的最优特征子集,设计了一种3维(特征–时间–工序)样本空间表征方法,工序内部以2维(特征–时间)形式表征,将工序作为通道构建3维(特征–时间–工序)样本,通过残差网络进行时间–空间特征提取,进而通过局部–全局特征融合得到最终的质量预测值;最后,通过一个实际制造过程——浮法玻璃生产过程,进行了实验验证。结果表明:在选择特征数相同的前提下,相较于其他4种基于相关性的特征选择方法(PCC、SCC、MI、MIC),本文所提多传感器过程变量筛选方法对于模型有更好的预测性能。以残差网络作为预测模型,本文所提3维样本构造方法,相较于传统的2维样本构造方法,对于模型的预测精度有了一定的提升,均方根误差ERMS、平均绝对误差EMA、对称平均绝对百分比误差E_(SMAP)分别提升9.2%、10.8%、9.8%,验证了所提方法的有效性。

基于注意力机制和软匹配的多标签遥感图像检索方法

针对卷积神经网络对于多标签遥感图像特征提取能力弱、不能准确反映遥感图像多标签复杂性的问题,提出基于注意力机制和软匹配的多标签遥感图像检索方法。在特征提取阶段,以密集卷积神经网络模型为基础,在每个密集块(Dense Block)后添加CBAM(Convolutional Block Attention Module)层,实现对多标签图像区域特征提取。在模型训练时,利用区分硬匹配与软匹配的联合损失函数,学习图像的哈希编码表示。通过评估遥感图像哈希编码间的汉明距离,实现相似图像的检索。实验结果表明,所提方法在数据集NUS-WIDE和多标签遥感图像数据集DLRSD上与其他基于全局特征的深度哈希方法相比,明显提升了检索准确率。

基于增强梯度算子的软阈值宽带频谱感知算法

为了改善信号梯度特征对幅度的损失以及寻求描述信号的最佳尺度问题,提出了一种基于增强梯度算子的软阈值宽带频谱感知算法。通过引入梯度增强算子还原信号幅值特征,结合信号本身梯度特征,使用不同的尺度描述信号梯度增量,得到软阈值判据,进一步加入尺度融合单元,利用硬阈值加软阈值联合判断的方法,得到描述信号的最佳尺度。理论分析和仿真实验结果表明,在高斯信道和瑞利衰落信道下,相较于MPSG算法,所提算法的检测概率和虚警概率均有明显改善,且复杂度更低。通过对比实测数据的检测效果,所提算法更适用于实际工程中。

计及改进粒子群算法优化BP神经网络的沼气产量软测量预测模型

为准确预测大中型沼气工程的日产气量,提出一种利用基于PSO-BP模型的软测量方法。首先,依托软测量技术选取参数;其次,以进料量、发酵温度、液位、罐内液压等参数作为输入量,沼气日产量为输出量进行模型建立。在此基础上,使用线性降低权重系数法和引入变异算子对粒子群算法进行改进,并对BP神经网络进行初始化来提高模型性能。通过实验比较改进PSO-BP模型、传统BP神经网络以及遗传算法优化的BP神经网络在预测沼气日产量方面的性能,采用改进的PSO-BP模型进行预测时,均方根误差(RMSE)、决定系数(R2)和平均绝对误差(MAE)分别为1.38440、0.84011和1.00910,证明改进PSO-BP模型结合软测量技术对进行复杂非线性牛粪高温厌氧发酵过程预测的可行性,同时可保证预测结果的精准性。

污水处理软测量仪表研究进展与应用

软传感器以其检测快速、成本低廉等优点应用于污水处理行业前景极佳,随着人工智能的进步,软测量仪表的精度与可靠度也在逐步提高。该文在总结污水处理软测量研究进展的基础上,研究了软测量仪表开发的关键技术,并分析了某膜曝气生物反应器(MABR)农村污水处理设施出水端的应用案例。

Double-kill contribution of high-roughness high-density porous carbon electrodes to mechanically self-sensing supercapacitors

Impact detecting and counting are fundamental functions of fuses used in hard target penetration weapons.However,detection failure caused by battery breakdown in high-g acceleration environments poses a vulnerability for such weapons.This paper introduces a novel supercapacitor that combines energy storage and high-g impact detection,called self-sensing supercapacitor.By deliberately inducing a transient soft short-circuit during shock in the supercapacitor,it is possible to detect external impact by its transient voltage drop.To realize this concept,firstly,by introducing the contact theory and force-induced percolation model,the electrode strength and roughness are found to have key impacts on the formation of soft circuits.Subsequently,to meet the needs for sensitivity and capacity,a high-density porous carbon(HDPC)that combines high mechanical strength and porosity,is selected as a suitable candidate based on the analysis results.Furthermore,a two-step curing method is proposed to prepare the high-roughness HDPC(HRHDPC)electrode and to assemble the self-sensing supercapacitor.Due to the rich specific surface of the electrodes and the high surface strength and roughness conducive to the formation of transient soft short circuits,the self-sensing supercapacitor not only possesses an excellent specific capacitance(171 F/g at 0.5 A/g)but also generates significant voltage response signals when subjected to high-g impacts ranging from 8000g to 31,000g.Finally,the self-sensing supercapacitor is applied to compose a successive high-g impact counting system and compared to traditional solutions(sensors and tantalum capacitors)in the military fuzes.The results show that the self-sensing supercapacitor-based system exhibits advantages in terms of size,power consumption,and counting accuracy.

基于光学触觉传感器的肿块检测研究

肿块检测可以帮助患者更早地发现潜在的疾病,提高患者的生存率。针对如何使用非侵入式方法检测组织样本中的肿块这一问题,制作了一款结构简单、校准方便的光学触觉传感器,并提出了一种基于表面重建的肿块判别算法。该方法对触觉传感器采集到的原始图像进行深度重建,并从重建图像中减去正常组织产生的基线图像,从而获得由肿块产生的异常触觉图像,并将处理完的图像送入神经网络中进行肿块检测。结果表明,该方法能够有效地分辨出直径在6~10 mm、嵌入深度在1~5 mm的肿块,验证了所述方法的实用性与可行性。最后,基于该方法与流程,讨论了肿块大小、肿块嵌入深度和正常组织硬度对肿块检测结果的影响。提出的非侵入式的肿块检测方法探索了光学触觉传感器在肿块检测领域中的可行性,为相关领域进一步的研究提供了新的思路。

基于软决策融合的CSS性能分析及优化

针对基于数据融合的软合并协作频谱感知占用通信资源较大问题,提出了一种基于软决策融合的协作频谱感知方案,推导了检测性能表达式,并基于NP准则给出了使检测率最大化的优化的协作频谱感知算法。分析和仿真结果表明:基于软决策融合的协作感知相比基于数据融合的协作感知可以有效节约系统资源;所提优化算法可以有效提高检测性能。

基于神经网络的FPSO原油来液量软测量方法

现有FPSO原油处理工艺流程中,由于原油携带沙子较多,经常导致流量计堵塞,从而影响原油来液测量的准确性。现场操作人员无法准确确定破乳剂的添加量,只能依靠经验进行操作,通常采用过量添加破乳剂的方式来确保处理后的原油符合要求。为了解决上述问题,采用LSTM神经网络对原油来液进行软测量建模,选取工艺流程中相关参数的测量值作为关联特征,通过训练模型并将预测结果与真实值进行对比。研究结果表明,所提方法输出结果的误差较小,可以进一步优化并推广至原油处理工艺流程的实时控制中,为实现破乳剂的自动控制提供了数据基础。

基于FRBPSO-RBF神经网络的污水BOD5软测量方法

污水处理过程中污水BOD5难以实时准确测量,故软测量方法逐渐被用于污水BOD5的预测,其中RBF神经网络软测量方法应用广泛,但存在训练过程易陷入局部极值等问题。为提高RBF神经网络的预测精度,提出了基于适应度排名的粒子群算法(fitness ranking based particle swarm optimization,FRBPSO),根据适应度排名与迭代次数确定惯性权重的大小,并根据粒子个体历史最优值的排名与迭代次数确定自我学习因子与社会学习因子的大小,并将FRBPSO算法引入RBF神经网络的参数训练中。基于13个基准测试函数与其他3个粒子群优化算法对比,实验结果显示FRBPSO算法的寻优能力相对较强。再将基于FRBPSO算法的RBF神经网络用于构建污水BOD5软测量模型,仿真结果表明,在测试数据中,FRBPSO-RBF软测量模型的平均绝对误差比PSO-RBF软测量模型、DAIW-RBF软测量模型、SCVPSO-RBF软测量模型分别降低了0.7178、0.2402、0.5851,平均绝对百分比误差分别降低了0.47%、0.15%、0.33%,均方根误差分别降低了0.0034、0.0015、0.0039。与其他3个基于PSO算法的BOD5软测量模型相比,FRBPSO-RBF模型具有较高的BOD5预测精度。