Dynamic evolution and trend prediction of multi-scale green innovation in China

Numerous studies deal with spatial analysis of green innovation(GI).However,researchers have paid limited attention to analyzing the multi-scale evolution patterns and predicting trends of GI in China.This paper seeks to address this research gap by examining the multi-scale distribution and evolutionary characteristics of GI activities based on the data from 337 cities in China during 2000-2019.We used scale variance and the two-stage nested Theil decomposition method to examine the spatial distribution and inequalities of GI in China at multiple scales,including regional,provincial,and prefectural.Additionally,we utilized the Markov chain and spatial Markov chain to explore the dynamic evolution of GI in China and predict its long-term development.The findings indicate that GI in China has a multi-scale effect and is highly sensitive to changes in spatial scale,with significant spatial differences of GI decreasing in each scale.Furthermore,the spatiotemporal evolution of GI is influenced by both geospatial patterns and spatial scales,exhibiting the“club convergence”effect and a tendency to transfer to higher levels of proximity.This effect is more pronounced on a larger scale,but it is increasingly challenging to transfer to higher levels.The study also indicates a steady and sustained growth of GI in China,which concentrates on higher levels over time.These results contribute to a more precise understanding of the scale at which GI develops and provide a scientific basis and policy suggestions for optimizing the spatial structure of GI and promoting its development in China.

GHM-FKNN:a generalized Heronian mean based fuzzy k-nearest neighbor classifier for the stock trend prediction

Stock trend prediction is a challenging problem because it involves many variables.Aiming at the problem that some existing machine learning techniques, such as random forest(RF), probabilistic random forest(PRF), k-nearest neighbor(KNN), and fuzzy KNN(FKNN), have difficulty in accurately predicting the stock trend(uptrend or downtrend) for a given date, a generalized Heronian mean(GHM) based FKNN predictor named GHM-FKNN was proposed.GHM-FKNN combines GHM aggregation function with the ideas of the classical FKNN approach.After evaluation, the comparison results elucidated that GHM-FKNN outperformed the other best existing methods RF, PRF, KNN and FKNN on independent test datasets corresponding to three stocks, namely AAPL, AMZN and NFLX.Compared with RF, PRF, KNN and FKNN, GHM-FKNN achieved the best performance with accuracy of 62.37% for AAPL, 58.25% for AMZN, and 64.10% for NFLX.

Grey GM(1,1) Model with Function-Transfer Method for Wear Trend Prediction and its Application

Trend forecasting is an important aspect in fault diagnosis and work state supervision. The principle, where Grey theory is applied in fault forecasting, is that the forecast system is considered as a Grey system; the existing known information is used to infer the unknown information's character, state and development trend in a fault pattern, and to make possible forecasting and decisions for future development. It involves the whitenization of a Grey process. But the traditional equal time interval Grey GM (1,1) model requires equal interval data and needs to bring about accumulating addition generation and reversion calculations. Its calculation is very complex. However, the non equal interval Grey GM (1,1) model decreases the condition of the primitive data when establishing a model, but its requirement is still higher and the data were pre processed. The abrasion primitive data of plant could not always satisfy these modeling requirements. Therefore, it establishes a division method suited for general data modeling and estimating parameters of GM (1,1), the standard error coefficient that was applied to judge accuracy height of the model was put forward; further, the function transform to forecast plant abrasion trend and assess GM (1,1) parameter was established. These two models need not pre process the primitive data. It is not only suited for equal interval data modeling, but also for non equal interval data modeling. Its calculation is simple and convenient to use. The oil spectrum analysis acted as an example. The two GM (1,1) models put forward in this paper and the new information model and its comprehensive usage were investigated. The example shows that the two models are simple and practical, and worth expanding and applying in plant fault diagnosis.

STUDY ON TREND PREDICTION AND VARIATION ONTHE FLOW INTO THE LONGYANGXIA RESERVOIR

The Longyangxia Gorge Key Water Control System is the first of the stairstep power sations along the Longyangxi-a-Qingtongxia river section. It has been playing an very important role in providing power, protecting flood and ice run supplying and irrigation etc. in the northwestern China. Therefore, the study on trend prediction, variation on the flow into the Longyangxia Reservoir are of the great social and economic benefits. In the medium-and-long-range runoff forecast, all kinds of regression equation are often used for predicting future hydrologic regime. However, these regression models aren’t appropriate to super long -range runoff forecast because of the restricting on weather data and so on. So a new super long-range runoff forecast model don’t depend on Reai-time weather data and called “Period correcting for residual error series GM (1, 1) model” is presented based on analyzing for the relational hydrologic data and the variation on the flow into the Longyangxia Reservoir, and the forecast model was applied successfully to predict the recent and super long -term trends of the flow into the Longyangxia Reservoir. The results indicate that the annual flow into the Longyangxia Reservoir is in the ending minimum period of the runoff history. The runoff increasing is expected in for the coming years.

Predicting Bitcoin Trends Through Machine Learning Using Sentiment Analysis with Technical Indicators

Predicting Bitcoin price trends is necessary because they represent the overall trend of the cryptocurrency market.As the history of the Bitcoin market is short and price volatility is high,studies have been conducted on the factors affecting changes in Bitcoin prices.Experiments have been conducted to predict Bitcoin prices using Twitter content.However,the amount of data was limited,and prices were predicted for only a short period(less than two years).In this study,data from Reddit and LexisNexis,covering a period of more than four years,were collected.These data were utilized to estimate and compare the performance of the six machine learning techniques by adding technical and sentiment indicators to the price data along with the volume of posts.An accuracy of 90.57%and an area under the receiver operating characteristic curve value(AUC)of 97.48%were obtained using the extreme gradient boosting(XGBoost).It was shown that the use of both sentiment index using valence aware dictionary and sentiment reasoner(VADER)and 11 technical indicators utilizing moving average,relative strength index(RSI),stochastic oscillators in predicting Bitcoin price trends can produce significant results.Thus,the input features used in the paper can be applied on Bitcoin price prediction.Furthermore,this approach allows investors to make better decisions regarding Bitcoin-related investments.

Feature Selection, Deep Neural Network and Trend Prediction

The literature generally agrees that longer-horizon(over a month) predictions make more sense than short-horizon ones. However, it's an especially challenging task due to the lack of data(in unit of long horizon)and economic data have a low S/N ratio. We hypothesize that the stock trend is largely dictated by driving factors which are filtered by psychological factors and work on behavioral factors: representative indicators from these three aspects would be adequate in trend prediction. We then extend the Stepwise Regression Analysis(SRA)algorithm to constrained SRA(c SRA) to carry out a further feature selection and lag optimization. During modeling stage, we introduce the Deep Neural Network(DNN) model in stock prediction under the suspicion that economic interactions are too complex for shallow networks to capture. Our experiments indeed show that deep structures generally perform better than shallow ones. Instead of comparing to a kitchen sink model, where over-fitting can easily happen with a shortage of data, we turn around and use a model ensemble approach which indirectly demonstrates our proposed method is adequate.

Adaptive Wavelets Based on Second Generation Wavelet Transform and Their Applications to Trend Analysis and Prediction

In order to make trend analysis and prediction to acquisition data in amechanical equipment condition monitoring system, a new method of trend feature extraction andprediction of acquisition data is proposed which constructs an adaptive wavelet on the acquisitiondata by means of second generation wavelet transform ( SGWT), Firstly, taking the vanishing momentnumber of the predictor as a constraint, the linear predictor and updater are designed according tothe acquisition data by using symmetrical interpolating scheme. Then the trend of the data isobtained through doing SGWT decomposition , threshold processing and SGWT reconstruction. Secondly,under the constraint of the vanishing moment number of the predictor, another predictor based on theacquisition data is devised to predict the future trend of the data using a non-symmetricalinterpolating scheme, A one-step prediction algorithm is presented to predict the future evolutiontrend with historical data. The proposed method obtained a desirable effect in peak-to-peak valuetrend analysis for a machine set in an oil refinery.

Intelligent HEV Fuzzy Logic Control Strategy Based on Identification and Prediction of Drive Cycle and Driving Trend

Real-time drive cycles and driving trends have a vital impact on fuel consumption and emissions in a vehicle. To address this issue, an original and alternative approach which incorporates the knowledge about real-time drive cycles and driving trends into fuzzy logic control strategy was proposed. A machine learning framework called MC_FRAME was established, which includes two neural networks for self-learning and making predictions. An intelligent fuzzy logic control strategy based on the MC_FRAME was then developed in a hybrid electric vehicle system, which is called FLCS_MODEL. Simulations were conducted to evaluate the FLCS_MODEL using ADVISOR. The simulation results indicated that comparing with the default controller on the drive cycle NEDC, the FLCS_MODEL saves 12.25% fuel per hundred kilometers, with the HC emissions increasing by 22.7%, the CO emissions reducing by 16.5%, the NOx emissions reducing by 37.5% and with the PM emissions reducing by 12.9%. A conclusion can be drawn that the proposed approach realizes fewer fuel consumption and less emissions.

未来30年中国耕地和高标准农田分布的省级预测

科学预测中国未来耕地和高标准农田分布对于保障我国粮食安全和提高耕地质量具有重要意义。该研究假设未来各省份耕地和高标准农田占全国总面积比例不变,根据各省份耕地和高标准农田实际面积、未来全国耕地及各省份高标准农田规划面积等数据,基于最小二乘法的二次多项式拟合预测未来30 a全国及各省份耕地面积和高标准农田面积,进而预测未来各省份耕地面积保有率和高标准农田占耕地比例。结果表明:(1)到2050年,我国耕地面积稳定在1.20×10^(8)hm^(2),耕地面积保有率稳定在100%,高标准农田面积达到1.03×10^(8)hm^(2),高标准农田占耕地比例从2020年的43.59%增长到2050年的85.89%,增长近一倍。(2)七大区域中,黄淮海区、长江中下游区、东南区的高标准农田占耕地比例较高,在2020年均已达到70%以上,东北区、西北区、青藏区、西南区高标准农田占耕地比例则处于40%~60%之间。到2050年,受耕地比例稳定中略有下降的趋势影响,黄淮海区、长江中下游区、东南区、青藏区高标准农田占耕地比例均达90%以上,东北区、西北区、西南区则处于70%~90%之间。(3)从各省份来看,江西省、浙江省、福建省、广东省等南方省份的高标准农田建设比例较高,而内蒙古自治区、黑龙江省、辽宁省、吉林省、山东省等北方省份的耕地面积保有率较高。研究得到的耕地面积及其保有率、高标准农田面积及其占耕地比例等指标的分省份预测值可以为未来相关部门制定国土空间规划等提供科学依据,也可以为未来耕地布局和农田利用等研究提供分省份的总量参考。

基于参数自适应SVR和VMD-TCN的水电机组劣化趋势预测

针对水电机组难以利用实时监测数据对机组劣化状态进行有效评估,以及水电机组不同运行工况对运行状态指标趋势预测模型参数影响显著的问题,提出一种基于参数自适应支持向量回归机(SVR)、变分模态分解(VMD)和时间卷积网络(TCN)的水电机组劣化趋势预测方法;首先按照功率和水头将机组运行工况细化为若干典型工况,在此基础上采用改进天鹰算法建立SVR模型,对各个工况下的预测参数进行寻优,建立起工况与最优参数的数据;再通过神经网络对工况和最优预测参数进行拟合,构建出映射两者复杂关系的非线性函数,然后将构建出的映射关系加入到传统的SVR中,实现适应于水电机组工况变化的自适应SVR健康模型;其次,根据健康模型输出的标准值和监测数据,计算出劣化趋势序列;最后,考虑到劣化趋势序列的非线性因素,建立了一个基于VMD-TCN的时间序列预测模型,以实现对劣化趋势的准确预测。并设计多组对比实验,验证所提出模型的精度更高,时间更快。

基于数实结合的测控装备健康监测系统设计与实现

针对测控业务领域装备种类多、型号杂、布站分散、技术保障人力不足、远程技术协助手段落后等实际问题,结合测控装备自身特点和运维保障规律,提出了采用数字技术构建功能级数字测控装备,并利用数字装备和实体装备间物联网络使数实装备状态数据一致,实现基于数实结合的测控装备健康监测,实现装备状态数实同步,并可利用数实同步有效的评估装备状态、进行故障诊断和健康趋势预估,实现了运维保障数字化。

某三甲中医医院ICU感染发生率时间序列分析及趋势预测

目的:了解某三甲中医医院ICU感染发生率的时序分布特征,预测其发生规律和趋势,为中医医院ICU感染监测提供数据支持。方法:收集某三甲中医医院2019年1月至2024年2月ICU医院感染数据。利用求和自回归滑动平均模型(Autoregressive integrated moving average,ARIMA)对ICU感染发生趋势进行预测并评价其预测效果。结果:2019年1月至2024年2月某三甲中医医院ICU医院感染发生率为2.61%(232/8895);时间序列分析显示,ICU医院感染发生率波动较大且存在一定周期性,总体呈下降趋势。根据赤池信息准则和贝叶斯信息准则拟合,ARIMA(0,1,1)为最优预测模型。经参数估计与效果评价,感染发生率实际值均在预测值95%可信区间内,模型预测效果较好。结论:运用ARIMA对某三甲中医医院ICU医院感染发生率的预测结果良好,可显示其长期发生规律与趋势,能为医院感染监测提供科学依据。

Dynamics modeling and optimal control for multi-information diffusion in Social Internet of Things

As an ingenious convergence between the Internet of Things and social networks,the Social Internet of Things(SIoT)can provide effective and intelligent information services and has become one of the main platforms for people to spread and share information.Nevertheless,SIoT is characterized by high openness and autonomy,multiple kinds of information can spread rapidly,freely and cooperatively in SIoT,which makes it challenging to accurately reveal the characteristics of the information diffusion process and effectively control its diffusion.To this end,with the aim of exploring multi-information cooperative diffusion processes in SIoT,we first develop a dynamics model for multi-information cooperative diffusion based on the system dynamics theory in this paper.Subsequently,the characteristics and laws of the dynamical evolution process of multi-information cooperative diffusion are theoretically investigated,and the diffusion trend is predicted.On this basis,to further control the multi-information cooperative diffusion process efficiently,we propose two control strategies for information diffusion with control objectives,develop an optimal control system for the multi-information cooperative diffusion process,and propose the corresponding optimal control method.The optimal solution distribution of the control strategy satisfying the control system constraints and the control budget constraints is solved using the optimal control theory.Finally,extensive simulation experiments based on real dataset from Twitter validate the correctness and effectiveness of the proposed model,strategy and method.

东北三省碳排放影响因素分析和趋势预测——基于STIRPAT模型和情景分析法

东北三省是“双碳”目标下碳减排工作的重点区域。利用STIRPAT(可拓展的随机性环境影响评估)模型对2000—2021年东北三省各省的碳排放影响因素进行分析,并利用情景分析法预测截止到2040年以前、均衡情景下东北三省各省的碳排放趋势。模型分析结果表明:各省只存在导致碳排放量增加的影响因素;能源消费总量、碳排放强度、人均GDP是各省的共同影响因素;各省的影响因素是不同的组合;各省影响因素的促进作用存在省份差异性。预测结果表明:碳排放量同期数值由高到低依次为辽宁省、黑龙江省、吉林省;各省的碳排放量曲线都呈现出倒“U”形并能够看出明显的峰值;黑龙江省和吉林省碳达峰的时间都为2012年,而辽宁省碳达峰的时间为2025年。针对分析和预测结果,提出能源、社会经济、区域协调3个方面的建议。

新世纪以来我国铀矿地质科技创新重要进展及展望

新世纪以来,我国铀矿地质勘查领域重大理论和技术创新成果,极大地改变了我国砂岩型、热液型铀矿的找矿思路、找矿方向,引领、支撑了系列找矿重大突破,重塑了铀矿勘查新格局。文章首先分析了我国20余年来铀矿地质科研布局的总体思路与演进方向;然后总结了铀矿重大基础地质与成矿理论研究进展,系统凝练了地质、物化探、遥感、钻探工艺、分析测试等领域的技术创新及核心勘查装备研制成果,梳理了铀资源预测评价技术研究进展及应用成效;最后,在重大基础前沿和成矿理论、先进铀矿勘查技术研发、非常规核能裂变资源勘查技术研究、数字铀矿勘查技术开发等方面提出了未来10~15年主要发展趋势和方向,指出在“双碳”目标和核能大发展对铀资源保障需求背景下,必须继续坚持并加强铀矿地质科技创新,高质量引领和推动“新区、新层位、新类型、新深度”找矿突破。

基于时间序列分析的潜在学科交叉前沿主题识别研究

[目的/意义]识别学科交叉前沿主题并预测其发展趋势,有助于了解学科内部结构,挖掘领域重点部署方向,为未来创新性、突破性研究提供参考。[方法/过程]以美国国家自然科学基金项目及其产出论文分别作为前端、后端数据,首先,从三个维度测度项目学科交叉度,遴选领域学科交叉项目;其次,从主题关注度、新颖度等方面构建研究前沿主题识别指标体系,对学科交叉主题进行二次遴选,满足阈值的即为学科交叉前沿主题;再次,对比时间序列分析模型ARIMA和LSTM主题拟合效果并选择误差最小模型对学科交叉前沿主题进行趋势预测分析;最后,以生物科学领域为例对方法的有效性和可行性进行实例验证。[结果/结论]生物科学领域在纳米生物学技术、全球变化和海洋环境生物学、生物信息学及壶菌病与两栖动物多样性方面有较好发展前景。经专家咨询和已有研究对比分析,该方法可有效识别领域学科交叉前沿主题,并对其未来研究趋势走向有一定参考借鉴。

震后趋势判定技术系统CAAFs震后余震预测效果评价

自2019年正式运行以来,震后趋势判定技术系统CAAFs为震后应急与趋势研判及时提供数据和技术支撑。本文梳理总结CAAFs系统持续4.5年产出的数据,从全国和不同地区两个方面评价系统产出的震后余震预测结果,得到以下三点认识:①震级上限预测(发生某震级以上地震的可能性不大)的正确率大多为90%及以上,显著优于震级区间预测(存在发生同等大小地震的可能或存在发生某震级左右地震的可能),各分区预测情况差异不大;②震级区间预测震级普遍比地震实况偏高,实际7日内发生的最大余震震级与预测震级的差值(震级偏差)落入[-0.5,0.5]区间的地震比例约为44%,落入[-1,1]区间的地震比例约为69%,越靠近[-0.5,0.5]区间,地震的比例越高;③震级区间预测中,各分区有一定差异,西南地区优于西北、华南和华北东北三个地区,总的来看4~5级地震预测情况不如其他震级区间,可能与该震级区间内地震的最大余震震级变化范围较大有关。

考虑特征重组与改进Transformer的风电功率短期日前预测方法

短期日前风电功率预测对电力系统调度计划制定有重要意义,该文为提高风电功率预测的准确性,提出了一种基于Transformer的预测模型Powerformer。模型通过因果注意力机制挖掘序列的时序依赖;通过去平稳化模块优化因果注意力以提高数据本身的可预测性;通过设计趋势增强和周期增强模块提高模型的预测能力;通过改进解码器的多头注意力层,使模型提取周期特征和趋势特征。该文首先对风电数据进行预处理,采用完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)将风电数据序列分解为不同频率的本征模态函数并计算其样本熵,使得风电功率序列重组为周期序列和趋势序列,然后将序列输入到Powerformer模型,实现对风电功率短期日前准确预测。结果表明,虽然训练时间长于已有预测模型,但Poweformer模型预测精度得到提升;同时,消融实验结果验证了模型各模块的必要性和有效性,具有一定的应用价值。

高质量发展背景下中国马铃薯产业新特点与发展趋势

高质量发展是适应新时代中国社会主要矛盾的必然要求,是经济社会发展的硬道理。在此必然要求和硬道理下,中国马铃薯产业在转型升级过程中呈现生产规模出现分化、发展方式转向内涵型增长、“北薯南移”区域布局持续演化、市场供应量季节波动逐渐弱化、价格波动程度越来越大、国际贸易优势日益凸显、直接食用向加工转化等新特点,资源环境约束趋紧、科技贡献趋缓、生产成本持续上涨、马铃薯消费增长乏力等问题的存在对马铃薯高质量发展形成不同程度的制约,预判中短期内中国马铃薯产业发展将呈现出规模基本稳定、市场波动季节波动性进一步弱化、产业发展加工导向逐渐凸显等趋势。推进马铃薯产业高质量发展,需通过加大马铃薯科技研发投入促进产业创新发展,发展加工践行大食物观促进产业协调发展,推广环境适应性生产模式促进产业绿色发展,强化行业管理部门引导促进产业稳定发展。

基于运行效率分布差异的水电机组劣化状态趋势预测

为解决水电机组劣化状态难以刻画及预测精度低的问题,需深入探究不同机组状态下运行效率的分布差异特性,提出了一种基于运行效率分布差异的水电机组劣化状态趋势预测方法。首先,综合考虑水电机组工况(水头、流量)与效率之间映射关系和状态监测数据随机性,利用高斯混合模型拟合多工况下机组健康状态运行效率的概率分布特性;在此基础上,计算观测样本在机组健康状态分布下的负对数似然概率,并以此作为水电机组劣化状态指标,表征观测样本与机组健康状态标准分布之间的偏差;进一步采用非因果原理和高斯误差线性单元,分别改进时间卷积网络(TCN)的膨胀卷积模块和残差模块,并融合门控循环单元(GRU),设计并构建水电机组劣化状态预测模型;最后,利用某水电站#6机组实际运行监测数据开展方法验证。结果表明,所提方法能有效提升机组劣化状态趋势预测精度。