基于SVM-STL-LSTM的区域短期电力负荷预测研究

针对区域电力负荷的时间序列数据随机性强、预测精度低及单一模型的数据特征提取能力差等问题,提出了一种支持向量机(SVM)、STL时序分解法、长短期记忆神经网络(LSTM)组合的电力负荷预测模型。该模型利用SVM对时间序列的电力负荷数据进行初始预测,并通过STL时序分解法对残差序列进行时序分解,从而提高残差序列的稳定性,减小其随机性,最后用LSTM对SVM的预测误差进行修正。试验结果证明,该方法利用误差修正可有效处理随机性强的数据,有利于预测结果的稳定性,提高预测精度。

基于STL-Holt-WOA-RF的拱坝位移预测模型

传统的混凝土拱坝位移预测模型主要关注水压、温度、时效等因素与拱坝位移之间的关系,未对拱坝位移数据中所包含的信息进行充分挖掘。为此,采用Seasonal and Trend decomposition using Loess算法(STL)将拱坝位移原始数据分解为趋势序列、周期序列及残差分量。在此基础上,采用鲸鱼优化算法(WOA)结合随机森林算法(RF)对三个分量进行预测,并使用Holt-Winters算法充分考虑趋势序列中的趋势信息对趋势序列的预测结果进行修正。最后将修正后的趋势序列预测结果和周期序列、残差分量预测结果相加,得出拱坝位移最终预测结果。工程实例表明,基于STL-Holt-WOA-RF的拱坝位移预测模型能够显著提高预测的准确性和稳定性,为拱坝位移预测提供了新的思路和方法。

一种基于STL-Prophet-Informer模型的太阳电池阵多变量趋势预测方法

为了提高太阳电池阵多变量预测的精度,解决阳电池阵遥测参数存在周期波动与增长性互相耦合的问题,提出一种基于STL-Prophet-Informer模型的太阳电池阵多变量预测算法.该算法首先应用局部加权周期趋势分解算法(seasonal and trend decomposition procedure based on loess,STL)对太阳电池阵的多个参数分解为趋势分量、周期分量和残差分量,然后采用对趋势性数据预测效果较好的Prophet预测趋势分量,Informer模型预测周期分量和残差分量,最后将各分量预测结果相加后得到总的太阳电池阵参数预测值.以某卫星太阳电池阵实际遥测数据做算例分析,提出算法的各项误差评价指标和单一的Informer模型、LSTM模型等相比有明显减小,将该组合预测模型用于太阳电池阵多变量参数预测中,可以提高参数预测精度,提升卫星自主运行性能.

Optimized nitrogen application for maximizing yield and minimizing nitrogen loss in film mulching spring maize production on the Loess Plateau,China

Excessive use of N fertilizers(driven by high-yield goals)and its consequent environmental problems are becoming increasingly acute in agricultural systems.A 2-year field experiment was conducted to investigate the effects of three N application methods(application of solid granular urea once(OF)or twice(TF),application of solid granular urea mixed with controlled-release urea once(MF),and six N rates(0,60,120,180,240,and 300 kg N ha^(-1))on maize yield,economic benefits,N use efficiency,and soil N balance in the maize(Zea mays L.)film mulching system on the Loess Plateau,China.The grain yield and economic return of maize were significantly affected by the N rate and application method.Compared with the OF treatment,the MF treatment not only increased the maize yield(increased by 9.0-16.7%)but also improved the economic return(increased by 10.9-25.8%).The agronomic N use efficiency(NAE),N partial factor productivity(NPFP)and recovery N efficiency(NRE)were significantly improved by 19.3-66.7,9.0-16.7 and 40.2-71.5%,respectively,compared with the OF treatment.The economic optimal N rate(EONR)of the OF,TF,and MF was 145.6,147.2,and 144.9 kg ha^(-1) in 2019,and 206.4,186.4,and 146.0 kg ha^(-1) in 2020,respectively.The apparent soil N loss at EONR of the OF,TF,and MF were 97.1-100.5,78.5-79.3,and 50.5-68.1 kg ha^(-1),respectively.These results support MF as a one-time N application method for delivering high yields and economic benefits,with low N input requirements within film mulching spring maize system on the Loess Plateau.

Coupling Coordination Analysis of Cropland Intensification and Agroecosystem Services:Evidence from Loess Plateau in Shaanxi Province,China

One of the greatest challenges in the agroecosystem is to improve cropland intensification while preserving agroecosystem services.While many studies have investigated the effect of cropland intensification on agroecosystem service,the interactive coupling and coordination among these factors remain largely unexplored.In view of this,this study performed a case study of the Loess Plateau in Shaanxi Province,China and constructed comprehensive evaluation models to quantify the cropland intensification and agroecosystem service in this area.Balance analysis and the coupling coordination degree model were used to evaluate the interactive relationship between cropland intensification and agroecosystem service,and statistical analysis and spatial autocorrelation were used to analyze the spatial characteristics and potential mechanism of the coupling coordination.Results show that both the cropland intensification and agroecosystem service in the study area were relatively low yet gradually increased from 2000 to 2020.Agroecosystem service lag was identified as the dominant unbalanced development type.Improving the supply capacity of agroecosystem services plays a key role in the balanced development of cropland in the Loess Plateau.The coupling coordination degree between cropland intensification and agroecosystem service ranges from basic coordination to serious incoordination.Therefore,cropland intensification practices in the area should be optimized to enhance this coordination degree.An upward trend was also observed in the coupling coordination degree from2000 to 2020.The withdrawal of marginal cropland in the Grain for Green program is one of the most important reasons for this trend,especially for the northern region.Around 83.6%of the high-high clusters are concentrated in the southern region of the Loess Plateau,whereas 70.5%of the low-low clusters are distributed in the northern region.These clustering characteristics are mainly attributed to the environmental suitability of these areas for agriculture and their degree of economic development.

Spatiotemporal variations in ecosystem services and their trade-offs and synergies against the background of the gully control and land consolidation project on the Loess Plateau,China

Studying the spatiotemporal variations in ecosystem services and their interrelationships on the Loess Plateau against the background of the gully control and land consolidation(GCLC)project has significant implications for ecological protection and quality development of the Yellow River Basin.Therefore,in this study,we took Yan'an City,Shaanxi Province of China,as the study area,selected four typical ecosystem services,including soil conservation service,water yield service,carbon storage service,and habitat quality service,and quantitatively evaluated the spatiotemporal variation characteristics and trade-offs and synergies of ecosystem services from 2010 to 2018 using the Integrated Valuation of Ecosystem Services and Trade-offs(InVEST)model.We also analysed the relationship between the GCLC project and regional ecosystem service changes in various regions(including 1 city,2 districts,and 10 counties)of Yan'an City and proposed a coordinated development strategy between the GCLC project and the ecological environment.The results showed that,from 2010 to 2018,soil conservation service decreased by 7.76%,while the other three ecosystem services changed relatively little,with water yield service increasing by 0.56% and carbon storage service and habitat quality service decreasing by 0.16% and 0.14%,respectively.The ecological environment of Yan'an City developed in a balanced way between 2010 and 2018,and the four ecosystem services showed synergistic relationships,among which the synergistic relationships between soil conservation service and water yield service and between carbon storage service and habitat quality service were significant.The GCLC project had a negative impact on the ecosystem services of Yan'an City,and the impact on carbon storage service was more significant.This study provides a theoretical basis for the scientific evaluation of the ecological benefits of the GCLC project and the realization of a win-win situation between food security and ecological security.

Unraveling the hydraulic properties of loess for landslide prediction:A study on variations in loess landslides in Lanzhou,Dingxi,and Tianshui,China

Loess has distinctive characteristics,leading to frequent landslide disasters and posing serious threats to the lives and properties of local re sidents.The involvement of water repre sents a critical factor in inducing loess landslides.This study focuses on three neighboring cities sequentially situated on the Loess Plateau along the direction of aeolian deposition of loess,namely Lanzhou,Dingxi,and Tianshui,which are densely populated and prone to landslide disasters.The variations in hydraulic properties,including water retention capacity and permeability,are investigated through Soil Water Characteristic Curve(SWCC)test and hydraulic conductivity test.The experimental findings revealed that Tianshui loess exhibited the highest water retention capacity,followed by Dingxi loess,while Lanzhou loess demonstrated the lowest water retention capacity.Contrastingly,the results for the saturated permeability coefficient were found to be the opposite:Tianshui loess showed the lowest permeability,whereas Lanzhou loess displayed the highest permeability.These results are supported and analyzed by scanning electron microscopy(SEM)observation.In addition,the water retention capacity is mathematically expressed using the van Genuchten model and extended to predict unsaturated hydraulic properties of loess.The experimental results exhibit a strong accordance with one another and align with the regional distribution patterns of disasters.

保留STL模型特征细节的3D打印自适应分层算法

针对传统的3D打印自适应分层算法存在计算过程复杂、无法消除阶梯效应影响、不能保留模型特征以及特征容易丢失等问题,提出了一种改进的保留STL模型特征细节的自适应分层算法。该算法首先判断模型当前层是否含有显著特征边;然后针对当前层含有显著特征边的情况,通过相邻2个三角网格法矢量的夹角来适应性确定层厚,对不含显著特征边的情况,利用三角网格法矢量与成型方向的夹角来自适应确定层厚,对于模型存在特征偏移的情况,则通过相邻切平面封闭曲线个数变化来适应性地确定层厚;最后将上述3种情况下自适应层厚的最小值作为当前分层的依据。研究结果表明:对于所采用的恐龙模型,所提算法不仅很好地保留了模型的细节特征,而且显著提升了模型成型质量;与基于三角面片法的矢量自适应分层算法和仅保留模型特征的自适应分层算法相比,所提算法的模型精度分别提升了28.6%和31.7%。该算法减小了阶梯效应以及模型特征丢失和偏移对打印质量的影响,提高了模型的打印精度。

Soil erosion and soil properties in reclaimed forestland of loess hilly region

Based on data observed from 1989 to 1998 in the Ziwuling survey station, changes of soil erosion and soil physico-mechanical properties were studied after forestland reclamation. When the man-induced factors changed the eco-environment by reclaiming forestlands, the intensity of man-made soil erosion in reclaimed lands was 1,000 times more than that of natural erosion in forestlands. From the analysis of soil physical and mechanical properties, the clay content and physical clay content decreased 2.74% and 3.01% respectively, and the >0.25 mm water stable aggregate content decreased 58.7%, the soil unit weight increased and the soil shear strength decreased, all of which were easier to cause soil erosion. The results of the correlation analysis showed that the >0.25 mm water stable aggregate content was the greatest influencing factor on soil erosion, the partial correlated coefficient was 0.9728, and then were soil coarse grain and soil shear strength, the partial correlated coefficients being 0.8879 and 0.6020 respectively. The relationships between the >0.25 mm water stable aggregate content, the soil sheer strength and the soil erosion intensity were analyzed, which showed that the first and seventh years were the turning years of the soil erosion intensity after the forestland reclamation. The degenerative eroded soil and eco-environment formed the peculiar erosion environment, which aggravated the soil erosion rapidly.

基于STL-LSTM-TCN模型的短期负荷预测方法

为提高模型的预测精度,引入STL算法将负荷序列分解为周期分量、趋势分量、残差分量,利用各分量训练LSTM和TCN模型。在得到LSTM和TCN模型的预测后,为进一步提高模型的预测精度,构建了集成模型对LSTM模型和TCN模型的预测值进行集成。以西班牙电力负荷数据集为例,对所提负荷预测方法进行了验证。实验结果表明,STL算法和集成模型的引入均提高了模型的预测精度,基于STL-LSTM-TCN的预测方法相较于LSTM、TCN、STL-LSTM、STL-TCN,其MAPE分别降低了2.8664%、2.1229%、0.37%、0.1%,所提负荷预测方法的预测误差最低,验证了所提预测方法的有效性与合理性。

基于局部简易拓扑结构的STL模型分层算法

为进一步提高3D打印切片效率,该文提出一种基于局部简易拓扑结构的STL模型分层算法。读取STL文件信息,确定模型坐标高度范围;依据高度信息和层切厚度确定每个切片层的高度;对三角形面片按高度进行排序。设计构造一种新型的简易半边数据结构,利用该数据结构对排好序的三角形在每个切片层分组建立并动态维护局部拓扑结构,通过拓扑结构快速得到切片轮廓。此方法不需要建立全局拓扑结构,并缩短了建立分组拓扑结构的时间,实验证明提高了切片的速度和效率。

Review on the fully mulched ridge–furrow system for sustainable maize production on the semi-arid Loess Plateau

The fully mulched ridge–furrow(FMRF) system has been widely used on the semi-arid Loess Plateau of China due to its high maize(Zea mays L.) productivity and rainfall use efficiency. However, high outputs under this system led to a depletion of soil moisture and soil nutrients, which reduces its sustainability in the long run. Therefore, it is necessary to optimize the system for the sustainable development of agriculture. The development, yield-increasing mechanisms,negative impacts, optimization, and their relations in the FMRF system are reviewed in this paper. We suggest using grain and forage maize varieties instead of regular maize;mulching plastic film in autumn or leaving the mulch after maize harvesting until the next spring, and then removing the old film and mulching new film;combining reduced/notillage with straw return;utilizing crop rotation or intercropping with winter canola(Brassica campestris L.), millet(Setaria italica), or oilseed flax(Linum usitatissimum L.);reducing nitrogen fertilizer and partially replacing chemical fertilizer with organic fertilizer;using biodegradable or weather-resistant film;and implementing mechanized production. These integrations help to establish an environmentally friendly, high quality, and sustainable agricultural system, promote highquality development of dryland farming, and create new opportunities for agricultural development in the semi-arid Loess Plateau.

Soil erosion and its causes in high-filling body:A case study of a valley area on the Loess Plateau,China

Large-scale land consolidation projects(LCPs)have been carried out on the Loess Plateau to increase the area of agriculture land.The newly created land is prone to soil erosion under the effects of water and gravity.Taking a typical high-filling body(HFB)formed by LCPs in Yan’an,China as the subject,this study comprehensively investigated the types and causes of soil erosion with multiple methods of field investigation,on-site monitoring and laboratory tests.Results showed that the HFB presented a composite pattern of soil erosion with multiple types mainly including underground erosion,mixed water-gravity erosion,seepage erosion,and scouring erosion.The type of erosion varied spatially in different parts of the HFB depending on the dominant factors,mainly including the groundwater state,rainfall,runoff,gravity action,topography,and soil erodibility.The underground erosion mainly occurred at the positions with higher groundwater level and larger hydraulic gradient,while scouring erosion mainly occurred at the positions with extensive interactions of surface runoff,channel slope gradient and soil properties.And near the leading edge of the top of the slope,a band of mixed watergravity erosion occurred owing to the effects of water and gravity.In addition,nearly saturated soils at the toe of HFB displayed groundwater exfiltration and slope-face slumping.Based on our findings on the causes and variation of soil erosion for the HFB,we proposed the following erosion prevention and control measures to protect the LCPs on the Loess Plateau:to construct drainage ditches and blind ditches to form a complete drainage system,plant alfalfa on the top platform to increase rainfall interception and reduce surface runoff,set seepage ditches and plant deep-rooted plants at the toe of the slope to improve slope toe stability,monitor groundwater level and slope deformation to learn the erosion dynamics and slope stability,and optimize the geometry of HFB such as the slope gradient and slope steps to reduce soil erosion.

Intelligent identification of landslides in loess areas based on the improved YOLO algorithm:a case study of loess landslides in Baoji City

Loess landslides are one of the geological hazards prevalent in mountainous areas of Loess Plateau,seriously threatening people's lives and property safety.Accurate identification of landslides is a prerequisite for reducing the risk of landslide hazards.Traditional landslide interpretation methods often have the disadvantage of being laborious and difficult to use on a large scale compared with the recently developed deep learning-based landslide detection methods.In this study,we propose an improved deep learning model,landslide detectionyou only look once(LD-YOLO),based on the existing you only look once(YOLO)model for the intelligent identification of old and new landslides in loess areas.Specifically,remote sensing images of landslides in Baoji City,Shaanxi Province,China are acquired from the Google Earth Engine platform.The landslide images of Baoji City(excluding Qianyang County)are used to establish a loess landslide dataset for training the model.The landslide data of Qianyang County is used to verify the detection performance of the model.The focal and efficient IoU(Focal-EIoU)loss function and efficient channel attention(ECA)mechanism are incorporated into the 7th version of YOLO(YOLOv7)model to construct the LD-YOLO model,which makes it more suitable for the landslide detection task.The experiments yielded an improved LD-YOLO model with average precision of 92.05%,precision of 92.31%,recall of 90.28%,and F1-score of 91.28%for loess landslide detection.The landslides in Qianyang County were divided into two test sets,new landslides and old landslides,which were used to test the detection performance of LD-YOLO for both types of landslides.The results show that LD-YOLO detects old landslides with a detection precision of 82.75%and a recall of 80%.When detecting new landslides,the detection precision is 94.29%and the recall is 91.67%.It indicates that our proposed LD-YOLO model has strong detection performance for both new and old landslides in loess areas.Through a proposed solution that can realize the accurate detection of landslides in loess areas,this paper provides a valuable reference for the application of deep learning methods in landslide identification.

ECO-ENVIRONMENT CHANGE AND SOIL EROSION PROCESS IN THE RECLAIMED FORESTLAND OF THE LOESS PLATEAU

Serious soil erosion has made the eco-environment fragile in the Loess Plateau. Based on the 10-year data observed from 1989 to 1998 in the Ziwuling Survey Station in loess hilly region, the eco-environment change and soil erosion process in reclaimed forestland were studied in this paper. The results showed that the intensity of man-made soil erosion caused by forestland reclamation was 1000 times more than that of the natural erosion. From the analysis of soil physical and mechanical properties, in the 10th year after forestland was reclaimed, the clay content and physical clay content decreased 2.74 percentage point and 3.01 percentage point respectively, the >0.25mm waterstable aggregate content decreased 31.59 percentage point, the soil bulk density increased and soil shear strength decreased, all of which were easier to cause soil erosion. The correlation analysis showed that >0.25mm waterstable aggregate content was the key factor affecting soil erosion, and the secondary factors were soil coarse grain and soil shear strength. The relation between the >0.25mm waterstable aggregate content, the soil sheer strength and the soil erosion intensity were analyzed, which showed that the first year and the seventh erosion year were the turn years of the soil erosion intensity after the forestland was reclaimed, revealed that the change of eco-environment was the main cause to accelerate soil erosion, and the worse environment caused soil erosion to be serious rapidly.

Regulation of vegetation pattern on the hydrodynamic processes of erosion on hillslope in Loess Plateau,China

As vegetation are closely related to soil erosion,hydrodynamic parameter changes under various vegetation pattern conditions can be used as an important basis for the research of the soil erosion mechanism.Through upstream water inflow experiments conducted on a loess hillslope,how the vegetation pattern influences the hydrodynamic processes of sediment transport was analyzed.The results show that the placement of a grass strip on the lower upslope can effectively reduce runoff erosion by 69%,relying on the efficiency of regulated hydrodynamic process.The effective location of grass strip for hillslope alleviating erosion is on the lower part of the upslope,mainly due to the grass strip measure used to regulate the hydrodynamic system.As a result,the underlying surface runoff resistance is increased by 5 times,runoff shear stress is decreased by more than 90%,and runoff power decreased by over 92%.The measure greatly separates the scouring energy of surface runoff that acts on the slope soil.Therefore,the use of grass strips effectively decreases the energy of runoff flowing along the slope,eliminating soil erosion to a great extent and thereby achieving a better regulation of hydrodynamic processe.

基于LOESS和ARIMA混合模型的PRS系统换热能力预测

探讨差分自回归移动平均模型(ARIMA)与局部加权回归模型(LOESS)的混合模型在核电厂蒸汽发生器二次侧非能动冷却系统(PRS)换热能力验证试验的应用,建立模型并进行参数估计和对换热功率预测。工程实践中时间序列往往具有线性和非线性双重特征,单一模型具有一定局限性,对此提出了混合模型进行时间序列预测。利用平稳自然循环阶段数据,采用SPSS软件建立ARIMA(1,1,0)模型,得到换热功率的线性部分;运用R软件建立LOESS模型,对偏差序列进行预测,得到换热功率的非线性部分;最后建立LOESS和ARIMA混合模型,利用混合模型对换热功率进行预测,并根据实测数据对预测结果进行对比验证。实验结果表明,LOESS和ARIMA混合模型可较好地拟合换热功率数据序列,并修正单一模型的误差,有效提高预测精度,为核电厂PRS换热能力验证提供参考。

Ecological stoichiometric comparison of plant-litter-soil system in mixed-species and monoculture plantations of Robinia pseudoacacia,Amygdalus davidiana,and Armeniaca sibirica in the Loess Hilly Region of China

We examined how afforestation patterns impact carbon(C),nitrogen(N),and phosphorus(P)stoichiometry in the plant-litter-soil system.Plant leaf,branch,stem,and root,litter,and soil samples were collected from mixedspecies plantations of Robinia pseudoacacia with Amygdalus davidiana(RPAD),R.pseudoacacia with Armeniaca sibirica(RPAS),and monocultures of R.pseudoacacia(RP),A.davidiana(AD),and A.sibirica(AS)in the Loess Hilly Region.The results showed that in mixed-species plantations,R.pseudoacacia had lower leaf N and P concentrations than in monocultures,while both A.davidiana and A.sibirica had higher leaf N and P concentrations.Soil P limited tree growth in both afforestation models.Mixing R.pseudoacacia with A.davidiana or A.sibirica reduced N-limitation during litter decomposition.Average soil total N and P concentrations were higher in RPAS than in RPAD,and both were higher than the corresponding monocultures.The average soil C:N ratio was the smallest in RPAS,while the average soil C:P ratio was larger in RPAS than in RP.A positive correlation between N and P concentrations,and between C:N and C:P ratios,was found in litter and all plant organs of mono-and mixedstands.Alternatively,for N concentration and C:N ratio,the correlations between plant(i.e.,leaf,branch,root)and litter and between plant and soil were inverse between plantation types.RPAD has an increased litter decomposition rate to release N and P,while RPAS has a faster rate of soil N mineralization.RPAD was the best plantation(mixed)to improve biogeochemical cycling,as soil nutrient restrictions,particularly for P-limitation,on trees growth were alleviated.This study thus provides insights into suitable tree selection and management by revealing C:N:P stoichiometry in the plant-litter-soil system under different afforestation patterns.

Leaf C:N:P stoichiometric homeostasis of a Robinia pseudoacacia plantation on the Loess Plateau

Homeostasis is the adaptability of a species to a changing environment.However,the ecological stoichiometric homeostasis of Robinia pseudoacacia L.in diff erent climatic regions is poorly understood but could provide insights into its adaptability in the loess hilly region.This study sampled 20 year-old R.pseudoacacia plantations at 10 sites along a north–south transect on the Loess Plateau.Variations in the ecological stoichiometric characteristics of leaf and soil carbon,nitrogen,and phosphorus were analysed and homeostatic characteristics of leaf ecological stoichiometric parameters in diff erent climates were identifi ed.Factors aff ecting leaf stoichiometry were assessed.The results show that R.pseudoacacia leaves were rich in nitrogen and defi cient in phosphorous during tree growth and development.Nitrogen and phosphorous levels in the soils of the loess region were lower than the average in soils in the rest of China.All ecological stoichiometric parameters of R.pseudoacacia leaves in two diff erent climates were considered“strictly homeostasis”.Precipitation,available phosphorus,and soil C:P were the main factors aff ecting the variation of C:N:P stoichiometry of R.pseudoacacia leaves.R.pseudoacacia in the loess hilly region has strong ecologically homeostatic characteristics and suggests that it is well-adapted to the area.

Regional differences in gully network connectivity based on graph theory:a case study on the Loess Plateau,China

Material exchange frequently occurs in gullies,and thus the relationship between a gullynetwork structure and sediment transport potential has attracted considerable interest.However,previous researches ignored the difficulty of material transport from sources to sinks,and did not quantify the connectivity of a network structure.In this study,we used a graph model structure to model gully networks of six typical sample areas in the Loess Plateau of China and quantified gully network connectivity using four indexes:average node strength,accessibility from sources to sinks,potential flow,and network structural connectivity index.Results show that:(1)Reflected by different quantitative indexes,the trends of gully network connectivity in different regions are similar.From north to south,the connectivity of a sample area first increases and then decreases.(2)The more mature gullies have stronger network connectivity.Small resistance is conducive to material transport in the gullies.(3)The node connectivity index of the gully network shows a significant aggregation distribution in space,and node connectivity on the main channel is often stronger than that on the branch trench.These results not only deepen the understanding of the process and mechanism of loess gully geomorphic development and evolution but also provide a reference for geomorphic studies.