Least Squares Support Vector Machine Research Articles

Quantitative Analysis of Pb in Soil Using Laser-Induced Breakdown Spectroscopy Based on Signal Enhancement of Conductive Materials

Studying efficient and accurate soil heavy-metal detection technology is of great significance to establishing a modern system for monitoring soil pollution, early warning and risk assessment, which contributes to the continuous improvement of soil quality and the assurance of food safety. Laser-induced breakdown spectroscopy (LIBS) is considered to be an emerging and effective tool for heavy-metal detection, compared with traditional detection technologies. Limited by the soil matrix effect, the LIBS signal of target elements for soil heavy-metal detection is prone to interference, thereby compromising the accuracy of quantitative detection. Thus, a series of signal-enhancement methods are investigated. This study aims to explore the effect of conductive materials of NaCl and graphite on the quantitative detection of lead (Pb) in soil using LIBS, seeking to find a reliable signal-enhancement method of LIBS for the determination of soil heavy-metal elements. The impact of the addition amount of NaCl and graphite on spectral intensity and parameters, including the signal-to-background ratio (SBR), signal-to-noise ratio (SNR), and relative standard deviation (RSD), were investigated, and the mechanism of signal enhancement by NaCl and graphite based on the analysis of the three-dimensional profile data of ablation craters and plasma parameters (plasmatemperature and electron density) were explored. Univariate and multivariate quantitative analysis models including partial least-squares regression (PLSR), least-squares support vector machine (LS-SVM), and extreme learning machine (ELM) were developed for the quantitative detection of Pb in soil with the optimal amount of NaCl and graphite, and the performance of the models was further compared. The PLSR model with the optimal amount of graphite obtained the best prediction performance, with an Rp that reached 0.994. In addition, among the three spectral lines of Pb, the univariate model of Pb I 405.78 nm showed the best prediction performance, with an Rp of 0.984 and the lowest LOD of 26.142 mg/kg. The overall results indicated that the LIBS signal-enhancement method based on conductive materials combined with appropriate chemometric methods could be a potential tool for the accurate quantitative detection of Pb in soil and could provide a reference for environmental monitoring.

Recognizing gender from images with facial makeup

<p>Recognizing the sex of an individual is a difficult task due to pose variation, occlusion, illumination effect, facial expression, plastic surgery, and makeup. In this manuscript, a novel approach for gender recognition with facial makeup is proposed. A novel Log-Gabor COSFIRE (LG-COSFIRE) filter is a shape-selective filter that is trained with prototype patterns of interest. The geometrical structure of the faces is acquired using the dual-tree complex wavelet transform (DT-CWT). Dense SIFT descriptor extracts the shape attributes of an image by building local histograms of gradient orientation. Finally, least square support vector machine (LS-SVM) is utilized to recognize the gender of an individual. The experiment was performed on self-built facial makeup for male and female (FMMF) database and achieves 89.7% accuracy.</p>

Novel optimized coupled rainfall model simulation based on stepwise decomposition technique

ABSTRACT Precipitation forecasting plays a pivotal role in guiding the effective management of regional water resources and providing crucial warnings for regional droughts and floods. Finding a monthly precipitation simulation model with robust fitting performance is a significant research endeavor in practical precipitation prediction. This paper introduces two modified African vulture optimization algorithms (MAVOA1 and MAVOA2). It provides hyperparameter optimization techniques for the least squares support vector machine (LSSVM), long short-term memory neural network (LSTM), and random forest (RF) models. These techniques are used to construct a monthly precipitation simulation model based on algorithmic optimization coupled with variational mode decomposition for full decomposition. The test results at five typical stations in the North China Plain reveal the following: (1) the LSSVM model demonstrates significantly better performance than the LSTM and RF models. (2) the MAVOA2-LSSVM model has the best-integrated effect: the average test fitting error is RMSE = 17.50 mm/month, MRE = 117.25%, NSE = 0.90, which shows its superiority in practical application and can significantly improve the accuracy of precipitation prediction; MAVOA2 is more suitable for machine learning models with more hyperparameters of its own, which provides a reference for hyperparameter optimization algorithms in the other fields.

Compressive strength detection of tunnel lining using hyperspectral images and machine learning

Traditional tunnel lining strength detection techniques are mostly contact-based, with relatively low detection efficiency. This study innovatively proposes hyperspectral imaging method to rapidly detect tunnel concrete lining strength from a machine vision perspective. Hyperspectral cameras were employed in indoor experiments to capture hyperspectral images of concrete specimens with different compressive strength levels. The differences of concrete strength based on hyperspectral reflectance characteristics were analysed using hyperspectral images and machine learning algorithms. Firstly, the K-Nearest Neighbors (KNN) classification algorithm was used to predict the classification of the concrete hyperspectral dataset with accuracy mostly exceeding 90 %. The results indicate distinctive differences in hyperspectral reflectance characteristics among concrete specimens. Furthermore, compressive strength prediction of different concrete specimens was carried out using Principal component regression (PCR), Partial least squares regression (PLSR), and Least Squares Support Vector Machine (LSSVM) machine learning models on both original and Savitzky-Golay(S-G) processed spectral data. LSSVM and PLSR models performed excellently in the visible light spectrums(400–1000 nm), with LSSVM excelling in the near-infrared spectrums(900–1700 nm). Finally, the feasibility of using Hyperspectral imaging(HSI) technology to detect tunnel lining strength was demonstrated at the shield tunnel model site. The predicted results were validated by combining with strength values measured by the rebound hammer, and presented promoising research path for automated non-destructive detection of concrete tunnel lining strength.

Machine learning models applied to altimetry era tide gauge and grid altimetry data for comparative long-term trend estimation: A study from Shikoku Island, Japan

Estimation of sea level trends is essential for understanding sea level rise dynamics. In this study, the performance of traditional Ordinary Least Squares (OLS) linear trend forecasting is compared with modern machine learning techniques, specifically Random Forests (RF) and Least Squares Support Vector Machines (LS-SVM).These methods are applied to 50 years of long-term tide gauge (TG) data from six tide gauge stations off the coast of Shikoku Island, Japan, and CMEMS Grid Altimetry data from 1993 to the present. The analysis uses OLS, RF, and LS-SVM to estimate trends from both data sets and compares the results. The objective is to determine the consistency and accuracy of RF and LS-SVM methods compared to the OLS method. The results indicate that machine learning algorithms (LS-SVM) effectively estimate sea level trends, offering potential improvements in precision for both long-term and medium-term analyses. Shikoku Island's coastal sea level trend is determined as 2.91±1.44 mm/yr using TG data and 3.00±1.52 mm/yr using CMEMS Grid Altimeter data with the OLS approach. Using the LS-SVM approach, the trend is found as 2.96±1.58 mm/yr with TG data and 3.02±1.60 mm/yr with CMEMS Grid Altimetry data. The novelty of this study lies in its thorough comparison of traditional and machine learning approaches for sea level trend estimation, providing valuable insights for future predictions of the sea level rise.

Data‐driven nonlinear state observation for controlled systems: A kernel method and its analysis

AbstractThis work proposes a data‐driven state observation algorithm for nonlinear dynamical systems, when the true state trajectory is not measurable and hence the states information needs to be reconstructed from input and output measurements. Such a reduction is formed by kernel canonical correlation analysis (KCCA), which (i) implicitly maps the available input–output data into a higher‐dimensional feature space, namely the reproducing kernel Hilbert space (RKHS); (ii) finds a projection of the past input–output data and a projection of the future input–output data with maximal correlation; and (iii) identifies the projected inputs and outputs, namely the canonical variates, as the observed states. We adopt a least squares support vector machine (LS‐SVM) formulation for KCCA, which imposes regularization on the vectors that specify the projections and is amenable to convex optimization. We prove theoretically that, based on the statistical consistency of KCCA, the observed states determined by the proposed state observer has a guaranteed correlativity with the actual states (when properly transformed). Furthermore, such observed states, when supplemented with the information of succeeding inputs, can be used to predict the succeeding outputs with guaranteed upper bound on the prediction error. Case studies are performed on two numerical examples and an exothermic continuously stirred tank reactor (CSTR).

A novel optimization rainfall coupling model based on stepwise decomposition technique

Traditional decomposition integration models decompose the original sequence into subsequences, which are then proportionally divided into training and testing periods for modeling. Decomposition may cause data aliasing, then the decomposed training period may contain part of the test period data. A more effective method of sample construction is sought in order to accurately validate the model prediction accuracy. Semi-stepwise decomposition (SSD), full stepwise decomposition (FSD), single model semi-stepwise decomposition (SMSSD), and single model full stepwise decomposition (SMFSD) techniques were used to create the samples. This study integrates Variational Mode Decomposition (VMD), African Vulture Optimization Algorithm (AVOA), and Least Squares Support Vector Machine (LSSVM) to construct a coupled rainfall prediction model. The influence of different VMD parameters α is examined, and the most suitable stepwise decomposition machine learning coupled model algorithm for various stations in the North China Plain is selected. The results reveal that SMFSD is relatively the most suitable tool for monthly precipitation forecasting in the North China Plain. Among the predictions for the five stations, the best overall performance is observed at Huairou Station (RMSE of 18.37 mm, NSE of 0.86, MRE of 107.2%) and Jingxian Station (RMSE of 24.74 mm, NSE of 0.86, MRE of 51.71%), while Hekou Station exhibits the poorest performance (RMSE of 25.11 mm, NSE of 0.75, MRE of 173.75%).

Inversion Study on Landslide Seepage Field Based on Swarm Intelligence Optimization Least-Square Support Vector Machine Algorithm

The permeability coefficient of landslide mass, a key parameter in the study of reservoir landslides, is commonly obtained through in situ and laboratory tests; however, the tests are costly and subject to high variability, leading to potential biases. In this paper, a new method was proposed to inversely estimate the permeability coefficient of landslide layers using monitoring data of groundwater level (GWL). First, the landslide transient seepage simulation was conducted to generate sample data for permeability coefficients and GWL during a reservoir operation cycle. Second, using GWL data as input and permeability coefficient data as output, the least-square support vector machine (LSSVM) was trained with two optimization algorithms, the particle swarm optimization (PSO) algorithm and the whale optimization algorithm (WOA), to construct the nonlinear mapping relationship between simulated GWL and permeability coefficients. Third, the accurate permeability coefficients for landslide seepage simulation were inverted or predicted based on the monitored GWL. Finally, using the inverted permeability coefficients for landslide seepage simulation, we compared simulation results with actual monitored GWL and achieved good consistency. In addition, this paper compared the inversion effects of three different algorithms: the standard LSSVM, PSO-LSSVM, and WOA-LSSVM. This study showed that these three algorithms had good nonlinear fitting effects in studying landslide seepage fields. Among them, using the inversion values from PSO-LSSVM for landslide seepage simulation resulted in the smallest relative error compared to actual monitoring data. Within a single reservoir operation cycle, the simulated water level changes were also largely consistent with the monitored water level changes. The results could provide a reference to determine landslide permeability coefficients and seepage.

Rapid and non‐destructive determination of total phenolic contents using UV–NIR spectroscopy of dehydrated mushroom (Lentinus edodes)

AbstractMultispectral imaging (MSI) is an emerging technique that ranges from light spectrum of UV–NIR (405–970 nm) used for rapid determination of phenolic contents. The current study focuses on the determination of total phenolic content (TPC) in a fast and non‐invasive way in hot‐air dehydrated Lentinus edodes. The spectral information of MSI has been combined with various chemometrics like partial least squares (PLS), back propagation neural networks (BPNN), and least squares‐support vector machines (LS‐SVM) for the quantitative prediction of phenolic contents. Fresh mushrooms possessed 1.49 and 1.73 GAE g kg−1 TPC in processed samples at 10% moisture content. Comparing models, PLS acquired significant Rp with a lower RMSEP of 0.9980 and 0.1039 and was considered an outstanding algorithm, followed by LS‐SVM of 0.9777 and 0.1417 coefficient of determination and error of prediction data set, respectively. It is concluded that MSI spectroscopy produces highly significant results to determine functional food properties in a rapid and non‐destructive way. The present study will provide a strong platform for the fast online determination of phenolic profile of agricultural commodities.Practical applicationsUtilization of spectroscopy for non‐destructive detection of phenolic contents is defined through the combination of chemometrics with spectra of multispectral imaging system and emphasizes control processing of postharvest horticulture produces. Prediction results R2 above 80% describe that these chemometrics are successfully capable of estimating total phenolic contents in control processing for quality preservation.

Rapid classification of rice according to storage duration via near-infrared spectroscopy and machine learning

Rice is the most important staple crop for more than half of the world's population. As rice quality can deteriorate during storage, methods that can effectively classify rice according to its storage duration are essential. However, existing methods of assessing rice storage time are time-consuming, laborious, and incompatible with modern industrial processing technologies. Therefore, we investigated the ability of near-infrared spectroscopy combined with machine learning algorithms to distinguish rice storage duration. A total of 482 rice samples were analyzed, which included 74, 100, and 308 samples produced during 2015–2016, 2017–2018, and 2020–2021, respectively. Five pre-processing methods were initially applied to the spectra to enhance the accuracy of the discrimination model. Subsequently, two-dimensional correlation spectroscopy and competitive adaptive reweighted sampling (CARS) were used to extract the characteristic spectra associated with storage time. Finally, three pattern recognition methods (K-nearest neighbor analysis, linear discriminant analysis, and least squares support vector machine (LS-SVM)) were compared for their effectiveness in constructing classification models. The results indicated that the best model for identifying the storage duration of rice was established after spectral pre-processing with the standard normal variate and first derivative, using the CARS algorithm to select feature wavelengths, and applying the LS-SVM modeling method, which together yielded correct identification rates of 99.72% and 91.67% for the calibration and validation sets, respectively. Thus, we propose near-infrared spectroscopy coupled with machine learning algorithms as an effective approach for classifying rice according to storage duration, which can facilitate evaluations of rice freshness in the market.

Determination of soluble solids content in loquat using near-infrared spectroscopy coupled with broad learning system and hybrid wavelength selection strategy

Soluble solids content (SSC) serves as a crucial metric in assessing loquat quality. To achieve fast, non-destructive, and accurate detection of SSC, the combination of near-infrared (NIR) spectroscopy and broad learning system (BLS) model was developed in this study. Meanwhile, the influence of six spectral preprocessing methods and six wavelength selection algorithms on SSC prediction was explored, and the performance of the BLS model was compared with the widely used partial least squares regression (PLSR) and least squares support vector machine (LS-SVM) models. The results showed that Savitzky-Golay smoothing combined with standard normal transformation (SG-SNV) offered the optimal pretreatment, and the hybrid method, namely interval variable iterative space shrinking analysis and successive projections algorithm (iVISSA-SPA), emerged as the most effective wavelength selection method. the BLS model outperformed both PLSR and LS-SVM models. Specifically, the BLS model based on the iVISSA-SPA method achieved the optimal SSC prediction results, with = 0.8646, RMSEP = 0.5104, and RPD = 2.7481. Therefore, NIR spectroscopy coupled with the BLS model and hybrid wavelength selection strategy could rapidly, non-destructively and accurately detect the SSC of loquat, providing a viable alternative for SSC prediction in fruit.

Pipeline Leak Detection: A Comprehensive Deep Learning Model Using CWT Image Analysis and an Optimized DBN-GA-LSSVM Framework.

Detecting pipeline leaks is an essential factor in maintaining the integrity of fluid transport systems. This paper introduces an advanced deep learning framework that uses continuous wavelet transform (CWT) images for precise detection of such leaks. Transforming acoustic signals from pipelines under various conditions into CWT scalograms, followed by signal processing by non-local means and adaptive histogram equalization, results in new enhanced leak-induced scalograms (ELIS) that capture detailed energy fluctuations across time-frequency scales. The fundamental approach takes advantage of a deep belief network (DBN) fine-tuned with a genetic algorithm (GA) and unified with a least squares support vector machine (LSSVM) to improve feature extraction and classification accuracy. The DBN-GA framework precisely extracts informative features, while the LSSVM classifier precisely distinguishes between leaky and non-leak conditions. By concentrating solely on the advanced capabilities of ELIS processed through an optimized DBN-GA-LSSVM model, this research achieves high detection accuracy and reliability, making a significant contribution to pipeline monitoring and maintenance. This innovative approach to capturing complex signal patterns can be applied to real-time leak detection and critical infrastructure safety in several industrial applications.

Feature extraction method for ultrasonic pipeline defects based on fractional-order VMD

ABSTRACT Ensuring the safety of pipeline transportation is vital for societal well-being. Traditional methods for identifying defects in steel pipelines through ultrasonic echo signal pattern recognition often fail to extract comprehensive and effective features crucial for accurate defect detection. This study introduces an innovative feature extraction method employing a fractional Fourier transform variational modal decomposition (FRFT-VMD), hereafter referred to as the fractional-order VMD algorithm. This method utilises the fourth-order central moment and envelope entropy to optimise several key parameters: the fractional order in the Fourier transform, the number of decomposition layers, and the penalty factor in variational modal decomposition. To evaluate the effectiveness of the proposed method, ultrasonic echo signals from both finite element simulations and experimental platforms were analysed using the FRFT-VMD technique. The features extracted were then classified using a Least Squares Support Vector Machine (LSSVM) to determine defect depths. The results show a recognition accuracy of 95.2% in simulated signals and 89.1% in experimentally measured signals across various defect depths, indicating a significant improvement over existing feature extraction methodologies. The fractional-order VMD algorithm proves to be superior in extracting features that enhance the accuracy of defect identification in steel pipelines.

Dynamic internal resistance modeling and thermal characteristics of lithium-ion batteries for electric vehicles by considering state of health

Attempting to solve the problem of inconsistent dynamic thermal characteristics caused by transient changes in internal resistance of lithium-ion batteries (LIBs) for electric vehicles (EVs) under service conditions, taking a 9.4Ah prismatic LIB as the study object, the internal resistance data under different state of health (SOH), state of charge (SOC), temperature (T), charge/discharge rates of LIB are obtained through the mixed-rate hybrid pulse power characterization (MR-HPPC) test. Based on this, a dynamic internal resistance model (DIRM) considering the SOH is proposed by the prediction of the least squares support vector machines (LS-SVM) model. In this study, a thermoelectric coupling model and a resistance-thermal co-simulation of the LIB are established based on DIRM, and the temperature rise characteristics under constant rates and dynamic operating conditions are investigated at different SOH states. The maximum DIRM error (DIRME) between DIRM-based simulated maximum transient temperature (Tmax-tra) and test Tmax-tra is 0.88 °C at constant charge/discharge rate operating conditions, while the maximum DIRME between DIRM-based simulated Tmax-tra and test Tmax-tra is 0.21 °C under dynamic operating conditions. The results can provide a theoretical support for improving the modeling precision of the heat generation of LIBs with different aging levels for accurate thermal management design.

Optimization of PV module ash accumulation prediction based on an improved vulture algorithm

Abstract In this research, an ash accumulation prediction model based on the Improved Bald Eagle Algorithm (IBES) optimization is proposed, and a least-squares support vector machine (LSSVM) is used to investigate the correlation between climatic parameters and the quantity of ash deposition on PV panel surfaces. Ash deposit affects PV panel output power over time and is influenced by meteorological factors. Following a thorough selection procedure, meteorological variables like wind speed and rainfall are used in the model to forecast the amount of ash that will accumulate on the slab. The original condor method was prone to local optimum convergence and had a slow convergence rate. These weaknesses are addressed by the optimal selection of the next iteration, which introduces the Cauchy variation operator to mutate the population’s optimal individuals. The optimization parameters of the revised algorithm are incorporated into the model and results are compared to models of other optimization algorithms. The outcomes demonstrate that the IBES-LSSVM ash accumulation prediction model has a higher degree of fit and a reduced prediction error.

Rapid monitoring of the spatial distribution of soil organic matter using unmanned aerial vehicle imaging spectroscopy

ABSTRACT An existing technique for quickly obtaining the spatial distribution data of soil organic matter (SOM) involves combining an unmanned aerial vehicle (UAV) platform and imaging spectroscopy technology; this technique is suitable for precision agricultural management and real-time monitoring. In this study, the average spectral data were first extracted via circular regions of interest (ROIs) of different radii and modelled using the measured SOM to determine the optimal extraction range for the spectral data. Partial least-squares regression (PLSR) and support vector machine (SVM) models were constructed based on the raw spectra, standard normal transform spectra, multiple scattering-corrected spectra, and first-order differential spectra. The difference index, ratio index, and normalized index were calculated and analysed for their correlation with SOM content, and the spectral indices were screened to establish the SOM prediction model. Finally, the accuracy of the model was evaluated, and we selected the optimal model to map the spatial distribution of SOM. The results of our study show that 1) the radius of the ROI significantly affects the model’s accuracy; 2) the accuracy of the nonlinear SVM model is much higher than that of the linear PLSR model; 3) the prediction model based on spectral indices is better than full-band model. This study can serve as a model for the application of UAV imaging spectroscopy technology to field-scale SOM estimations and rapid monitoring, as well as provide technical support for precision agriculture and climate change studies.

A Realistic Method for Diagnosing Bearing Faults of Three-Phase Induction Motor using Advanced Support Vector Machine Algorithm

Bearings are one of the most crucial components of any induction motor. The faults or failure of bearings can lead to high maintenance and operational costs. To ensure the induction motor operates consistently, it is crucial to identify and diagnose bearing faults at an early stage. In this perspective, this manuscript presents a method based on least square support vector machine (LSSVM) to identify the bearing defects of an induction motor. The IM parameters, such as stator currents, input voltage, and rotor speed, were extracted using experimental setup at different loads and health conditions of bearing. Then these parameters were used to train and validate the LSSVM algorithm in MATLAB to diagnose the bearing defects. The proposed method provides 97.14% fault prediction accuracy with an RBF kernel.

Effective approach to assess higher heating value of biomass from ultimate and proximate analysis

AbstractFor proper design and operation of biomass‐based energy systems, it is important to determine the higher heating value (HHV) of biomass. In this paper, two machine learning (ML) approaches, namely extra trees (ET) and least squares support vector machine (LSSVM), are used to predict the value of HHV associated with biofuels. The data required for HHV calculation, including proximate and ultimate analyses datasets, were collected from the literature. The performances of these two ML approaches for predicting biomass HHV were then compared with other smart models available in the literature. Even though the available empirical models can predict the biomass HHV with acceptable precision, it was found that our proposed ML techniques have a superior performance based on the error analysis; the proposed approaches also consider all key biomass characteristics in the developed models. In addition, the ET model proved to be slightly more accurate compared to the LSSVM model. Additionally, the developed proximate‐based ET model showed better performance compared to the ultimate‐based ET model. The most influential parameters in the developed ET models for the proximate and ultimate approaches were determined to be ash fraction and carbon fraction, respectively. Finally, it was concluded that the smart modelling techniques can be utilized as a robust and reliable alternative predictive methodology to replace direct laboratory measurement of the biomass HHV.

Numerical solutions to one dimensional linear Volterra–Fredholm integral equations based on LS-SVM model

This paper introduces a novel numerical approach using least squares support vector machine (LS-SVM) to handle one dimensional linear Volterra–Fredholm integral equations. This method can also handle weak singular kernel Fredholm/Volterra integral equations and linear systems of Volterra–Fredholm integral equations. By minimizing errors and utilizing least squares support vector machines, the approach derives continuous closed-form solutions to the origin problem. Ultimately, the method offers an efficient numerical approximation for solving Volterra–Fredholm integral equations through the resolution of linear equations. The paper includes multiple numerical examples that demonstrate the effectiveness and superiority of the proposed method.

Discrimination of New and Aged Seeds Based on On-Line Near-Infrared Spectroscopy Technology Combined with Machine Learning

The harvest year of maize seeds has a significant impact on seed vitality and maize yield. Therefore, it is vital to identify new seeds. In this study, an on-line near-infrared (NIR) spectra collection device (899–1715 nm) was designed and employed for distinguishing maize seeds harvested in different years. Compared with least squares support vector machine (LS-SVM), k-nearest neighbor (KNN), and extreme learning machine (ELM), the partial least squares discriminant analysis (PLS-DA) model has the optimal recognition performance for maize seed harvest years. Six different preprocessing methods, including Savitzky–Golay smoothing (SGS), standard normal variate transformation (SNV), multiplicative scatter correction (MSC), Savitzky–Golay 1 derivative (SG-D1), Savitzky–Golay 2 derivative (SG-D2), and normalization (Norm), were used to improve the quality of the spectra. The Monte Carlo cross-validation uninformative variable elimination (MC-UVE), competitive adaptive reweighted sampling (CARS), bootstrapping soft shrinkage (BOSS), successive projections algorithm (SPA), and their combinations were used to obtain effective wavelengths and decrease spectral dimensionality. The MC-UVE-BOSS-PLS-DA model achieved the classification with an accuracy of 88.75% using 93 features based on Norm preprocessed spectral data. This study showed that the self-designed NIR collection system could be used to identify the harvested years of maize seed.