Particle Swarm Optimization-based Support Vector Machine Research Articles

Interpretable machine learning-based analysis of hydration and carbonation of carbonated reactive magnesia cement mixes

This study explored the influence of different input variables on the hydration and carbonation degree of carbonated reactive magnesia cement (RMC) system by employing six machine learning algorithms. These included support vector machine (SVM), particle swarm optimization-based SVM (PSO-SVM), extreme learning machine (ELM), grey wolf optimizer-based SVM (GWO-SVM), kernel extreme learning machine (KELM), and extreme gradient boosting (XGBoost). The followed approach enabled the deep learning of the relevant database to achieve parameter prediction. Two feature analysis methodologies, i.e. partial dependence plot (PDP) and SHapley Additive exPlanations (SHAP), were applied to uncover the operating laws underpinning the black box operation characteristics of machine learning models. Results revealed that GWO-SVM and XGBoost outperformed all other models in predicting the hydration and carbonation degree of the complete database set (R2 of the total database set was 0.9470/0.9775 and 0.9663/0.9727 for hydration and carbonation degree, respectively). Factors such as carbonation duration, CO2 concentration, pre-curing temperature, and w/b directly influenced the degree of hydration and carbonation. Among them, carbonation duration and CO2 concentration were two of the most influential factors, resulting in the promotion of brucite consumption and accelerating the formation of carbonation products.

Intelligent Diagnosis Based on Double-Optimized Artificial Hydrocarbon Networks for Mechanical Faults of In-Wheel Motor

To avoid the potential safety hazards of electric vehicles caused by the mechanical fault deterioration of the in-wheel motor (IWM), this paper proposes an intelligent diagnosis based on double-optimized artificial hydrocarbon networks (AHNs) to identify the mechanical faults of IWM, which employs a K-means clustering and AdaBoost algorithm to solve the lower accuracy and poorer stability of traditional AHNs. Firstly, K-means clustering is used to improve the interval updating method of any adjacent AHNs molecules, and then simplify the complexity of the AHNs model. Secondly, the AdaBoost algorithm is utilized to adaptively distribute the weights for multiple weak models, then reconstitute the network structure of the AHNs. Finally, double-optimized AHNs are used to build an intelligent diagnosis system, where two cases of bearing datasets from Paderborn University and a self-made IWM test stand are processed to validate the better performance of the proposed method, especially in multiple rotating speeds and the load conditions of the IWM. The double-optimized AHNs provide a higher accuracy for identifying the mechanical faults of the IWM than the traditional AHNs, K-means-based AHNs (K-AHNs), support vector machine (SVM), and particle swarm optimization-based SVM (PSO-SVM).

Integrated Sensing and Computing for Wearable Human Activity Recognition with MEMS IMU and BLE Network

Abstract The miniature sensor devices and power-efficient Body Area Networks (BANs) for Human Activity Recognition (HAR) have gained increasing interest in different fields, including Daily Life Assistants (DLAs), medical treatment, sports analysis, etc. The HAR systems normally collect data with wearable sensors and implement the computational tasks with a host machine, where real-time transmission and processing of sensor data raise a challenge for both the network and the host machine. This investigation focuses on the hardware/software co-design for optimized sensing and computing of wearable HAR sensor networks. The contributions include (1) design of a miniature wearable sensor node integrating a Micro-Electro-Mechanical System Inertial Measurement Unit (MEMS IMU) with a Bluetooth Low Energy (BLE) in-built Micro-Control Unit (MCU) for unobtrusive wearable sensing; (2) task-centric optimization of the computation by shifting data pre-processing and feature extraction to sensor nodes for in-situ computing, which reduces data transmission and relieves the load of the host machine; (3) optimization and evaluation of classification algorithms Particle Swarm Optimization-based Support Vector Machine (PSO-SVM) and Cross Validation-based K-Nearest Neighbors (CV-KNN) for HAR with the presented techniques. Finally, experimental studies were conducted with two sensor nodes worn on the wrist and elbow to verify the effectiveness of the recognition of 10 virtual handwriting activities, where 10 recruited participants each repeated an activity 5 times. The results demonstrate that the proposed system can implement HAR tasks effectively with an accuracy of 99.20 %.

Multi-Stage Feature Extraction and Classification for Ship-Radiated Noise.

Due to the complexity and unique features of the hydroacoustic channel, ship-radiated noise (SRN) detected using a passive sonar tends mostly to distort. SRN feature extraction has been proposed to improve the detected passive sonar signal. Unfortunately, the current methods used in SRN feature extraction have many shortcomings. Considering this, in this paper we propose a new multi-stage feature extraction approach to enhance the current SRN feature extractions based on enhanced variational mode decomposition (EVMD), weighted permutation entropy (WPE), local tangent space alignment (LTSA), and particle swarm optimization-based support vector machine (PSO-SVM). In the proposed method, first, we enhance the decomposition operation of the conventional VMD by decomposing the SRN signal into a finite group of intrinsic mode functions (IMFs) and then calculate the WPE of each IMF. Then, the high-dimensional features obtained are reduced to two-dimensional ones by using the LTSA method. Finally, the feature vectors are fed into the PSO-SVM multi-class classifier to realize the classification of different types of SRN sample. The simulation and experimental results demonstrate that the recognition rate of the proposed method overcomes the conventional SRN feature extraction methods, and it has a recognition rate of up to 96.6667%.

ESVM-SWRF: Ensemble SVM-based sample weighted random forests for liver disease classification.

Recently, a significant way to diagnose the disease is using the model of medical data mining. The most challenging task in the healthcare field is to face a large amount of data during disease analyzes and prediction. Once the data are transformed into valuable data by means of data mining models then the actual prediction and decision making is easier. The existing studies met few shortcomings because of higher execution time, more computational complexities, less scalability, slow convergence, and lack of providing the solution. In this article, we have proposed an ensemble SVM-based sample weighted random forests (eSVM-swRF) with novel improved colliding body optimization (NICBO) algorithm to predict liver diseases. The extraction, loading, transformation, and analysis (ELTA) are used to pre-process the patient data. The significant feature with a suitable model is generated depending upon the filter-based method. Based on eSVM-swRF, the parameter values such as penalty parameter (P), threshold (T), and mTry are optimized via a novel improved colliding boding optimization (NICBO) algorithm. The UCI dataset provides liver disease data for this study. The implementation platform of RapidMiner Studio version 7.6 with different evaluation measures is used to validate the performance of eSVM-swRF with the NICBO method. Anyway, the proposed method yields outstanding performance than other existing methods such as Particle Swarm Optimization-based Support Vector Machine (PSO-SVM), fuzzy adaptive, and neighbor weighted k-NN (FuzzyANWKNN), Naïve Bayes-based Support Vector Machine (NB-SVM), and Neural network.

Rolling Bearing Fault Diagnosis Based on VMD-MPE and PSO-SVM

The goal of the paper is to present a solution to improve the fault detection accuracy of rolling bearings. The method is based on variational mode decomposition (VMD), multiscale permutation entropy (MPE) and the particle swarm optimization-based support vector machine (PSO-SVM). Firstly, the original bearing vibration signal is decomposed into several intrinsic mode functions (IMF) by using the VMD method, and the feature energy ratio (FER) criterion is introduced to reconstruct the bearing vibration signal. Secondly, the multiscale permutation entropy of the reconstructed signal is calculated to construct multidimensional feature vectors. Finally, the constructed multidimensional feature vector is fed into the PSO-SVM classification model for automatic identification of different fault patterns of the rolling bearing. Two experimental cases are adopted to validate the effectiveness of the proposed method. Experimental results show that the proposed method can achieve a higher identification accuracy compared with some similar available methods (e.g., variational mode decomposition-based multiscale sample entropy (VMD-MSE), variational mode decomposition-based multiscale fuzzy entropy (VMD-MFE), empirical mode decomposition-based multiscale permutation entropy (EMD-MPE) and wavelet transform-based multiscale permutation entropy (WT-MPE)).

Generalized Composite Multiscale Diversity Entropy and Its Application for Fault Diagnosis of Rolling Bearing in Automotive Production Line

This paper considers the entropy based feature extraction method for the fault diagnosis of rolling bearings in automobile production line, where the fault information is difficult to identify due to the strong nonlinear and non-stationary characteristics of the fault vibration signals. In our work, a novel entropy based method called generalized composite multiscale diversity entropy (GCMDiEn) is developed. This method can effectively track the inside pattern changes of the time series by the description of cosine similarity between adjacent orbits. Unlike most of the existing entropy based results which concentrate on the static orderliness, we analyze the dynamic complexity characteristics of the arbitrary time series. Moreover, compared with the multiscale diversity entropy, GCMDiEn calculates all coarse-grained time series entropy value with the same scale and extends the first-order moment to second-order moment to mitigate the shortcomings of the short time series instability and losing part of the frequency region information. This paper proposed a new rolling bearing fault diagnosis framework which combined the GCMDiEn with the empirical wavelet transform (EWT), Laplacian score (LS), and particle swarm optimization-based support vector machine (PSO-SVM). Finally, the simulation results show the superiority of the GCMDiEn method over the multiscale diversity entropy method. The proposed framework has a higher fault recognition rate (99.38%) than the existing methods.

Cultural Emperor Penguin Optimizer and Its Application for Face Recognition

Face recognition is an important technology with practical application prospect. One of the most popular classifiers for face recognition is support vector machine (SVM). However, selection of penalty parameter and kernel parameter determines the performance of SVM, which is the major challenge for SVM to solve classification problems. In this paper, with a view to obtaining the optimal SVM model for face recognition, a new hybrid intelligent algorithm is proposed for multiparameter optimization problem of SVM, which is a fusion of cultural algorithm (CA) and emperor penguin optimizer (EPO), namely, cultural emperor penguin optimizer (CEPO). The key aim of CEPO is to enhance the exploitation capability of EPO with the help of cultural algorithm basic framework. The performance of CEPO is evaluated by six well-known benchmark test functions compared with eight state-of-the-art algorithms. To verify the performance of CEPO-SVM, particle swarm optimization-based SVM (PSO-SVM), genetic algorithm-based SVM (GA-SVM), CA-SVM, and EPO-SVM, moth-flame optimization-based SVM (MFO-SVM), grey wolf optimizer-based SVM (GWO-SVM), cultural firework algorithm-based SVM (CFA-SVM), and emperor penguin and social engineering optimizer-based SVM (EPSEO-SVM) are used for the comparison experiments. The experimental results confirm that the parameters optimized by CEPO are more instructive to make the classification performance of SVM better in terms of accuracy, convergence rate, stability, robustness, and run time.

Feature Extraction of Ship-Radiated Noise Based on Enhanced Variational Mode Decomposition, Normalized Correlation Coefficient and Permutation Entropy.

Due to the complexity and variability of underwater acoustic channels, ship-radiated noise (SRN) detected using the passive sonar is prone to be distorted. The entropy-based feature extraction method can improve this situation, to some extent. However, it is impractical to directly extract the entropy feature for the detected SRN signals. In addition, the existing conventional methods have a lack of suitable de-noising processing under the presence of marine environmental noise. To this end, this paper proposes a novel feature extraction method based on enhanced variational mode decomposition (EVMD), normalized correlation coefficient (norCC), permutation entropy (PE), and the particle swarm optimization-based support vector machine (PSO-SVM). Firstly, EVMD is utilized to obtain a group of intrinsic mode functions (IMFs) from the SRN signals. The noise-dominant IMFs are then eliminated by a de-noising processing prior to PE calculation. Next, the correlation coefficient between each signal-dominant IMF and the raw signal and PE of each signal-dominant IMF are calculated, respectively. After this, the norCC is used to weigh the corresponding PE and the sum of these weighted PE is considered as the final feature parameter. Finally, the feature vectors are fed into the PSO-SVM multi-class classifier to classify the SRN samples. The experimental results demonstrate that the recognition rate of the proposed methodology is up to 100%, which is much higher than the currently existing methods. Hence, the method proposed in this paper is more suitable for the feature extraction of SRN signals.

Particle Swarm Optimization-Based SVM for Classification of Cable Surface Defects of the Cable-Stayed Bridges

The machine vision system was employed to inspect the surface defects of bridge cables of cable-stayed bridges. After the acquisition and preprocessing of the defect images, it is necessary to classify and identify the defects of the cables to meet the requirements of non-destructive testing and evaluation. In this paper, feature extraction for defect images was performed using mathematical statistical methods. After that, 10 feature parameters including shape features, grayscale features and texture features of the defect images were obtained and selected for a classification model of support vector machine (SVM). To improve the SVM classification performance, the particle swarm optimization algorithm (PSO) was adopted to obtain the punish factor c and the kernel parameter g of the SVM model, namely the PSO-SVM algorithm. Finally, our PSO-SVM classification model was employed to implement the classification of real surface defect images of the bridge cables. Longitudinal crack, transverse crack, surface corrosion, and pothole defect can be automatically identified and the classification accuracy reached 96.25%. The experimental results showed that the PSO-SVM model can improve the classification performance of the surface defects. Based on the effective classification, we can find the distribution characteristics of the surface defects of the cable. It is very important to analyze the relationship between the type of surface defects and the material of the protective layer, so as to adopt appropriate materials and reasonable maintenance measures. Thus, it has a great significance in the structural health monitoring of bridge cables.

A memetic algorithm using emperor penguin and social engineering optimization for medical data classification

Gene selection and classification of microarray data play an important role in cancer diagnosis and treatment. One of the most popular and faster classification model is support vector machine (SVM). However, the major challenge in SVM lies in the selection of its two parameters, namely, regularization parameter C and kernel parameter γ. Attempts have been made to improve the performance of SVM by tuning these two parameters with the help of metaheuristics. Although existing metaheuristics can search the promising regions of the search space, they are unable to explore the global optimum efficiently. In this paper, a memetic algorithm-based SVM (M-SVM) is presented for simultaneous feature selection and optimization of SVM parameters. The memetic algorithm is a fusion of local search strategy using social engineering optimizer (SEO) and global optimization framework using emperor penguin optimizer (EPO). The idea of embedding SEO in EPO is to enhance the exploitation capability of EPO. The performance of our algorithm is evaluated on seven standard benchmark datasets. To prove the efficacy of the method, it is compared with particle swarm optimization based SVM (PSO-SVM), genetic algorithm based SVM (GA-SVM), and fifteen other state-of-the-art methods. The experimental results confirm that the proposed method significantly outperforms other existing techniques in terms of accuracy and number of selected genes. The proposed method is validated using a statistical analysis, namely, ANOVA.

Intelligent Fault Diagnosis of Rotating Machinery Using ICD and Generalized Composite Multi-Scale Fuzzy Entropy

A new intelligent fault diagnosis algorithm of rotating machinery based on intrinsic characteristic-scale decomposition (ICD), generalized composite multi-scale fuzzy entropy (GCMFE), Laplacian score (LS), and particle swarm optimization-based support vector machine (PSO-SVM) is proposed in this paper. First, ICD is applied to decompose a vibration signal into a sum of product components. Second, GCMFE is proposed to evaluate the complexity of the decomposed vibration signals. GCMFE can overcome the drawbacks of the MFE method, and the superiority of GCMFE is validated using a simulated signal. Third, the LS method is utilized to select the extracted fault features. In the end, the selected features are input into the PSO-SVM to classify different health conditions. The simulated and experimental results validate the superiority of the proposed method in fault feature extraction compared with three other methods: ICD-MFE, ICD-CMFE, and GCMFE.

Particle Swarm Optimization Based Support Vector Machine (P-SVM) for the Segmentation and Classification of Plants

With the rapid growth in urbanization and population, it has become an earnest task to nurture and grow plants that are both important in sustaining the nature and the living beings needs. In addition, there is a need for preserving the plants having global importance both economically and environmentally. Locating such species from the forest or shrubs having human involvement is a time consuming and costly task to perform. Therefore, in this paper, a novel method is presented for the segmentation and classification of the seven different plants, named Guava, Jamun, Mango, Grapes, Apple, Tomato, and Arjun, based on their leaf images. In the first phase, both real-time images and images from the crowdAI database are collected and preprocessed for noise removal, resizing, and contrast enhancement. Then, in the second phase, different features are extracted based on color and texture. The third phase includes the segmentation of images using a k-means algorithm. The fourth phase consists of the training of support vector machine, and finally, in the last phase, the testing is performed. Particle swarm optimization algorithm is used for selecting the best possible value of the initialization parameter in both the segmentation and classification processes. The proposed work achieves higher experimental results, such as sensitivity = 0.9581, specificity = 0.9676, and accuracy = 0.9759, for segmentation and classification accuracy = 95.23 when compared with other methods.

A novel optimized SVM classification algorithm with multi-domain feature and its application to fault diagnosis of rolling bearing

Abstract Sensitive feature extraction from the raw vibration signal is still a great challenge for intelligent fault diagnosis of rolling bearing. Current fault classification framework generally concentrates on the pattern of classifier with single-domain feature, which is easy to induce insufficient feature extraction and low recognition accuracy. Therefore, to address this issue and improve intelligent diagnostic accuracy of rolling bearing, this paper proposes a novel fault classification algorithm based on optimized SVM with multi-domain feature, which mainly consists of three stages (i.e. multi-domain feature extraction, feature selection and fault identification). In this first stage, three approaches (i.e. statistical analysis, FFT and VMD) are separately applied to extract the fault feature information from multi-domain aspect (e.g. time-domain, frequency-domain and time-frequency domain), which can excavate comprehensively the condition information and intrinsic property of the raw vibration signal. Secondly, Laplace score algorithm is introduced to select automatically the meaningful sensitive feature according to the importance of each feature, which is aimed at removing some redundant information and improving the calculation efficiency. Finally, particle swarm optimization-based support vector machine (PSO-SVM) classification model is employed to implement the identification of multiple fault condition of rolling bearing. Performance of the proposed method is evaluated on two experimental examples of rolling bearing fault diagnosis. Experimental results show that the proposed method achieves high diagnosis accuracy for different working conditions of rolling bearing and outperforms some traditional methods both mentioned in this paper and published in other literature.

A quantitative approach to design alternative evaluation based on data-driven performance prediction

Design alternative evaluation in the early stages of engineering design plays an important role in determining the success of new product development, as it influences considerably the subsequent design activities. However, existing approaches to design alternative evaluation are overly reliant on experts’ ambiguous and subjective judgments and qualitative descriptions. To reduce subjectivity and improve efficiency of the evaluation process, this paper proposes a quantitative evaluation approach through data-driven performance predictions. In this approach, the weights of performance characteristics are determined based on quantitative assessment of expert judgments, and the ranking of design alternatives is achieved by predicting performance values based on historical product design data. The experts’ subjective and often vague judgments are captured quantitatively through a rough number based Decision-Making Trial and Evaluation Laboratory (DEMATEL) method. In order to facilitate performance based quantitative ranking of alternatives at the early stages of design where no performance calculation is possible, a particle swarm optimization based support vector machine (PSO-SVM) is applied for historical data based performance prediction. The final ranking of alternatives given the predicted values of multiple performance characteristics is achieved through Višekriterijumska Optimizacija I kompromisno Rešenje (VIKOR). A case study is carried out to demonstrate the validity of the proposed approach.

An Improved Grey Wolf Optimization Strategy Enhanced SVM and Its Application in Predicting the Second Major

In order to develop a new and effective prediction system, the full potential of support vector machine (SVM) was explored by using an improved grey wolf optimization (GWO) strategy in this study. An improved GWO, IGWO, was first proposed to identify the most discriminative features for major prediction. In the proposed approach, particle swarm optimization (PSO) was firstly adopted to generate the diversified initial positions, and then GWO was used to update the current positions of population in the discrete searching space, thus getting the optimal feature subset for the better classification purpose based on SVM. The resultant methodology, IGWO-SVM, is rigorously examined based on the real-life data which includes a series of factors that influence the students’ final decision to choose the specific major. To validate the proposed method, other metaheuristic based SVM methods including GWO based SVM, genetic algorithm based SVM, and particle swarm optimization-based SVM were used for comparison in terms of classification accuracy, AUC (the area under the receiver operating characteristic (ROC) curve), sensitivity, and specificity. The experimental results demonstrate that the proposed approach can be regarded as a promising success with the excellent classification accuracy, AUC, sensitivity, and specificity of 87.36%, 0.8735, 85.37%, and 89.33%, respectively. Promisingly, the proposed methodology might serve as a new candidate of powerful tools for second major selection.

An improved SVM classifier based on double chains quantum genetic algorithm and its application in analogue circuit diagnosis

Support Vector Machine (SVM) classifier is widely used in analogue circuit diagnosis. However, the penalty parameter C and the kernel parameter γ of SVM classifier with the radial basis function (RBF) affect the classification performance seriously. A double-chains-quantum-genetic-algorithm (DCQGA) based method is proposed to optimize C and γ. In DCQGA, each chromosome carries two gene chains, and each of gene chains represents an optimization solution, which can accelerate the search process and help to find the global solution. Thereafter, the optimal parameters C and γ are obtained by optimizing the parameter searching process with DCQGA. Two common datasets named Iris and Wine from UCI Machine Learning Repository are used to test the performance of the presented SVM classifier. The simulation results illustrate that the population's best fitness and the classifying accuracy of the proposed DCQGA-SVM are higher than that of the Particle-Swarm-Optimization based SVM (PSO-SVM), the Quantum Genetic Algorithm based SVM (QGA-SVM) and the classifier based on grid search method (GS-SVM). Finally, the proposed DCQGA-SVM is applied to analogue circuit diagnosis, a Sallen–Key bandpass filter circuit and a four-opamp biquad high-pass filter are chosen as circuits under test (CUT). Wavelet packet analysis is performed to extract the fault features before classifying. The experimental results show that the SVM parameters selected by DCQGA-SVM contribute to higher diagnosis accuracy than other methods referred in this paper.

An evolvable self-organizing neuro-fuzzy multilayered classifier with group method data handling and grammar-based bio-inspired supervisors for fault diagnosis of hydraulic systems

Purpose – The purpose of this paper is to apply a hybrid neuro-fuzzy paradigm called self-organizing neuro-fuzzy multilayered classifier (SONeFMUC) to classify the operating faults of a hydraulic system. The main motivation behind the use of SONeFMUC is to attest the capabilities of neuro-fuzzy classifier for handling the difficulties associated with fault diagnosis of hydraulic circuits. Design/methodology/approach – In the proposed methodology, first, the neuro-fuzzy nodes at each layer of the SONeFMUC are trained separately using two well-known bio-inspired algorithms, i.e. a semi deterministic method with random walks called co-variance matrix adaptation evolutionary strategy (CMA-ES) and a swarm-based explorer with adaptive fuzzified parameters (SBEAFP). Thereafter, a revised version of the group method data handling (GMDH) policy that uses the Darwinian concepts such as truncation selection and elitism is engaged to connect the nodes of different layers in an effective manner. Findings – Based on comparative numerical experiments, the authors conclude that integration of neuro-fuzzy method and bio-inspired supervisor results in a really powerful classification tool beneficial for uncertain environments. It is proved that the method outperforms some well-known classifiers such as support vector machine (SVM) and particle swarm optimization-based SVM (PSO-SVM). Besides, it is indicated that an efficient bio-inspired method can effectively adjust the constructive parameters of the multi-layered neuro-fuzzy classifier. For the case, it is observed that designing a fuzzy controller for PSO predisposes it to effectively balance the exploration/exploitation capabilities, and consequently optimize the structure of SONeFMUC. Originality/value – The originality of the paper can be considered from both numerical and practical points of view. The signals obtained through the data acquisition possess six different features in order for the hydraulic system to undergo four types of faults, i.e. cylinder fault, pump fault, valve leakage fault and rupture of the piping system. Besides, to elaborate on the authenticity and efficacy of the proposed method, its performance is compared with well-known rival techniques.

Recognition of concurrent control chart patterns using wavelet transform decomposition and multiclass support vector machines

Statistical process control charts have been widely utilized for monitoring process variation in many applications. Nonrandom patterns exhibited by control charts imply certain potential assignable causes that may deteriorate the process performance. Though some effective approaches to recognition of control chart patterns (CCPs) have been developed, most of them only focus on recognition and analysis of single patterns. A hybrid approach by integrating wavelet transform and improved particle swarm optimization-based support vector machine (P-SVM) for on-line recognition of concurrent CCPs is developed in this paper. A statistical correlation coefficient is used to determine whether the input pattern is a single or concurrent CCP. Based on wavelet transform, a raw concurrent pattern signal is decomposed into two basic pattern signals, which can be recognized by multiclass SVMs. The performance of the hybrid approach is evaluated by simulation experiments, and numerical and graphical results are provided to demonstrate that the proposed approach can perform effectively and efficiently in on-line CCP recognition task.

Chinese Grain Production Forecasting Method Based on Particle Swarm Optimization-based Support Vector Machine

Forecasting of grain production is an important resource for establishing agriculture policy. Particle swarm optimization-based support vector machine (PSO-SVM) is applied to forecast grain production in this paper. In PSOSVM model, particle swarm optimization (PSO) is used to determine free parameters of support vector machine. PSO is a new optimization method, which is motivated by social behavior of bird flocking or fish schooling. The optimization method not only has strong global search capability, but also is very easy to implement. The Chinese grain production is used to illustrate the performance of proposed PSO-SVM model. The experimental results indicate that the PSO-SVM method can achieve greater forecasting accuracy than grey model, artificial neural network in Chinese grain production forecasting. Consequently, PSO-SVM is a proper method in Chinese grain production forecasting.