185 results about "Relevance vector machine" patented technology

The invention provides a universal circuit breaker mechanical fault diagnosis method based on feature fusion of vibration and sound signals. The method includes steps of 1, collecting machine vibration signals and machine sound signals during an engaging and disengaging process of a universal circuit breaker; 2, adopting an improved wavelet packet threshold value denoising algorithm for denoising; 3, adopting a complementary total average empirical mode decomposition algorithm for extracting a plurality of solid mode function components reflecting state information of engagement and disengagement actions of the circuit breaker from the denoising signals; 4, determining the number Z of the solid mode function components; 5, calculating the energy ratio, the sample ratio and the power spectrum entropy as three types of features; 6, adopting a combination core principal component analysis method for performing dimension reduction on a feature sample with unified three types of features of the vibration and the sound signals and obtaining M principle components; 7, establishing a related vector machine based sequence binary tree multiple classifier model.

A variational Relevance Vector Machine (RVM) is disclosed. The RVM is a probabilistic basis model. Sparsity is achieved through a Bayesian treatment, where a prior is introduced over the weights governed by a set of what are referred to as hyperparameters—one such hyperparameter associated with each weight. An approximation to the joint posterior distribution over weights and hyperparameters is iteratively estimated from the data. The posterior distribution of many of the weights is sharply peaked around zero, in practice. The variational RVM utilizes a variational approach to solve the model, in particular using product approximations to obtain the posterior distribution.

The invention discloses the technical field of wind power prediction, particularly, relates to an ultrashort-term power prediction method. The method comprises the following steps of: firstly, acquiring the wind speed, the wind direction and the wind power of a wind power farm to form a sample set; then, preprocessing the data of the sample set; reducing dimensions of the preprocessed sample set by a depth autocoder network; and finally, training a relevance vector machine regression model by the sample set with reduced dimensions, and predicting the ultrashort-term wind power through the trained relevance vector machine regression model. The method reduces the training time of a prediction model, satisfies the requirements on precision and real-time property in system status estimation, and enables the prediction method to be more accurate.

The invention discloses a network anomaly traffic monitoring method, belonging to the technical filed of information security. The network anomaly traffic monitoring method comprises the steps of: capturing a network data stream passing by; according to generation time of the network data stream, selecting n network data streaming data which are closest to the current time, wherein n is determined according to the calculating capacity of a system; training the captured n network data streaming data as input of a relevance vector machine, establishing a data model; and monitoring the current network traffic data according to the data model. By the method and the device, the classifying monitoring precision can be improved so that anomaly traffic is monitored more rapidly and effectively, and lower false drop rate and fallout ratio are ensured.

The invention discloses a wind power range short-term prediction method based on variation mode decomposition and a relevant vector machine. The method comprises: variation mode decomposition is carried out on a wind power sequence to obtain a plurality of components having different center frequencies; all components are processed by using a relevant vector machine algorithm to establish range prediction models respectively; and prediction results of all components are superposed to obtain a general range prediction result under a certain confidence level. With the method, the prediction precision and range coverage rate of the model are improved; and the range width is narrowed obviously, so that the short-term prediction result of the wind power range is improved obviously.

The invention provides a satellite lithium ion battery residual life prediction system and a satellite lithium ion battery residual life prediction method based on RVM (relevance vector machine) dynamic reconfiguration, and relates to a lithium ion battery residual life prediction system and a lithium ion battery residual life prediction method. The uncertainty expression of the lithium ion battery predication is realized, and the lithium ion battery residual life prediction method is more applicable to satellite system environment with limited resources. A dynamic reconfiguration module of the prediction system comprises a reconfiguration unit A and a reconfiguration unit B, the reconfiguration unit A and the reconfiguration unit B realize the time sharing multiplex of logic resources of the dynamic reconfiguration module, and the RVM training and predication is realized; and the Gaussian kernel function flowing water calculation is realized by a multistage flowing water segmented linear proximity method and a parallel computing structure, and the computational efficiency is enabled to be fully improved. The inverse calculation of symmetric positive definite matrices is realized by a Cholesky decomposition method, the computing resources consumption is reduced by a multiplying and gradually decreasing device, and the computing delay is reduced. Experiments show that the system and the method have the advantages that FPGA (field programmable gate array) finite computing resources are utilized for realizing the computational accuracy similar to a PC (personal computer) platform, the four-times computing efficiency improvement relative to the PC platform is obtained, and the utilization rate of hardware resources is effectively improved through dynamic reconfiguration strategies.

The invention provides a water quality parameter time series prediction method based on relevance vector machine regression. The water quality parameter time series prediction method comprises the following steps of 1 acquiring water quality parameter historical data from an automatic water quality monitoring station and performing data pre-processing; 2 using front 2 / 3 data in the pro-processed water quality parameter historical data as a training sample set and using rear 1 / 3 data as a testing sample set; 3 using the training sample set to train an RVM, using the testing sample set to test the trained RVM so as to obtain a water quality parameter time series prediction model based on the RVM regression; 4 using the water quality parameter time series prediction model based on the RVM regression to predict new water quality parameters. The water quality parameter time series prediction method can perform time series prediction, is large in prediction range, high in accuracy and good in prediction stability, and can provide probabilistic output, give a predicted confidence interval while performing prediction, reduce the prediction time and timely observe water quality parameter change.

The invention discloses an online lithium ion battery residual life predicting method based on relevance vector regression, belongs to the technical field of lithium ion battery life prediction, and solves the problem that the residual life of the existing lithium ion battery is predicted by an offline method with low precision. The method comprises the following steps: firstly selecting original samples, performing phase-space reconstruction to construct a training sample set; initializing the model parameters of RVM (relevance vector machine); performing RVM training to obtain a RVM prediction model; comparing the obtained prediction value with ynew, if yes, the constructed novel training set WS equal to WSUINS, retraining RVM, and updating the RVM prediction model; otherwise, keeping the RVM prediction model stable; performing recurrence prediction until the prediction value is smaller than the invalid threshold value U, and finishing the online prediction of the residual life of the predicted lithium ion battery. The method is suitable for prediction of the lithium ion battery residual life.

A method of measuring a location of a user equipment (UE) located indoors in a wireless network includes receiving signals from a plurality of access points (APs), performing training for machine learning using the received signals or information acquired from the received signals, setting a weight vector to be applied to a relevance vector machine (RVM) method using data subjected to the training for machine learning, and applying RVM regression to the set weight vector and measured strengths of the received signals and determining whether the signals received from the plurality of APs are line of sight (LOS) signals or non line of sight (NLOS) signals.

The invention provides an opening-closing fault diagnosis method for an air circuit breaker based on vibration signals, wherein an acceleration sensor is used to collect machine body vibration signals generated during opening-closing courses of the air circuit breaker. The method comprises the steps that firstly, the acceleration sensor is used to collect the machine body vibration signals generated during opening-closing actions of the air circuit breaker and transform the vibration signals into digital signals, so that initial vibration signals are obtained; secondly, an improved wavelet packet threshold de-noising algorithm is used to process the collected vibration signals; thirdly, a complementary ensemble-average empirical mode decomposition algorithm is used to extract intrinsic mode function components from the de-noising vibration signals; fourthly, the quantity Z of the intrinsic mode function components is determined; fifthly, the intrinsic mode function components of the first Z orders are selected and extracted as sample entropies of a characteristic quantity; sixthly, binary tree multi-classifiers based on a relevance vector machine are established; and seventhly, the binary tree multi-classifiers based on the relevance vector machine obtained at the sixth step are used to establish a fault recognition model of the air circuit breaker.

The invention belongs to the technical field of circuit-breaker failure diagnosis, and particularly relates to a method and device for diagnosing mechanical characteristic failures of a high-voltage circuit-breaker. The device comprises the circuit-breaker and further comprises a vibration sensor, a voltage conditioning element, an AD conversion element, a clock element, a power element, a central processing unit, a communication unit and a failure diagnosis upper computer. According to mechanical vibration signals in the motion process of the circuit-breaker, the vibration sensor, the voltage conditioning element, the AD conversion element, the clock element, the power element, the central processing unit, the communication unit and the failure diagnosis upper computer are utilized for achieving the mechanical characteristic failure diagnosis of the circuit-breaker. The method for diagnosing mechanical characteristic failures comprises the steps of conducting wavelet packet decomposition on vibration signals in the operation process of the high-voltage circuit-breaker, extracting characteristic vectors of the vibration signals in spectral entropy of each frequency band, and adopting a relevance vector machine algorithm to conduct failure diagnosis on the mechanical characteristics of the high-voltage circuit-breaker. The method and device can effectively diagnoses the mechanical characteristic failures of the circuit-breaker, and provide a basis for the state maintenance of the circuit-breaker.

The invention provides a relevance vector machine-based multi-class data classifying method, which mainly solves the problem that the traditional multi-class data classifying method cannot integrally solve classifying face parameters and needs proximate calculation. The relevance vector machine-based multi-class data classifying method comprises a realizing process comprising the following steps of: partitioning a plurality of multi-class data sets and carrying out a normalizing pretreatment; determining a kernel function type and kernel parameters; setting basic parameters; calculating the classifying face parameters; calculating lower bounds of logarithms and solving variant values of the lower bounds of the logarithms and adding 1 to an iterative number; if the variant values of the lower bounds of the logarithms are converged or the iterative number reaches iterating times, finishing updating the classifying face parameters, and otherwise, continuing to updating; and obtaining a prediction probability matrix according to the updated classifying face parameters, wherein column numbers corresponding to a maximum value of each row of the matrix compose classifying classes for testing the data sets, and samples which have the prediction probability less than a false-alarm probability and the detection probability corresponding to a false-alarm probability value set in a curve are rejected. The relevance vector machine-based multi-class data classifying method has the advantages of obtaining classification which is comparable to that of an SVM (Support Vector Machine) by using less relevant vectors and rejecting performance and can be used for target recognition.

The invention provides a method for Bayes compressed sensing signal recovery based on a self-adaptive measurement matrix and relates to the field of the information and communication technology. The method aims at solving the problem that an existing compressed sensing signal recovery method is low in accuracy. Based on the design of the self-adaptive measurement matrix in compressed sensing and combined with the Bayes compressed sensing algorithm, a design scheme of the compressed sensing method is obtained. The method is characterized in that the designed measurement matrix can be generated in a self-adaptive mode according to different signals, the purposes of determinacy and storage of the matrix are both achieved, and combined with the Bayes compressed sensing recovery algorithm of a relevant vector machine, the priority of a layered structure is introduced. The design scheme passes simulation verification, it is confirmed that the good signal recovery effect can be obtained, and the error range of recovered signals can be evaluated. The method is suitable for wireless signal transmission occasions in the information and communication technology.

The invention relates to a circuit breaker failure diagnosis method based on a circuit breaker dynamic property test instrument, which comprises the following steps: collecting arbitrary sample signals from the test instrument; converting the collected signals to digital signals through digital-analog conversion; shaping and filtering the digital signals to form finishing signals W(t); extracting the wavelet characteristic entropy of the finishing signals W(t), and inputting the wavelet characteristic entropy into a relevance vector machine model to obtain the posterior probability of a corresponding relevance vector RVM; and by adopting the strategy of maximum probability win (MPW), attributing failure to the sort of signals having maximum posterior probability. The invention has the following advantages: wavelet decomposition is carried out on the collected signals to extract the wavelet characteristic entropy as a characteristic value, and the characteristic value is input to a failure diagnosis model established according to a relevance vector machine principle for diagnosis; by adopting the posterior probability diagnosis method, power equipment can be monitored in time; and the calculation amount of the kernel function is greatly reduced, and the diagnosis efficiency and accuracy are improved.

The invention relates to a lithium ion battery cycle life prediction technology, in particular, a deep belief network and relevance vector machine fusion-based lithium battery residual life prediction method. An existing lithium battery residual life prediction method relies on accurate physical models or complex signal processing technologies, as a result, the existing lithium battery residual life prediction method needs heavy investment, or the existing method is based on a shallow structures, as a result, the performance of fault prediction will be limited, and the existing method is vulnerable to curse of dimensionality, while with the method of the invention adopted, the problems of the existing method can be solved. A charging and discharging period-based lithium battery capacity degradation data set is obtained; the data are pre-processed; the fusion models of a DBN (deep belief network) and an RVM (relevance vector machine) are built; a DBN model and a RVM model are trained; and the trained fusion models of the DBN and the RVM is adopted to predict the residual life of a lithium battery. The method of the invention is suitable for predicting the residual life of the lithium battery.

The invention discloses an analogue circuit failure prediction method which comprises the following steps: performing Monte Carlo analysis on various elements of an analogue circuit in a failure-free section and extracting various frequency band signal energy, normalizing the extracted frequency band signal energy to obtain a feature vector; training a BP neural network; judging failure modes with occurrence trends, extracting a failure prediction feature vector when the element is at the initial value, extracting the failure prediction feature vector when a detected circuit is in work, computing the cosine angle distance to represent the health degree of the element, computing the health degree threshold value when the element is in failure, and optimally selecting a kernel function width factor of a relevance vector machine algorithm, and performing the failure prediction on the analogue circuit. The method can be used for a real-time system, and can be further used for a non-real-time system, a failure of the linear analogue circuit can be predicted, and a failure of the non-linear analogue circuit can be predicted, and failures of main elements such as resistor, inductor and the capacitor in the analogue circuit can be predicted.

The invention relates to the computer vision technology field, and provides a multi-human face tracking method based on the relevant vector machine. The method which can improve the studying quality comprises the following steps that: initialization detection is carried out to a scene, and the detected human face is constructed with a human face motion model and a color model, which are stored into a human face model database; at the same time, the state of the detected human face is initialized, and then is recorded into a human face state database; during the multi-human face tracking process, different tracking methods are adopted according to the different states of the human face, and detection is carried out to the tracking result, so as to change the state information of the human face according to the detection result; during the tracking process, a whole image searching is carried out once a plurality of frames, so as to detect the human face which is failed in being tracked and the new human face which enters into the scene. The invention requires no manual intervention, and can simultaneously detect and track random multi-human faces at a quick operating speed, thereby satisfying the real-time processing requirement.

The invention provides a pre-stack seismic data retrieval method and system. The pre-stack seismic data retrieval method includes: acquiring pre-stack seismic data of the current stratum; selecting an advantageous frequency band range from the pre-stack seismic data; extracting angle gather seismic data and seismic wavelets from the advantageous frequency band range; determining angle gather reflection coefficient according to a convolution model, the angle gather seismic data and the seismic wavelets; determining hyper-parameters according to relevant vector machines and the angle gather reflection coefficient; determining a super-resolution sparse pulse reflection coefficient according to the hyper-parameters, the angle gather reflection coefficient and the Bayes criterion; comprehensively analyzing the super-resolution sparse pulse reflection coefficient to obtain layer distribution of the current stratum. By the pre-stack seismic data retrieval method and system, resolution and accuracy of retrieval results are improved, and accuracy in identifying stratum relation and property of the thin layer and the thin-interbed lithologic oil-gas reservoir is improved.

The invention discloses a high-precision wind electric field power interval forecasting method based on a relevance vector machine, which comprises the steps of collecting data, carrying out normalization on the data, and selecting a training sample of a relevance vector machine forecasting model; optimizing parameters of the relevance vector machine forecasting model to obtain an optimized iteration initial value of kernel function width and the relevance vector machine forecasting model; obtaining a kernel function, and then obtaining relevance vector machine forecasting model parameters after convergence; and finally obtaining a forecasting value and a variance of a wind electric field, so as to obtain a forecasted interval of wind electric field power. The method can improve adaptability of the model, improve forecasting accuracy, reduce training sample size and reduce training time.

The invention discloses a method for predicting faults of a power electronic circuit based on an FRM-RVM (fuzzy rough membership-relevant vector machine), and the method comprises the following steps: monitoring voltage and current signals, and carrying out wavelet threshold denoising on the signals so as to form multidimensional circuit parameter vectors; carrying out dimensionality reduction on the multidimensional circuit parameter vectors so as to obtain multidimensional fault feature vectors; obtaining a fault feature vector sample set within a health tolerance range of the circuit; obtaining a fault feature vector of the circuit in the process of real-time operation at a periodic interval; computing the health degree of the fault feature vector to the fault feature vector sample set at each time point so as to form a health degree-time sequence of the circuit; giving out the threshold value of the health degree of the circuit; carrying out prediction on the health degree-time sequence of the circuit by using an RVM (Relevance Vector Machine) algorithm so as to obtain the health degree of the circuit in some future time, comparing the obtained health degree with the threshold value of the health degree, and determining the health situation of the circuit in some future time, thereby realizing the fault prediction of the circuit. By using the prediction method disclosed by the invention, the real-time state monitoring and health-status estimation on the power electronic circuit can be realized, thereby realizing the prediction on the future state of the circuit, and then predicting the time of fault occurrence in advance.

The invention discloses a health management decision-making method suitable for a complex process system. The method comprises the steps of 1, acquiring a system measuring point parameter; 2, pre-processing the system measuring point parameter; 3, performing real-time fault detection on the pre-processed system measuring point parameter through an adaptive threshold analysis method, and performing fault diagnosis on the fault detection result by using a knowledge reasoning method in combination with historical data, a fault mode and an influence analysis table; 4, evaluating the health degrees of single-parameter sensors by combining the fault diagnosis result and using a grey theory method, and fusing the health degrees of the single-parameter sensors by using a fuzzy set fusion theory to obtain a fault mode health parameter; 5, predicting a fault mode health parameter by using the obtained fault mode health parameter through a relevance vector machine method; and 6, fusing maintenance decisions of multiple decision theories by using a grey group decision-making theory to obtain a maintenance decision result. The method can be used for fault prediction for next operation of the system, and provides a maintenance advice for each fault mode.

The invention discloses a rotor system fault diagnosis method and device based on vibration analysis. A sensor acquires normal conditions of a rotor system and vibration signals under fault conditions; the acquired vibration signals are decomposed by an improved inherent time scale decomposition method to generate a plurality of rotational components and residual signals; related rotational components capable of reflecting fault information are selected from the rotational components; energy of each related rotational component is calculated; related vector machine multi-classification models are built by an improved directed acyclic method; fault characteristics are inputted to the related vector machine multi-classification models for training and fault diagnosis. A motor, a first bearing block, a second bearing block and a third bearing block are arranged on a test bed base, the first bearing block, the second bearing block and the third bearing block respectively support a first rotating shaft and a second rotating shaft which are sequentially connected with an output shaft of the motor, both the first rotating shaft and the second rotating shaft are provided with a disk, and a sensor group is arranged at the end of the second rotating shaft. Rotor system fault types can be rapidly and accurately recognized, and the method and the device are applicable to online diagnosis of the rotor system.

The invention provides a method for predicting the remaining service life of a lithium battery based on wavelet denoising and a relevant vector machine and relates to a method for estimating the health condition of the lithium battery and predicting the remaining service life of the lithium battery. The method comprises the following steps of measuring health condition data of the lithium battery along with charging and discharging periods, carrying out secondary wavelet denoising on measured lithium battery capacity data; calculating a capacity threshold that lithium battery loses effect; carrying out optimization selection on a width factor of a relevant vector machine algorithm through a differential evolutionary algorithm based on a lithium battery capacity data sequence and a charging and discharging period data sequence; predicting the remaining service life of the lithium battery through the relevant vector machine algorithm optimized by the differential evolutionary algorithm. The method for predicting the remaining service life of the lithium battery based on wavelet denoising and the relevant vector machine is easy to operate and effective, and the remaining service life of the lithium battery can be accurately predicted.

The invention discloses a support vector machine (SVM) based prediction method for the degradation tendency of super-capacitor capacitance value. The prediction method utilizes the regression function of the support vector machine to predict the degradation tendency of the super capacitor capacitance value and comprises: 1) pre-processing the input value and the output value; 2) carrying out trainings to the training set data for a regression estimation function; 3) using the particle swarm optimization algorithm to automatically optimize the relevant parameters of the support vector machine; 4) according to the optimization result, configuring the corresponding parameter values of the support vector machine; substituting the training set data into a correlation vector machine model to obtain a regression prediction model for the degradation tendency of the capacitance value; and 5) substituting the training set data into the regression prediction model to obtain the degradation tendency of the capacitance value. According to the invention, it is possible to conduct online prediction to the degradation tendency of the capacitance value. Through the introduction of a particle swarm optimization algorithm to modify the parameter optimization method, the prediction efficiency and accuracy of the algorithm are increased so that it can be applied in a larger scope.

The invention provides a combination-kernel-function RVM hyperspectral classification method integrated with multi-scale morphological characteristics. The method comprises the steps that 1) dimensions of hyperspectral images are reduced via principal component transform; 2) spatial characteristics of the hyperspectral images after principal component transform are extracted via mathematical morphological transform; 3) according to theories of the combination kernel function, combination kernel functions based on addition, multiplication and weighted addition are respectively established, and spectral and spatial characteristic of the images are integrated by means of a combination kernel function method; and 4) the hyperspectral images are classified via a combination-kernel-function RVM classifier, and classification experiments are carried out on the hyperspectral images via an AVIRIS (Airborne Visible Infrared Imaging Spectrometer). Compared with a traditional RVM classifier based on spectral characteristics, the classification precision of the combination-kernel-function RVM is greatly increased without substantial increase of training time; and the method of the invention is strongly stable and is not sensitive to the number of samples.

The invention relates to a roller bearing service life predicting model of a self-adaptive multi-kernel combination relevance vector machine. The roller bearing service life predicting model comprises the following steps of utilizing particle filter to initialize a combination kernel function weighting matrix, so as to obtain a combination kernel function set; establishing a multi-kernel combination relevance vector machine set; performing iteration predicting, weight updating and re-sampling self-adaption to obtain an optimal multi-kernel combination relevance vector machine model; finally, predicting the running state and remaining life of the rolling bearing. The roller bearing service life predicting model has the advantages that the excellent characteristics of a plurality of single kernel functions are adaptively integrated, the reliance of the single kernel function relevance vector machine model on the parameters is reduced, the predicting accuracy is improved, the predicting stability is better, the robustness of the model is higher, and the engineering application value is higher.

The invention belongs to the technical field of power system disaster warning, and particularly relates to a power transmission line icing prediction method based on a relevance vector machine. According to the prediction method, according to features of an icing phenomenon, the input quantity and weight index of an icing prediction model are selected and processed in a targeted mode; a power transmission line icing prediction model is built through a relevance vector machine method; training is conducted on the model through sample data, and the model is optimized by the adoption of a quantum particle swarm optimization and a K-fold cross-validation method; the icing thickness and probability distribution of a power transmission line are predicted according to the test data, and correction is further conducted on the model through repeated training to improve prediction accuracy. The prediction method considers the influences, on power transmission line icing, of various factors comprehensively, can predict the icing thickness of the power transmission line precisely and has very high prediction accuracy and generalization ability.

The invention discloses a FastRVM(fast relevance vector machine) wastewater treatment fault diagnostic method. The method includes the following steps that: 1) samples with incomplete properties in samples to be recognized in wastewater data are removed, since the dimensions of the properties of the samples are different, the samples are normalized to an interval of [0, 1]; 2) based on a clustering fast relevance vector machine, the majority of types of data are compressed; 3) the synthetic minority over samplingtechnique is adopted to expand the minority of types of data; 4) a "one-to-one" fast relevance vector machine multi-classification model is established; and 5) fast relevance vector machine wastewater fault diagnosis modeling is carried out. According to the FastRVM wastewater treatment fault diagnosis method of the invention, the majority of types of data are compressed based on the clustering fast relevance vector machine, and the minority of types of data are expanded through the synthetic minority over sampling technique, and therefore, the imbalance of wastewater data can be decreased; and the fast RVM is adopted to establish a multi-classification model for a wastewater biochemical treatment process, and therefore, the accuracy of fault diagnosis on a wastewater biological wastewater treatment system can be effectively improved.

The invention discloses an intelligent early warning method for dam safety monitoring data. The method comprises the steps of early warning model establishment, threshold value setting and mutual feedback type early warning. Gross error identification and gross error processing are carried out, model sample data quality is improved, according to the monitoring items, independent variable relevance, historical monitoring data quantity and historical monitoring data distribution, different early warning models and indexes are established, including a stepwise regression model, a correlation vector machine model and a gray system model; the established models can reflect the relationship between the independent variable and the dependent variable more truly and are wide in application range,according to a measuring instrument, measuring point attributes, a threshold value, an early warning model and indexes, real-time early warning is carried out on monitoring data, monitoring instrumentabnormity early warning is sent to monitoring personnel, or dam safety early warning is sent to dam safety management personnel, experts with professional knowledge and rich experience are not needed, the workload is small, the early warning speed is high, and the early warning result is more accurate and reliable.

The present invention discloses a sewage water quality detection method and an apparatus, wherein data training is performed to pre-establish a prediction model based on a relevance vector machine, and the total nitrogen and the total phosphorus in sewage are predicted by using the prediction model. Compared to the manual detection method, the method and the apparatus have the following characteristics that: online prediction can be achieved, real-time monitoring and regulation are easily performed, and contribution is provided for real-time monitoring automation of sewage water quality. In addition, the prediction model adopted by the method and the apparatus is a soft measurement method based on the relevance vector machine, and has better applicability and higher prediction accuracy compared with a model established by adopting a neural network and support vector machine modeling method.