831 results about "Gaussian process" patented technology

The invention provides a fuzzy curve analysis based soft sensor modeling method using time difference Gaussian process regression, it is suitable for application in chemical process with time delay characteristics. This method can extract stable delay information from the historical database of process and introduce more relevant modeling data sequence to the dominant variable sequence. First of all, the method of fuzzy curve analysis (FCA) can intuitively judge the importance of the input sequence to the output sequence, estimate the time-delay parameters of process, and such offline time-delay parameter set can be utilized to restructure the modeling data. For the new input data, based on the historical variable value before a certain time, the current dominant value can be predicted by time difference Gaussian Process Regression (TDGPR) model. This method does not encounter the problem of model updating and can effectively track the drift between input and output data. Compared with steady-state modeling methods, this invention can achieve more accurate predictions of the key variable, thus improving product quality and reducing production costs.

The invention relates to a condition monitoring data stream anomaly detection method, in particular to a condition monitoring data stream anomaly detection method based on an improved gaussian process regression model. The problem that an existing method for processing monitoring data stream anomaly detection is poor in effect is solved. The method comprises the steps that firstly, the historical data sliding window size is determined; secondly, the types of a mean value function and a covariance function are determined; thirdly, the hyper-parameter initial value is set to be the random number from 0 to 1; fourthly, q data closest to the current time t are extracted; fifthly, the gaussian process regression model is determined; sixthly, prediction is conducted by means of the nature of the gaussian process regression model; seventhly, PI of normal data at the time t+1; eighthly, monitoring data are compared with the PI; ninthly, whether the real monitoring data need to be marked to be abnormal or not is judged; tenthly, beta (xt+1) corresponding to the monitoring value at the time t+1 is calculated; eleventhly, the real value or prediction value and the t+1 are added into DT; twelfthly, new DT is created. The condition monitoring data stream anomaly detection method based on the improved gaussian process regression model is applied in the field of network communication.

The invention discloses a mobile robot autonomous cruise method for reliable WIFI connection. The method comprises the following steps that 1, a robot traverses the whole environment through autonomous exploration navigation, and a WIFI two-dimensional distribution probabilistic model is built according to WIFI signal intensity data of limited measuring points at the access position by means of a Gaussian process regression model; 2, environmental grid maps are built at the same time and mixed with WIFI signal intensity distribution to generate a mixed map, namely a WIFI map; 3, the built WIFI map is utilized for carrying out obstacle avoidance navigation, and therefore it is ensured that the path through which a robot passes bypasses the WIFI signal weak area while the optimal path obstacle avoidance navigation is achieved. The WIFI signal distribution of the whole indoor room can be estimated only based on the data of the limited WIFI signal intensity measuring points through a machine learning algorithm, and the method is applicable to application occasions with high requirements for real-time wireless network connection in remote mobile robot cruise monitoring.

An improved system and method is provided for sparse Gaussian process regression using predictive measures. A Gaussian process regressor model may be construction by interleaving basis vector set selection and hyper-parameter optimization until the chosen predictive measure stabilizes. One of various LOO-CV based predictive measures may be used to find an optimal set of active basis vectors for building a sparse Gaussian process regression model by sequentially adding basis vectors selected using a chosen predictive measure. In a given iteration, a predictive measure is computed for each of the basis vectors in a candidate set of basis vectors and the basis vector with the best predictive measure is selected. The iterative addition of basis vectors may stop when predictive performance of the model degrades or no significant performance improvement is seen.

The invention belongs to the field of lithium ion batteries, and discloses an online SOH estimation method of a lithium ion battery for solving the problems that characteristic parameters are difficult to be obtained online, the dependency of a model on training data is high, the required data size is large, the complex function relationship between the battery capacity and the characteristic parameters is difficult to be described by simple linear regression, and the estimation accuracy is difficult to be guaranteed in an implementation process of the existing SOH estimation technology. According to the online SOH estimation method disclosed by the invention, the characteristic parameters are obtained from a capacity increment curve by using a capacity increment method. The method does not require the battery to undergo a complete charging and discharging process, the feature parameter extraction is simpler, and the application of the method in the BMS is facilitated. The establishment of a characteristic parameter and SOH function model is completed by using a multi-output Gaussian process regression model method, the potential correlation between different outputs is better used, and the estimation accuracy of SOH is improved. Meanwhile, the dependency of the method on the training data is small, and the online SOH estimation method has very good adaptability on different types of lithium ion batteries.

A method for modelling a dataset includes a training phase, wherein the dataset is applied to a non-stationary Gaussian process kernel in order to optimize the values of a set of hyperparameters associated with the Gaussian process kernel, and an evaluation phase in which the dataset and Gaussian process kernel with optimized hyperparameters are used to generate model data. The evaluation phase includes a nearest neighbour selection step. The method may include generating a model at a selected resolution.

The invention discloses a Gaussian process regression-based method for predicting state of health (SOH) of lithium batteries, relates to a method for predicting the SOH of the lithium batteries, belongs to the fields of electrochemistry and analytic chemistry and aims at the problem that the traditional lithium batteries are bad in health condition prediction adaptability. The method provided by the invention is realized according to the following steps of: I. drawing a relation curve of the SOH of a lithium battery and a charge-discharge period; II, selecting a covariance function according to a degenerated curve with a regeneration phenomenon and a constraint condition; III, carrying out iteration according to a conjugate gradient method, then determining the optimal value of a hyper-parameter and bringing initial value thereof into prior distribution; IV, obtaining posterior distribution according to the prior part; V, obtaining the mean value and variance of predicted output f' without Gaussian white noise; and VI, together bringing the practically predicted SOH of the battery and the predicted SOH obtained in the step V into training data y to obtain the f', then determining the prediction confidence interval and predicting the SOH of the lithium battery. The method provided by the invention is used for detecting lithium batteries.

The invention discloses a robot under-actuated hand autonomous grasping method based on stereoscopic vision, and relates to a robot autonomous grasping method. The problems that a grasping point can not be calculated through an existing robot grasping method until a three-dimensional model of an object is obtained in advance and the existing robot grasping method can only recognize a simple object and can not obtain a corresponding grasping point for a complicated object are solved. The method includes the steps of obtaining RGB-D point cloud of the object and the environment through a Kinect sensor and conducting filtering on the point cloud for a to-be-grasped object and the environment of the object; extracting normal vector included angle characteristics, coplanar characteristics, distance characteristics, grasping stability characteristics, collision detecting characteristics and corresponding constraint equations for the RGB-D point cloud; establishing a grasping planning scheme on the basis of Gaussian process classification; driving an under-actuated hand for grasping according to the grasping scheme, then judging whether the under-actuated hand has already grasped the object or not according to current detection till the under-actuated hand grasps the object, and releasing the object after completing the grasping task. The method is suitable for the field of robot grasping.

The present invention discloses a short-term load prediction method based on variable selection and Gaussian process regression. The method includes the following steps that: 1) bad data elimination, supplementation and normalization pre-processing are performed on sample data; 2) candidate input variables are selected from the perspectives of historical load, temperature and humidity, and the date type of a prediction date, and the scores of the importance of the variables are calculated through a random forest algorithm, and the scores of the importance of the variables are sequenced; 3) an optimal variable set is determined through adopting a sequence forward search strategy and based on a Gaussian process regression model; 4) the Gaussian process regression model is trained based on the determined optimal variable set, and the parameters of the model are optimized based on improved particle swarm optimization; and 5) the predictive performance of the model is verified in a test set. With the method provided by the invention adopted, prediction accuracy can be effectively improved, and the load prediction problem of a power system can be solved.

The invention discloses a Gaussian process modeling based wind turbine shafting state monitoring method in the field of wind turbine state monitoring. The technical scheme includes: collecting values of normal temperature of a bearing to be monitored and of correlated variables of the bearing temperature from historical data of a wind turbine SCADA system to form a bearing temperature vector set; building a bearing temperature model by the aid of a Gaussian process regression method; using the bearing temperature model for monitoring the bearing in real time, and using difference between the measured bearing temperature and the predicated temperature outputted by the model as predicated model residual; comparing the predicated model residual with a set residual threshold, and when the predicated model residual is larger than the residual threshold, judging the bearing to be abnormal; and otherwise, judging the bearing to be in a normal state. The method has the advantages that under the operation conditions of random changing of wind speed and time varying of rotating speed of a wind turbine shafting, states of bearings on the wind turbine shafting are analyzed and judged accurately, bearing fault alarm is sent timely, and maintenance complexity and cost are lowered.

A method of the present invention using a prediction process including a battery equivalent circuit model used to predict a voltage and a state of charge of a battery. The equivalent circuit battery model includes different equivalent circuit models consisting of at least an ideal DC power source, internal resistance, and an arbitrary number of representative parallel resistors and capacitors. These parameters are obtained a priori by fitting the equivalent circuit model to battery testing data. The present invention further uses a correction process includes determining a corrected predicted state of charge of the battery; and storing the corrected state of charge of the battery in a storage medium. In the present invention, an expectation of the predicted voltage of the battery and an expectation of the predicted state of charge of the battery are obtained by an unscented transform with sigma points selected by a Gaussian process optimization.

The invention discloses a power system short-term load probability forecasting method, a device and a system. The short-term load probability density forecasting model of Gaussian process quantile regression is established by selecting an optimal input variable set affecting the load. Firstly, the importance score of input variables is given by stochastic forest algorithm, and the influence degreeof each input variable is sorted. Secondly, particle swarm optimization algorithm is used to search the super-parameters of the model to form the optimal Gaussian process quantile regression prediction model, avoiding the adverse effect of artificial experience setting initial parameters on the prediction performance of the model. The invention can avoid the shortcomings of manual experience selection, the load forecasting model established in the optimal input variable set has low error, which further reduces the forecasting error, and overcomes the problems that the common conjugate gradient method is easy to fall into the local optimal solution, the iterative number is difficult to determine, and the optimization performance is greatly affected by the initial value selection, so that the self-searching and group cognitive ability can be brought into full play.

Described herein is a method and system for characterizing in-ground rock types from measurement-while-drilling data in a mining environment. The method includes the steps of drilling holes at a plurality of selected locations within a region of interest; collecting measurements while drilling to obtain an array of data samples (162) indicative of rock hardness at various drilling depths in the drill hole locations; obtaining a characteristic measure (163) of the array of data samples; performing Gaussian Process regression (164) on the characteristic measure; and applying boundary detection (166) to the rock hardness output data obtained from the Gaussian process model to identify the distribution (280) of at least one cluster of rock type within the region of interest.

The invention discloses an unmanned vehicle urban intersection left-turn decision-making method based on conflict resolution. The method comprises the steps of track prediction for straight vehicles at an intersection, decision-making process selection of a behavior decision-making module corresponding to different scenes, and vehicle control parameter selection corresponding to an action selection module. According to the invention, the decision-making framework of the left turn of the unmanned vehicle at the intersection is divided into environment assessment, behavior decision-making and action selection; prediction of intersection straight driving motion tracks is realized by using a Gaussian process regression model, decision-making processes under different left-turn scenes are formulated, and an unmanned vehicle driving action selection method considering multiple factors is provided; and the decision-making process of the left turn of the unmanned vehicle at the intersection isstructured and clarified, so that the reasonability and the adaptability of the decision-making model are improved.

The invention discloses a JITL (just-in-time learning) based multi-model fusion modeling method adopting GPR (Gaussian process regression). The method is used for a complex and changeful multi-stage chemical process and is a multi-model strategy which is continuously updated online; a Gaussian mixture model is adopted to identify different stages of the process, and a self-adaptive learning method is adopted to continuously update an established GPR model; when new data arrive, partially similar data are selected based on Euclidean distance and angle principle at each stage and used for establishing a partial GPR model; finally, new data obtained through calculation belong to posterior probability of each stage, and the partial model is subjected to fusion output. According to the method, key variables can be predicated accurately, so that the product quality is improved, and the production cost is reduced.

The invention belongs to the technical field of automatic vehicle driving environment cognition and decision-making, and especially relates to a vehicle trajectory predicting method based on hybrid dynamic bayesian networks and gaussian process. According to the method, parameters of MDBN and GP are learned through natural vehicle driving data, and a plurality of vehicle kinematic models are combined through utilizing MDBN, so that short-term trajectory prediction and estimated probabilities of driving intention and driving characteristics are obtained, and then long-term trajectory predictionand representation of uncertainty prediction are conducted through using GP. By adopting the method, short-term prediction characteristics based on a vehicle physical movement model as well as long-term trajectory prediction and representation of uncertainty prediction according to driver information can both taken into account. Compared to an existing vehicle trajectory predicting method, vehicle models, abstract intention and data driving are combined together, and the expansibility of the MDBN model and the GP model are strong, and thus the method is suitable for different driving scenarios and can combine more effective situational information like road information and traffic information.

The invention relates to a method for calculating the expected car following distance in driver car following behavior analysis. On the basis that drivers design multiple different driving scenes during data collection on a car surrounding simulation testing platform, driving data is collected aiming at the drivers; the car following behavior feature parameter data set of each driver is extracted from the driving data; the feature behavior parameter data sets of all the drivers are clustered into several different driving classifications, and are adopted as training data sets; then the training data sets are utilized for carrying out classification judgment on the current drivers to be classified. Different car following behaviors are classified, the efficiency of collecting car following behavior data is improved, the cost is low, and good safety is achieved. The longitudinal driving behaviors of the drivers can be simulated through the Gussian process, the individualized expected car following distance can be provided for the current drivers, and the active adaptive capacity of an auxiliary driving system to different drivers is improved.

The invention relates to a monitoring and locating method based on an abnormal electricity consumption detection module of the deep noise reduction self-coding network-Gaussian process. Electricity consumption and meter event information of all the detected users in a transformer area is inputted to the abnormal electricity consumption detection module of the deep noise reduction self-coding network-Gaussian process, the features of the data are extracted from a time-frequency domain and classified, and the suspected abnormal electricity consumption users of the detected users are selected through screening by the model. The abnormal electricity consumption detection module outputs the suspected abnormal degree coefficient and orders the probability of the suspected abnormal degree of the users so as to obtain a suspected abnormal electricity consumption user list. The multiplatform electricity consumption data are analyzed by combining the artificial intelligence field leading-edge technology, the hidden user electricity consumption behavior mode in the mass data is deeply mined and the suspected abnormal electricity consumption users are located so that abnormal electricity consumption detection is enabled to be more intelligent and more efficient.

The invention discloses a gaussian process regression method for predicting a network security situation in the technical field of network information security. According to the invnetion, a hierarchical network security situation evaluation index system is structured by using an analytic hierarchy process; the damage degree of various network security threats to the network security situation is analyzed by the system so as to calculate a network security situation value of each time monitoring point and structure a time sequence and then structure into a training sample set; the training sample set is subjected to iterative training by utilizing gaussian process regression so as to obtain a prediction model meeting an error requirement; an optimal training parameter of the gaussian process regression is dynamically searched by utilizing an particle swarm optimization in the training process so as to reduce a prediction error, and finally the prediction of the network security situation value of the time monitoring point in the future is finished by utilizing the prediction mode. The gaussian process regression method provided by the invnetion has the beneficial effects of better adaptability and lower prediction error in the respect of reducing the prediction error of the network security situation.

The invention provides an SOC online measurement method for a lithium battery based on the Gaussian mixture process and the dynamic OCV correction. According to the invention, the Gaussian mixture regression (GMR) process is integrated with a Gaussian mixture model and a Gaussian process regression model, so that the time series of dynamic non-linearity can be effectively represented. The dynamic OCV correction method can calibrate an OCV-SOC curve according to external factors, so that the accurate OCV is obtained. Therefore, the SOC is corrected, and the accumulative error is eliminated. In this way, a battery model can be updated in real time according to the appropriate algorithm difficulty under the complex working condition of an automobile. Meanwhile, battery characteristics can be accurately tracked and the accumulated estimation error is corrected. The long-term precision is guaranteed.

The invention relates to an Internet of Things data uncertainty measurement, prediction and outlier-removing method based on the Gaussian process. The method is a dynamical system method of estimating and collecting the standard deviation of Internet of Things perception sensor measurement errors and combining the Gaussian process modeling theory with autoregression model representations; prediction values and uncertainty measurement of observation data effective time sequence data are given, whether the data are missing values or outlier data is judged according to the information, and data supplement is correspondingly carried out. The method is a non-parameterized probability prediction method. Due to the fact that training set learning has the feature of tracing system dynamic states, judgment, early-warning and data supplement can be carried out on data exception and data missing phenomena in time according to the prediction value uncertainty and the sensor calibration standard deviation, the prediction error is small, and the accuracy is high. The Internet of Things data uncertainty measurement, prediction and outlier-removing method is used for controlling the quality of Internet of Things automatic observation data, and can ensure accuracy of collected data.

The invention discloses a humanoid robot gait control method based on model correlated reinforcement learning. The method comprises steps of 1) defining a reinforcement learning framework for a stable control task in forward and backward movements of a humanoid robot; 2) carrying out gait control of the humanoid robot with a model correlated reinforcement learning method based on the sparse online Gaussian process; and 3) improving a motion selection method of a reinforcement learning humanoid robot controller by a PID controller, and taking the improved operation as an optimizing initial point for the PID controller obtaining the motion selection operation of the reinforcement learning controller. The invention utilizes reinforcement learning to control gaits of the humanoid robot in movement, and thus the movement control of the humanoid robot can be automatically adjusted via interaction with the environment, a better control effect is achieved, and the humanoid robot is enabled to be stable in forward and backward directions.

The invention discloses a method for predicting a bearing fault based on Gaussian process regression. The method comprises the following five steps of: step 1, setting prediction system parameters, initializing a Gaussian process regression model; step 2, collecting a bearing vibration signal regularly, extracting characteristics of a vibration signal to obtain time domain characteristic parameters of the bearing vibration signal, and carrying out fault symptom judgment; step 3, judging whether a fault symptom exists; step 4, calculating and storing the characteristic parameters, and carrying out dynamic updating of the Gaussian process regression model; and step 5, predicting the fault of a bearing. According to an actual use condition of a product, small amount of data is collected, time that the product possibly has the fault is given quantificationally, a calculation speed and prediction accuracy are improved by using the Gaussian process regression, a whole life cycle of the bearing is divided into three time ranges, such as a health time range, a sub-health time range and a fault time range by use of an idea of health management, fault prediction is carried out in the sub-health state, usage management capacity of the bearing is improved.

The present invention discloses an auto building method and system of a Wi-Fi position fingerprint map. The method comprises the steps of acquiring crowdsourcing data, building the Wi-Fi position fingerprint map by using a pedestrian navigation reckoning method and a machine learning method, and screening according to an indoor map to obtain Wi-Fi strength information with a position tag; and using the Wi-Fi strength information with the position tag as a training sample of a Gaussian process, so as to obtain a function relationship between the signal strength and the position information, resolving a super parameter in the Gaussian process, and predicting according to the super parameter so as to obtain Wi-Fi strength information of a poor constraint area of the indoor map. Regression is performed by using the Gaussian process, the Wi-Fi position fingerprint of a wide area is predicted based on the Wi-Fi position fingerprint information of a high constraint area of the map, so that the Wi-Fi position fingerprint map of the whole indoor area is built automatically.

The invention provides a training method of multi-moving object action identification, comprising the following steps of: extracting the movement track information of each moving object from video data; layering the movement track information of the moving objects; modeling for the movement mode of the multi-moving object action on each layer; carrying out characteristic description on the model of the movement mode by synthesizing the overall and local movement information in a video, wherein the characteristic at least comprises a three-dimensional hyper-parameter vector for describing the movement track by using a gaussian process; and training a grader according to the characteristic. The invention also provides a multi-moving object action identification method which identifies the multi-moving object action in the video by utilizing the grader obtained by using the training method. In the invention, the movement track of an object is represented by using the gaussian process from a probability angle, and a model is established for a multi-people action mode from three granularity layers, and the characteristics are extracted, which makes the representation of the multi-people action more practical.

The invention provides a multi-model comprehensive prediction method for photovoltaic powder based on synchronous extrusion wavelet transformation. The multi-model comprehensive prediction method comprises the following steps: dividing photovoltaic historical data into four types including sunny day, cloudy day, rainy day and cloudy day according to different weather conditions; preprocessing eachtype of photovoltaic powder data by virtue of a synchronous extrusion wavelet transformation method, and decomposing the data into a series of modal functions with mutually exclusive characteristics;carrying out normalization processing on each modal function; determining an input variable set of each modal function; establishing a BP neural network, support vector machine and Gaussian process regression integrated multi-model comprehensive prediction method for each modal function; and overlapping prediction results of different modal functions, so as to obtain a final photovoltaic power short-term predicted value. According to the multi-model comprehensive prediction method, the prediction precision is effectively increased, the reliability of a prediction result is improved, and the problem of short-term prediction of the photovoltaic powder of a power system can be well solved.

The invention discloses a short-term wind speed prediction method of Gaussian process regression and particle filtering, thereby realizing on-line dynamic detection and correction of abnormal values and improving wind speed prediction accuracy. According to the method, an input variable set having the highest correlation with a wind speed value at a to-be-predicted time is determined by using a partial autocorrelation function, a state vector is determined, and a proper training sample set is constructed; a Gaussian-process-regression-based short-term wind speed prediction model is establishedin the training sample set and a fitting residue during the training process is given; on the basis of combination of the state vector and the Gaussian process regression model, a particle filteringstate space equation is established and state estimation is carried out on a current measurement value by using a particle filtering algorithm; and the estimation value and the measurement value residual of particle filtering are analyzed, determination is carried out based on a 3 sigma principle, and an abnormal value is corrected. According to the method provided by the invention, the abnormal value can be detected and corrected effectively; the short-term wind speed prediction precision is improved; and a wind speed prediction problem of the power system is solved.

The invention discloses an estimation method and device of a lithium battery health status, and a storage medium. The method comprises the following steps: serving multiple health indicators (HI) corresponding to multiple effective charging cycles as input variables, and multiple health statuses (SOH) corresponding to multiple effective charge cycles as output variables, and training to obtain a Gaussian process regression GPR model, wherein the GPR model is obtained by using multiple groups of data through the machine learning training, each group of data in multiple groups of data comprisesthe HI of each effective charge cycle and the SOH corresponding to the HI; acquiring the HI of the to-be-detected charge cycle, inputting the HI into a GPR model, and enabling the GPR model to outputthe SOH corresponding to the HI. The technical problem that the method for detecting the health status of the lithium battery is comparatively complex and hard to adapt to the collection data with badquality in the related technology is solved.

The invention discloses a gaussian process regression soft measurement modeling method based on an EGMM (Error Gaussian Mixture Model), which is used for a complex and changeable chemical process with non-gaussian noise. Prediction errors are frequently generated by a soft measurement prediction model established in an industrial process, however, the model prediction errors frequently contain rich useful information, and therefore, information can be extracted from the prediction errors so as to compensate the output of the model, thereby improving the established soft measurement model. Firstly, appropriate variables are selected to form error data, so as to be optimized to obtain appropriate number of gaussian components; then fitting is performed on the error data by using the EGMM; when new data arrive, prediction output is performed by using established GPR (Gaussian Process Regression), the mean conditional error is obtained through the EGMM, and the output is compensated, so as to obtain more accurate results. Key variables can be accurately predicted, thereby increasing the quality of products and reducing the production cost.