4182 results about "Model prediction" patented technology

A nonlinear model-based predictive temperature control system is described for use in thermal process reactors. A multivariable temperature response is predicted using a nonlinear parameterized model of a thermal process reactor. The nonlinear parameterized model is implemented using a neural network. Predictions are made in an auto-regressive moving average fashion with a receding prediction horizon. Model predictions are incorporated into a control law for estimating the optimum future control strategy. The high-speed, predictive nature of the controller renders it advantageous in multivariable rapid thermal processing reactors where fast response and high temperature uniformity are needed.

A predictive modeling method may include determining a time interval of time-series data; identifying one or more variables of the data as targets; determining a forecast range and a skip range associated with a prediction problem represented by the data; generating training data and testing data from the time-series data; fitting a predictive model to the training data; and testing the fitted model on the testing data. The forecast range may indicate a duration of a period for which values of the targets are to be predicted. The skip range may indicate a temporal lag between the time period corresponding to the data used to make predictions and the time period corresponding to the predictions. The skip range may separate input data subsets representing model inputs from subsets representing model outputs, and separate test data subsets representing model inputs from subsets representing validation data.

A nonlinear model-based predictive temperature control system is described for use in thermal process reactors. A multivariable temperature response is predicted using a nonlinear parameterized model of a thermal process reactor. The nonlinear parameterized model is implemented using a neural network. Predictions are made in an auto-regressive moving average fashion with a receding prediction horizon. Model predictions are incorporated into a control law for estimating the optimum future control strategy. The high-speed, predictive nature of the controller renders it advantageous in multivariable rapid thermal processing reactors where fast response and high temperature uniformity are needed.

A method and system for optimizing post prandial blood glucose excursions for persons with diabetes via computation of time-varying optimal insulin profiles are disclosed. The method builds upon a mathematical model based on the physiologic description of the human glucoregulatory system. Model parameter values are specific for each patient and meal type, and are determined from measured data by minimizing differences between both measurements and model predictions, and which are used to provide the optimal insulin profiles that are useful to diabetics in managing their chronic disease.

Optimization algorithms and techniques to predict and simulate motion and various performance of a digital human model. The human body is modeled as a kinematics system represented by a series of segments connected by joints that represent musculoskeletal joints such as the wrist, elbow, shoulder, clavicle and pelvis. Optimization tools are used to determine the rotation at each degree of freedom of each joint that minimizes a performance measure.

The invention discloses an adaptive model prediction control (MPC) device, comprising a MPC control module controlling the controlled process, an online recognition module and a control detection module, wherein the MPC control module is connected with the controlled process, the online recognition module and the control detection module, the online recognition module is connected with the control detection module and the controlled process, the control detection module is connected with the controlled process. When the MPC control module works, the online recognition module and the MPC control module operate together to automatically effectively complete the steps for the application of the MPC control module. When the MPC control module works online, the control detection module continuously detects the property of MPC and the quality of the model. When the control property is worse and the model has low quality, the control detection module starts the online recognition module to automatically maintain the MPC control module.

Disclosed are a model prediction control method and a model prediction control system for all working conditions of a wind generating set. The system comprises an MPC (model prediction control) device, a feedback information measurer, a wind wheel, a driving chain, a tower, a generating unit, a variable propeller driver and a converter. The feedback information measurer is used for detecting status variables of the wind wheel, the driving chain, the tower and the generating unit and transmitting detecting results to the MPC device, the MPC device is used for computing targets of the blade pitch angle and the generator torque, and the variable propeller driver and the converter are used for adjusting the blade pitch angle and the wind generator torque. The method is used for computing control increment by means of a variable propeller control prediction model and a torque control prediction model, takes the status variables including driving chain torsional displacement, driving chain torsional speed, blade plane external first-order flap displacement, blade plane external first-order flap speed, tower front-back first-order swing displacement, tower front-back first-order swing speed, mechanical loads of the unit and the like, and two prediction models can be automatically switched in different working conditions, so that the wind generating set can be operated in all working conditions.

The invention discloses a multi-gesture and cross-age oriented face image authetication method. The method comprises the following steps of: rapidly detecting a face, performing key point positioning, performing face alignment, performing non-face area filtration, extracting the face features by blocks, performing feature dimension reduction and performing model prediction. The method provided by the invention can perform the face alignment, realize the automatic remediation for a multi-gesture face image, and improves the accuracy rate of the algorithm, furthermore, the feature extraction and dimension reduction modules provided by the invnetion have robustness for aging changes of the face, thus having high use value.

The invention belongs to the field of power converter control and relates to a model prediction control method of a voltage source type rectifier when a network voltage is unbalanced. The model prediction control method of the voltage source type rectifier when the network voltage is unbalanced comprises the steps of converting three-phase network voltage and a three-phase current into network voltage and a current which are under a two-phase static coordinate system through an abc / alpha beta coordinate, obtaining network positive sequence voltage and negative sequence voltage by carrying positive and negative sequence component separation of the network voltage under the two-phase static coordinate system, calculating reference values of active power, calculating current reference values of the rectifier under the two-phase static coordinate system, and carrying out model prediction current control and selecting an optimal on-off state. The model prediction control method of the voltage source type rectifier when the network voltage is unbalanced is capable of quickly and precisely tracing the current under the two-phase static coordinate and effectively eliminating fluctuation of direct current side voltage at the same time, power quality of the system is improved, a reliable network-connected operation of the rectifier when transient state network voltage is unbalance is achieved, and dynamics and stability of the whole system are improved.

The invention relates to a horizontal and vertical coordination control method for trajectory tracking of an intelligent vehicle. For association and coupling characteristics of horizontal and vertical dynamics of the intelligent vehicle, a horizontal and vertical coordination controller for trajectory tracking of the intelligent vehicle is designed. By applying a model prediction control and sliding mode control algorithm, the opening degree of a throttle valve of an engine, the pressure of a main braking cylinder and the deflection angle of a front wheel are cooperatively controlled. In thedesign of a horizontal MPC, the state amount of the vehicle is selected at the formula which is shown in the description, and the state amount (vertical speed vx) is real-time changable vehicle speedoutput by the vehicle after vertical control; vy is a horizontal speed at the mass center of the vehicle; the formulas which are shown in the description are the heading angle and heading angle speedat the mass center of the vehicle; Y and X are a horizontal position and a vertical position under world coordinates. According to a horizontal and vertical coordination control system, the intelligent vehicle efficiently and stably tracks an expectation trajectory at the expected speed. Large-steering operation can be remarkably improved, and the horizontal stability of the intelligent vehicle inthe trajectory tracking process is improved.

The invention discloses a vehicle stability integrated control method based on a variable-weight model prediction algorithm. The method is mainly aimed at yawing motion and side-tipping motion of a vehicle to improve the stability and the riding comfort of the vehicle. The method mainly comprises steps as follows: step one, vehicle driving state information is acquired and subjected to estimation processing; step two, reference state values for keeping the yawing stability and the side-tipping stability of the vehicle are decided respectively; step three, a variable-weight model prediction control algorithm is used, a front wheel rotating angle and tire braking force are taken as control variables, and vehicle yawing and side-tipping stability control are integrated; step four, expected tire braking force is compared with actual pressure of a brake tube, an action instruction of an electromagnetic valve is determined, an additional front wheel rotation angle and steering action of a driver are superposed, and a final front wheel rotation angle is obtained; step five, a brake executer and a steering executer execute the action instruction of the electromagnetic valve and a front wheel rotation angle instruction, so that the vehicle keeps the yawing stability and side-tipping stability.

The invention relates to an intelligent electric vehicle path tracking model prediction control method, belonging to the technical field of control. The aim of the invention is to adopt the intelligent electric vehicle path tracking model prediction control method, wherein the model prediction control method can take account of the safety constraints of a whole vehicle simultaneously and realizeseclectic optimization among vehicle path tracking performance, safety and whole vehicle performance effectively. The method gives thought to intelligent electric vehicle path tracking model building in limited conditions and yaw stabilization controller design based on model prediction control as well as the tracking performance of the vehicle, the safety of the vehicle, the whole vehicle performance, driving comfort and saving and control of energy in the process of carrying out control strategy deduction, and improves the dynamic performance of the whole vehicle. The method gives thought tothe tracking performance of the vehicle (path following and speed following), the safety of the vehicle (preventing slipping, locking, sideward inclining or drifting), the whole vehicle performance (accelerating and braking performance), driving comfort (the change of torque cannot be too great) and saving and control of energy (saving energy on the premise of meeting the performance). The dynamicperformance of the whole vehicle is improved.

Method for evaluating an underground reservoir production scheme taking account of uncertainties.Flow simulator input parameters characterizing the reservoir and the production scheme are selected. An approximate analytical model allowing the reservoir responses to be predicted is constructed. A desired degree of accuracy Dp is defined, this degree of accuracy Dp measuring the difference between the responses of the model and those of the simulator. The degree of accuracy Dp(M) of the predictions of the model is calculated. A design of experiments is constructed so as to select simulations to be performed, pertinent for adjustment of the model. The simulations selected by the design of experiments are carried out, then, for each response simulated by the simulator, the analytical model is adjusted by means of an approximation method. This operation is repeated until the desired degree of accuracy Dp is reached. Finally, the production scheme is evaluated by analyzing the reservoir responses predicted by the approximate analytical model.Application: notably to the development of petroleum reservoirs for example.

The invention discloses a system for intelligent management on the basis of the prediction of future road traffic conditions (speed and flow). The system comprises a data fusion processing module, a mathematic model prediction computation module and a traffic guidance management module based on prediction. The system creates standardized information input by implementing data fusion on the diversified traffic information collected in real-time, judges, by means of historic data, interconnection of road network traffic flows to establish a parameterized statistical model and further computes the traffic conditions (such as speed and flow) of each road in the road network in the coming 90 minutes, wherein, the time precision is consistent with the time interval of the real-time data collection. Then the system further judges and creates quantifiable traffic indicators in accordance with the prediction result, comprising: the traffic jam probability of each road in the future, the time the jammed roads are clear and traffic flow abnormal alarms, thus providing policy support for traffic participants and managers.

The invention discloses a model prediction control-based active and reactive coordinated control method of a power distribution network. The method comprises the steps of (1) building a recent optimization and control model according to an optimization object and constraint conditions; (2) solving the recent optimization and control model and making a recent active and reactive control plant of the power distribution network; (3) taking a recent active and reactive control plan of the power distribution network as a reference value, considering current and future running constraint conditions on the basis of a running state measurement value of the power distribution network at a current moment, building a prediction model and an active and reactive coordinated optimization and control model at an intra-day rolling correction stage by and solving an intra-day active and reactive control instruction sequence in a limited period in the future; and (4) executing an intra-day active and reactive control instruction at a first moment, moving a time window backwards at a time interval and repeating the intra-day rolling correction optimization process. According to the method, the running cost of a system is reduced on the premise of ensuring the running security of the system, the network loss of the system is reduced, and maximization of the running benefit of the active power distribution network is achieved.

A method, system, and computer program product is provided, for example, for predicting a location of a road sign. In an example embodiment, the method may include receiving a set of pre-processed road observations and extracting a plurality of features from the set of pre-processed road observations, wherein the plurality of features comprise at least a plurality of sensor based features and a plurality of map based features. Further, the method may include associating a set of sensor based features from the plurality of sensor based features and a set of map based features from the plurality of map based features with at least one of a plurality of ground truth points in ground truth data. Additionally, the method may include training a machine learning model based on the association of the ground truth data with the set of sensor based features and the set of map based features with at least one of a plurality of ground truth points in the ground truth data and further predicting the location of the road sign based on the trained machine learning model.

Methods for predicting states of a subject are presented. For example, a method for predicting states of a subject includes obtaining training data comprising a plurality of variables, obtaining training states associated with the training data, and forming a predictive model according to the training data and the training states, the predictive model predictive of the training states. The forming of the predictive model includes extracting one or more hidden components from the training data. The extracting of the one or more hidden components includes regression analysis including determining one or more relationships between the one or more hidden components and the plurality of variables, and determining one or more relationships between the one or more hidden components and the training states. A number of the one or more hidden components is less than a number of the plurality of variables and greater than a number of the training states.

The invention discloses an automotive adaptive cruise control method taking multiple targets into consideration. A layer control strategy is utilized, upper layer control can decide an expected longitudinal acceleration according to a target vehicle and a current state of a controlled vehicle, and lower layer control can track the expected longitudinal acceleration through a reverse recursion method. The automotive adaptive cruise control method comprises the following steps: a mutual longitudinal dynamitic model between two vehicles is established, a design model predicating controller can obtain the expected distance between two vehicles according to a constant time headway strategy, a model prediction control algorithm is utilized to decide an expected longitudinal acceleration for tracking the expected distance between the two vehicles, vehicle control work conditions are divided into a driving work condition and a braking work condition, reverse longitudinal dynamic models for the two work conditions are established respectively according to a vehicle driving equation, an expected throttle percentage is obtained according to the vehicle reverse longitudinal dynamic model and the expected acceleration in the driving work condition, and an expected brake pedal openness is obtained according to the expected acceleration in the braking work condition.

The invention discloses a model prediction based cut tobacco dryer outlet moisture control method. Aiming at characteristics of complicated state changes of a cut tobacco dryer during working and diversity of production process modes, an intelligent integrated optimizing control system for the cut tobacco dryer based on an intelligent prediction model and an artificial intelligent operating mode is constructed so as to achieve comprehensive optimization and automation during cut tobacco drying. Aiming at different stages and different production process modes in production, a model capable of describing dynamic process characteristics depending on feed quantity and feed moisture is constructed. On the basis of the model, an on-line optimizing control algorithm which is capable of simultaneously or selectively adjusting multiple process variables, adapting to changes of the feed quantity and the feed moisture and overcoming mutual interference among the variables and influences of various uncertainties during cut tobacco drying and has self-adaptive and self-adjustment functions is designed, and strict requirements on outlet cut tobacco moisture in different working conditions can be met.

The invention discloses a human-vehicle cooperative steering control method considering real-time allocation of driving rights. The method solves the problem of driving right allocation between drivers and vehicle automatic driving controllers in a human-vehicle cooperative driving process. The method comprises the specific steps of 1, establishing a vehicle dynamics model and a vehicle kinematicsmodel; 2, establishing a vehicle automatic driving controller; 3, establishing a human-vehicle co-driving system model; 4, adopting a model prediction method for designing a human-vehicle co-drivingsystem controller; 5, allocating the driving rights, calculating a control amount, executing the control amount, and achieving the process of controlling a vehicle to steer through the cooperation ofa driver and the vehicle automatic driving controller. By means of the method, under the condition that the vehicle automatic driving controller and the driver drive the vehicle together, the allocation of the steering driving rights of the vehicle automatic driving controller and the driver can be optimized in real time online, and the operation of steering the vehicle is cooperatively completedby the vehicle automatic driving controller and the driver.

The invention belongs to the control field of electric power electron converters and relates to a model prediction control method for a three-level voltage-source-type converter. The model prediction control method for the three-level voltage-source-type converter comprises the following steps: step one, building models of relations between currents of the alternating current side of the three-level voltage-source-type converter and switch functions; step two, building models of relations between capacitance voltage offsets of the direct current side and switch functions; step three, obtaining current instruction values through outer voltage loops and obtaining a current instruction value of the next moment by calculating current instruction values of several moments before; step four, setting value functions of a model prediction control algorithm for the three-level voltage-source-type converter; step five, calculating on-off state subsets; step six, achieving real-time prediction control. The model prediction control method for the three-level voltage-source-type converter is simple in algorithm, easy to achieve and capable of being adopted in a higher-level converter, and has generality.

A method and system for optimizing post prandial blood glucose excursions for persons with diabetes via computation of time-varying optimal insulin profiles are disclosed. The method builds upon a mathematical model based on the physiologic description of the human glucoregulatory system. Model parameter values are specific for each patient and meal type, and are determined from measured data by minimizing differences between both measurements and model predictions, and which are used to provide the optimal insulin profiles that are useful to diabetics in managing their chronic disease.

A method and an apparatus for dynamically adjusting a sampling rate relating to wafer examination. A process step is performed upon a plurality of workpieces associated with a lot. A sample rate for acquiring metrology data relating to at least one of the processed workpiece is determined. A dynamic sampling rate adjustment process is performed to adaptively modify the sample rate. The dynamic sampling rate adjustment process includes comparing a predicted process outcome and an actual process outcome and modifying the sampling rate based upon the comparison.

The embodiment of the invention discloses a training method of a machine learning model, an exception prediction method and a related device. The method includes: processing The training data of the T-M time to the T time by the trained first machine learning model to obtain the predicted data of the T+N time;; Determining credible abnormity according to the prediction data at the T + N moment andthe real data at the T + N moment, and processing The training data of the T-M time to the T time by using the second machine learning model to obtain a prediction abnormality,; and according to theerror of the prediction abnormity degree and the credible abnormity degree, transmitting the error back to the second machine learning model, adjusting the current model parameters of the second machine learning model, and achieving unsupervised learning of the second machine learning model capable of predicting the data abnormity at the future moment. Moreover, the relationship between continuousdata is considered in training of the training data based on the time period, and the accuracy of model prediction can be improved.

The invention discloses an unsupervised anomaly prediction method for two-stage cloud server, which is used for solving the anomaly prediction problem of a cloud server environment. The method comprises a prediction stage and an anomaly detection stage, the prediction stage is used for training a many-to-many time sequence prediction model according to preprocessed historical cloud server key performance index data, and the model is used for predicting cloud server KPIs data at the future moment; in the anomaly detection stage, a multivariable anomaly detection model is trained according to preprocessed historical cloud server KPIs data, anomaly detection is conducted on the predicted KPIs data at the future moment through the model, the anomaly probability of data points at the future moment is obtained, finally, an anomaly probability threshold value is set, and the data points with the anomaly probability threshold value larger than the threshold value are regarded as abnormal datapoints. Otherwise, the data points are normal data points, and an anomaly prediction result is obtained. The invention has the advantages of being independent of label data, wider in applicability andexcellent in performance.

The invention provides a refrigeration system with a compressing unit, and a method of controlling a refrigeration system. To facilitate a better control, the capacity of the compressing unit is controlled based on a predicted future cooling demand rather than an actually determined cooling demand. The invention further provides a system wherein a cost value for changing the cooling capacity of the system is taken into consideration.

The invention discloses a training method of an SMPL parameter prediction model, a server and a storage medium. The method comprises the steps of obtaining a sample picture; inputting the sample picture into an attitude parameter prediction model to obtain an attitude prediction parameter; inputting the sample picture into a morphological parameter prediction model to obtain a morphological prediction parameter; according to the posture prediction parameters and the form prediction parameters, constructing a human body three-dimensional model through an SMPL model; calculating model predictionloss according to the SMPL prediction parameters and / or the human body three-dimensional model in combination with the annotation information of the sample picture; and reversely training an attitudeparameter prediction model and a morphological parameter prediction model according to the model prediction loss. In the embodiment of the invention, the sample picture is directly used as model input for model training, and a model for extracting the human body information in the picture does not need to be independently trained, so that the complexity of model training is reduced, and the efficiency of model training is improved.

The invention discloses a keyword extraction method based on a Seq2seq framework. The method comprises: creating a sequence model by utilizing a Seq2seq framework; introducing an attention mechanism,extracting features of keywords in the text; fusing a pointer network model and a Cover mechanism at a decoding end to improve the attention distribution of potential keywords; then using a softmax loss function to train a network model, and finally in a model prediction stage, using a Beam Search cluster search algorithm to generate a key word sequence with the maximum probability to serve as a key word result set to obtain appropriate key words. According to the method, deep semantics in the long text are well considered, the word distribution probability is calculated in combination with the context information context, the problem of repetition of low-frequency words and generative tasks is better solved, and the keyword extraction accuracy is improved.

The invention discloses an unmanned aerial vehicle autonomous formation evaluation index and model prediction control (MPC) formation control method and belongs to the technical field of flight control. According to the method, a definition relevant to unmanned aerial vehicle autonomous formation, a computing method of the collision prevention probability of unmanned aerial vehicle autonomous formation and an unmanned aerial vehicle autonomous formation evaluation index are provided, and the MPC formation control method applicable to unmanned aerial vehicles is provided. The method solves the problem how to give out the collision prevention quantitative index which should be followed by design of an autonomous formation flight control system when it is known that the unmanned aerial vehicle autonomous formation scale is M, the average expectation distance between unmanned aerial vehicles in formation is dr, and the collision probability of the whole formation is required to be smaller than Pc. The method provides a quantitative index basis for engineering design of the unmanned aerial vehicle autonomous formation control system, and the collision problem the unmanned aerial vehicles in dense formation is solved.

The invention discloses an energy consumption model prediction method for a battery electric vehicle based on road information and driving style optimization. The energy consumption model prediction method comprises the steps that a vehicle sensor, geographic information software, an electronic map and a weather forecasting system are utilized to acquire vehicle state parameters, road informationparameters and environmental information parameters; according to the acquired parameters, parameter estimation is performed on rolling resistance coefficients, air density and road gradient parameters; a working condition prediction model based on road information and driving style optimization is established to predict the working conditions to enable energy consumption of the predicted workingconditions to be accurately approximated to energy consumption of actual working conditions; and a battery electric vehicle energy consumption prediction model is established for energy consumption prediction, specifically, a battery electric vehicle energy consumption calculation model is established based on a battery electric vehicle performance test, parameter estimation results and working condition prediction results are used as the input of the battery electric vehicle energy consumption calculation model to form the battery electric vehicle energy consumption prediction model, the battery electric vehicle energy consumption prediction model outputs predicted energy consumption, and energy consumption of future path information is predicted.