37 results about "Dominant factor" patented technology

The statistical analysis described and claimed is a predictive statistical tree model that overcomes several problems observed in prior statistical models and regression analyses, while ensuring greater accuracy and predictive capabilities. Although the claimed use of the predictive statistical tree model described herein is directed to the prediction of a disease in individuals, the claimed model can be used for a variety of applications including the prediction of disease states, susceptibility of disease states or any other biological state of interest, as well as other applicable non-biological states of interest. This model first screens genes to reduce noise, applies k-means correlation-based clustering targeting a large number of clusters, and then uses singular value decompositions (SVD) to extract the single dominant factor (principal component) from each cluster. This generates a statistically significant number of cluster-derived singular factors, that we refer to as metagenes, that characterize multiple patterns of expression of the genes across samples. The strategy aims to extract multiple such patterns while reducing dimension and smoothing out gene-specific noise through the aggregation within clusters. Formal predictive analysis then uses these metagenes in a Bayesian classification tree analysis. This generates multiple recursive partitions of the sample into subgroups (the “leaves” of the classification tree), and associates Bayesian predictive probabilities of outcomes with each subgroup. Overall predictions for an individual sample are then generated by averaging predictions, with appropriate weights, across many such tree models. The model includes the use of iterative out-of-sample, cross-validation predictions leaving each sample out of the data set one at a time, refitting the model from the remaining samples and using it to predict the hold-out case. This rigorously tests the predictive value of a model and mirrors the real-world prognostic context where prediction of new cases as they arise is the major goal.

The statistical analysis described and claimed is a predictive statistical tree model that overcomes several problems observed in prior statistical models and regression analyses, while ensuring greater accuracy and predictive capabilities. Although the claimed use of the predictive statistical tree model described herein is directed to the prediction of a disease in individuals, the claimed model can be used for a variety of applications including the prediction of disease states, susceptibility of disease states or any other biological state of interest, as well as other applicable non-biological states of interest. This model first screens genes to reduce noise, applies k-means correlation-based clustering targeting a large number of clusters, and then uses singular value decompositions (SVD) to extract the single dominant factor (principal component) from each cluster. This generates a statistically significant number of cluster-derived singular factors, that we refer to as metagenes, that characterize multiple patterns of expression of the genes across samples. The strategy aims to extract multiple such patterns while reducing dimension and smoothing out gene-specific noise through the aggregation within clusters. Formal predictive analysis then uses these metagenes in a Bayesian classification tree analysis. This generates multiple recursive partitions of the sample into subgroups (the “leaves” of the classification tree), and associates Bayesian predictive probabilities of outcomes with each subgroup. Overall predictions for an individual sample are then generated by averaging predictions, with appropriate weights, across many such tree models. The model includes the use of iterative out-of-sample, cross-validation predictions leaving each sample out of the data set one at a time, refitting the model from the remaining samples and using it to predict the hold-out case. This rigorously tests the predictive value of a model and mirrors the real-world prognostic context where prediction of new cases as they arise is the major goal.

The invention discloses a finite-element analysis, monitoring and support method of soft-rock slope stability. The method includes the following steps: step one, obtaining rock slope failure types, slope stability influence factors and slope reinforcement support measures by induction, establishing a mechanical model of engineering geology simulation analysis under the action of the different influence factors, and analyzing a dominant factor controlling the soft-rock slope stability and a potential mode thereof; and step two, using an approximate equivalent physical model, which is formed bya plurality of mutually associated unit bodies, by a finite-element analysis method of the slope stability to replace actual structures or continua, and establishing equations, which characterize force and displacement relationships, through basic principles of structure and continuum mechanics and physical characteristics of units. The finite-element analysis, monitoring and support method of thesoft-rock slope stability uses the finite-element method to analyze the stability thereof, carries out support and treatment on places where danger may further occur, and has certain guiding significance for solving slope stability problems.

The invention discloses a blasting vibration control prediction method based on a fault tree and a genetic algorithm, which collects all factors influencing the blasting vibration design and construction process in the history as fault tree events and constructs a blasting vibration fault tree. Identification of dominant factors; Genetic algorithm is used to optimize the initial weights and network structure of the neural network to replace the general neural network initial weights and the random selection and blind selection of the network structure. Then the neural network algorithm is usedto fine-tune the network weights and search the optimal solution or the optimal approximate solution in the network structure; constructing a BP artificial neural network model; predicted results: within the control range (1, 0, 0, 0); reaching the critical value (0, 1, 0, 0); exceeding the control value (0, 0, 1, 0); seriously exceeds control value (0, 0, 0, 1). The invention controls the occurrence of the blasting vibration accident from the source, and effectively reduces the harmful effect caused by the blasting vibration. The rationality and scientificity of blasting design are improved.

The invention discloses a human-robot formation control method applicable to a human-aimed navigation system. The navigation system comprises a mobile robot and a portable wearable device based on vibro tactile feedback. A laser radar scans a surrounding environment, the distance information with a surrounding object is transmitted to an upper computer, and the upper computer processes the distance information to obtain a two-dimensional environment map; according to the map information, the upper computer uses software to carry out optimal path planning in real time, and a planning speed is generated; an RGB-D camera collects the skeleton information of a user at a certain sampling frequency and transmits the information to the upper computer, and after processing, the upper computer obtains the position and speed information of the user; a human-robot formation control algorithm takes the user as a dominant factor, the speed of the user and the planning speed are integrated, and theactual control speed of the robot is generated; and according to the position information of the user, the upper computer instructs the user to adjust the self position through a vibration device anda voice prompting device, a certain relative distance and a relative angle are kept between the robot and the user, and the function of human-aimed navigation is realized.

The invention, which belongs to the technical field of the power system and automation thereof, discloses a load-shedding control method with consideration of system voltage stability after photovoltaic access, so that the adverse influence on high-voltage grid voltage recovery after a fault by large-scale photovoltaic access to the grid is eliminated. An equivalent local grid static voltage stability index of an external network is calculated to determine a voltage instability situation; a dominant factor of voltage instability is identified based on a characteristic change of reactive power absorbing by a photovoltaic aggregation station after the fault; and then load-shedding control strategies are formulated on the condition of using a photovoltaic factor as the dominant instability factor and using a cross-section factor as the dominant instability factor, thereby realizing load-shedding control of the voltage weak node load. Therefore, the power system scheduling operation staff can obtain the inherent operation rule of the photovoltaic access system and the foundation is laid for formulating a load-shedding optimization strategy cooperatively, so that the effectiveness and automation level of secure and stable control of a large grid are improved.

The present invention provides a method and a system for optimizing primary frequency modulation (FM), and a terminal device. The method for optimizing the primary FM comprises: obtaining a unit system of the primary frequency failure period and an operating state of the unit system; obtaining the unit speed, and the measured parameters of the measured model of the primary FM, wherein the measuredparameters comprise the characteristic parameter value of the steam turbine model, the PI controller parameter value, the rotational speed difference, the load delay parameter value, the FM instruction delay value, the PI controller output rate limit value, and the pressure pullback loop parameter value; inputting the different values of the measured parameters in the preset range and the unit speed to the measured model of the primary FM to obtain the simulation result; and optimizing the primary FM according to the primary FM performance reflection function and the simulation result. According to the technical scheme of the present invention, by using the simulation means to determine the dominant factors and optimize the control strategy, the blindness of the conventional method is avoided, the labor intensity and the test risk are reduced, the coupling influence can be solved, the key parameters can be determined, and the pertinence of the primary FM analysis optimization can be improved.

The invention discloses a missing traffic data repairing method based on Bayesian enhanced tensor decomposition. The method comprises the following steps of organizing the road network vehicle speed data into a third-order data tensor to introduce a dominant factor structure for modeling; inputting the data tensor and indicating the tensor; updating posterior distribution of a global parameter mu;updating posterior distribution of a hyper-parameter; updating posterior distribution of a bias parameter phi and posterior distribution of a factor matrix parameter U until i is equal to m; updatingposterior distribution of a bias parameter theta and posterior distribution of a factor matrix parameter V until j is equal to n; updating posterior distribution of a bias parameter eta and posteriordistribution of a factor matrix parameter X until t is equal to f; repeating the steps S5-S9 until the difference deltatau between the accuracy parameter tau and the previous parameter tau is less than epsilon, converging the model and proceeding to the next step; substituting the updated {mu, phi, theta, eta, U, V, X} parameter values into the expressions of yijt to calculate and estimate the tensor

The invention relates to a method for estimating suspended matter concentration in turbid lake water based on a VIIRS sensor. The method comprises: based on lake field measured spectrum and substance concentration data, analyzing optical properties of water body and water quality dominant factors; screening wave bands which are sensitive to suspended matter concentration variation in the turbid lake water, using curve fitting to determine an optimal model of suspended matter concentration and ground measured spectrum (Rrs); using Rrc data which is removed with water vapor and ozone absorption and corrected by Rayleigh scattering to establish a suspended matter concentration quantitative model which is suitable to be used for the VIIRS; and finally, using the established model on a VIIRS image, to accurately obtain interannual and inter-monthly variation rules and spatial distribution of the suspended matter concentration of the turbid lake water. Using the method can realize long-term high-precision monitoring of suspended matter concentration of a lake, and the method is helpful to scientifically evaluate interannual suspended matter concentration variation and development trends thereof, and effectively evaluate ecological environment changes of lake water.

The invention discloses a boiler characteristic calibration method based on operating data. The method comprises construction of a steady state model of a boiler system to determine a minimum measuredvariable parameter set, data acquisition, data preprocessing, calibration calculation, and the like. Through the actual operating characteristic parameters under different operating conditions, the dominant factors of boiler operating characteristic parameters can be analyzed, and a dominant factor model of the characteristic parameters can be established. The method overcomes the uncertainty ofthe conventional calculation method based on design data, and the use of a large amount of operating data can obtain the actual operating characteristics of a boiler more precisely and describe the actual operating state and performance of the system more accurately.

The invention discloses a multi-advantage factor double-person shared remote operation method, which can effectively reduce the operation error caused by the operator's muscle tremor, operational proficiency, and psychological factors through the double-person shared control method, and the control accuracy is more accurate and reliable. Sex is higher. Secondly, by designing multiple advantageous factors, the present invention can freely adjust the transparency and motion followability of the operating system, and the transparency and motion tracking effects under the optimal conditions are better than the traditional two-person shared control operation. Finally, the present invention provides system stability conditions through Raisbeck passive stability criteria, and then calculates system parameters from the stability conditions to ensure absolute stability of the system.

Belonging to the technical field of measuring and testing, the invention relates to ultra-high cycle fatigue of an ultrasonic fatigue test simulation material, and a high temperature alloy ultra-high cycle fatigue damage simulation method adopting a microanalysis technique for studying micromechanical properties and material deformation to conduct damage assessment. In the invention, a neutron diffraction method is employed to test high temperature alloy funnel-shaped sample surface and subsurface residual stress absolute difference, thus obtaining a relation curve of a cycle number and the residual stress absolute difference; utilizing a nanoindentor to test the elastic modulus absolute difference between dendrite arms and dendrites in the material, thus obtaining a relation curve of the cycle number and the elastic modulus absolute difference; and analyzing the crystal deformation condition by electron back scattering diffraction so as to obtain a relation curve of the cycle number and small angle subboundary crystal deformation. Through comprehensive comparison of the three curves, the dominant factors determining high temperature alloy ultra-high cycle fatigue damage can be understood clearly, and the ultra-high cycle fatigue damage condition of similar other alloys can be predicted.

The invention provides a thermal generator set online performance monitoring method based on component characteristics, which belongs to the technical field of power system simulation and control. The method comprises the steps that 1) required historical measurement data of a power plant are selected, and coarse error rejection and steady-state screening are carried out on the data; 2) a data coordination technology is used to reduce the uncertainty of measured data to provide an accurate data source for modeling; 3) the whole working condition exact model of the key components of a thermal generator set is established based on a dominant factor method; 4) according to the deviation between a monitoring variable target value calculated by a component model and the actual measured value, a device performance change quantitative judgment method based on a statistical control technology is used to judge whether a component is failed, so as to realize online performance monitoring of the thermal generator set. According to the method, data preprocessing and data reconciliation are carried out on the data used for modeling; the modeling data are reliable; the device health can be published online in real time; and a failed component can be accurately located.