179results about How to "Reduce fit error" patented technology

Described is a method for recognizing a shape from a path of a digitizing device. A set of points along a path of the digitizing device is acquired and a parametric representation is fit to the path. The parametric representation is compared to a set of known shapes to determine that the path of the digitizing device represents a particular shape. Fitting the parametric representation to the path includes generating a vector distance field representation of a set of objects corresponding to the set of digitized points, initializing a parametric representation to fit the set of objects, determining a fitting error from the vector distance field representation and adjusting the parametric representation to reduce the fitting error. The fitting error indicates an accuracy of the fit of the parametric representation to the set of objects.

The invention discloses a GNSS carrier loop circuit tracking method based on a stochastic resonance algorithm. According to the invention, the signal to noise ratio of receiving signals is increased by integral accumulation and multiplication feedback treatment on input signals; rough tracking of frequency of the receiving signals is performed by utilizing a second-order frequency locking loop; and precise tracking is performed on the signals subjected to rough tracking by adopting a third-order phase locking loop. Therefore, precise carrier loop circuit tracking on the GNSS signals is realized in a low carrier-to-noise ratio. According to the invention, an advantage of the stochastic resonance algorithm in an aspect of improving the signal to noise ratio and an advantage of anti-interference and robustness of the cooperation of the second order frequency locking loop and the third-order phase locking loop are integrated effectively. The method is simple in hardware implementation and stable in signal tracking, and can realize stable and reliable tracking of highly dynamic satellite signals by a receiver in a severe environment.

The present invention relates to a classification and aggregation sparse representation face identification method based on a nuclear space. The method comprises the following steps: employing a convolutional neural network to extract facial features of a facial image, training a classification and aggregation dictionary, and identifying the image. The classification and aggregation sparse representation face identification method based on the nuclear space considers that the weighting of each training sample to the subspace construction is different and the train samples closed to a class center should have bigger weight to the subspace construction when the train samples are configured to perform sparse representation of test samples, a [Phi](Xc)Wc matrix is adopted to construct a new sparse representation dictionary, and classification concentration constraint terms are added in a sparse representation constraint. Compared with the prior, the classification and aggregation sparse representation face identification method based on the nuclear space is able to effectively reduce the fitting error of test samples in corresponding subspaces to allow samples with the same type to aggregate in the sparse representation so as to improve the face identification performance, enhance the capabilities of processing the non-linear structure and relation, effectively excavate hiding features of complex data and further improve the face identification performance.

The invention particularly discloses a method for extracting surface nuclear magnetic resonance signals on the basis of power-frequency harmonic modeling and self-correlation. The method includes steps of (1), acquiring a group of MRS (magnetic resonance sounding) noised data by the aid of surface nuclear magnetic resonance groundwater detecting instruments; (2), judging whether spike noise exists or not by the aid of a statistic process, removing the spike noise if the spike noise exists, substituting an interpolation result for the spike noise, and keeping the measurement data unchanged if the spike noise does not exist; (3), using a harmonic modeling process to remove power-frequency harmonic noise from the data after the spike noise is removed from the data; (4), carrying out self-correlation and superposition processing on the result obtained in the step (3) to reduce random noise; (5), extracting MRS signal parameters from the result obtained in the step (4). The method has the advantages that the problem of difficulty in effectively extracting MRS signals due to strong power-frequency harmonic interference and random noise in magnetic resonance depth measuring water exploration work can be solved by the aid of the method, and the fitting errors of the key feature parameters of the acquired MRS signals are small.

The invention discloses a tread contour fitting method capable of automatically extracting segmentation points, which comprises the steps of installing laser displacement sensors inside and outside a track according to a mirror symmetry mode; acquiring coordinate data of tread detection points, and converting the coordinate data of each laser displacement sensor into a coordinate system of a vertical plane parallel to the track direction; integrating the converted coordinate data corresponding to the two laser displacement sensors into the same coordinate; performing feature point extraction; determining initial segmentation intervals according to extracted feature points; performing curve fitting according to the initial segmentation intervals, and solving a fitting determination coefficient; comparing the fitting determination coefficient with a preset curve fitting determination coefficient threshold, and determining accurate segmentation points; and determining fitting intervals according to the accurate segmentation points, and performing curve fitting on each interval respectively so as to acquire a complete tread contour. The tread contour fitting method has the characteristics of automatic extraction, high fitting accuracy and high fitting speed.

The invention discloses a generalized load joint probability modeling method considering node spatial correlation of a wind power plant. The method comprises the following steps: step 1 of dividing a root bus node connected with each wind power plant into a power characteristic and a load characteristic according to the respective power flow; step 2 of performing interval refinement on each root bus node respectively according to the active power, and making a statistic of the probability information of each root bus node; step 3 of calculating the relevant characteristic parameters between adjacent node power intervals by using a spatial correlation method aiming at a node region correlation and bringing the relevant characteristic parameters into the characteristic leaning of the node; adopting the RBF (Radial Basis Function) neural network learning train and extracting the node characteristics of an interval set, and establishing a joint probability model structure. The relevant parameter matrixes are blended into the RBF neural network modeling, and the adjacent node voltage is included in the node characteristic learning, thus the established model is more comprehensive. Simulation shows that the fitting error of each segment is smaller and the fitting effect is good.

The invention discloses a time series data nearest-neighbor classifying method based on subsection orthogonal polynomial decomposition. The time series data nearest-neighbor classifying method includes dividing a time sequence into subsequences comprising complete fluctuation trends on the basis of time sequence coded identification turning points; extracting Chebyshev coefficient as subsequence features by means of a first type Chebyshev polynomial decomposition subsequences, and constructing subsequence feature vectors; finally in the nearest-neighbor classifier, classifying by the dynamic planning algorithm based on local mode matching as distance metric function. Classifying accuracy and efficiency are superior to other nearest-neighbor classifiers to the great extent, and the time series data nearest-neighbor classifying method plays an important role in daily activity of people and industrial production, such as in applications of banking transactions, traffic control, air quality and temperature monitoring, industrial process monitoring, medical diagnosis and the like, massive sampling data or high-speed dynamic data can be classified and predicted, abnormalities can be detected and online modes are identified.

The invention provides a solder paste printing performance influence factor analysis method based on SMT big data, and solves the problems of incomplete analysis and low precision in solder paste printing performance influence factor analysis. The method comprises the following implementation steps: collecting solder paste printing parameters and performance indexes to construct a solder paste printing data set; Processing the data by using a Mahalanobis distance and a null value; Calculating correlation coefficients among the features, and filtering redundant features; Dividing training and testing sample sets; Randomly extracting a part of features and constructing a random forest model; Setting a model termination condition; Estimating a characteristic importance degree score accordingto the model mean square error increment, and performing sorting; and determining a subset of key impact factors. According to the method, the key influence factors of the SMT solder paste printing performance are mined through random forest feature selection in combination with the big data processing technology, the correlation between the performance indexes and the printing parameters is determined, the solder paste printing performance is optimized, and the circuit board printing quality is improved. The method is used for process optimization and solder paste printing performance improvement in the surface mounting technology solder paste printing process.

The invention discloses a quasi-optimal energy management method for day running of a fuel cell tramcar. According to target route information, site information and the train scheduling scheme, the SOC reduction standard of every adjacent sections of an energy storage system in the fuel cell tramcar is obtained; the auto-adapted plan algorithm based on an extreme learning machine is adopted for calculating the optimal power distribution relations of a train in various running sections under different loading conditions, and the optimal power distribution relations comprise a power time curve and a power system SOC change curve and are stored to a car automatic running system in a matrix manner; and the optimal power distribution scheme of the next running section is acquired based on a fuzzy control method in the car automatic running system. The specific parameters of the energy management method of train practical running can be adjusted self-adaptively along with changes of the loadcapacity, the energy consumption level of the train is effectively reduced, and the feasibility that the offline energy management method can be adopted during practical running of the train is improved.

The invention discloses a digital image analysis method based on fractal dimension. The method is characterized by comprising the following steps: extracting all color attribute information included in a digital image through an image information matrix U; calculating the distance between pixel points by use of the image information matrix U; performing increment judgment on elements in the image information matrix U; selecting an increment manifold algorithm or non-increment manifold algorithm according to the increment judgment result to obtain a low-dimensional pixel point space of the image information matrix U; calculating the distance between any two pixel points in the low-dimensional pixel point space by use of the low-dimensional pixel point space; and calculating the fractal dimension of the digital image by use of the obtained distance between the pixel points so as to realize classification of different digital images. The method disclosed by the invention can effectively obtain the fractal dimension from the image to be processed, ensures relatively least fitting error, and realizes good classification ability on the digital image.

The invention provides a spectrum Gaussian peak feature extraction algorithm based on a genetic algorithm. The main steps include: S1: inputting a spectrum curve, smoothing the spectrum curve and correcting the baseline; S2: population initialization stage; S3: individual variation stage; S4: population crossing stage; S5: population gene decoding stage; 6, calculating fitness fit; 7, judging whether the fitness fit meets the preset condition: if so, executing step S9; if not, executing step S8; S8: natural selection stage; S9: Gaussian peak characteristics being obtained and output by fittingusing the best fitting interval. The invention divides the spectral curve into a plurality of sub-intervals by taking a wave trough as a dividing point, and then combines the sub-intervals to carry out Gaussian peak fitting, obtains an optimal fitting interval through a genetic algorithm, reduces the fitting error, and improves the accuracy of Gaussian peak parameter extraction and the recognition rate of terahertz time domain spectral material identification. The best fitting Gaussian peak number is obtained by the optimization condition of peak number increment, which enhances the resolvingpower of spectral aliasing peaks.

The invention puts forward a lithium battery capacity and service life prediction method based on a generalized degradation model and multiscale analysis. The method comprises the following steps of: firstly, carrying out phase space reconstruction on historical capacity degradation data, utilizing reconstructed data to fit the generalized degradation model, obtaining a model parameter, and fitting a root-mean-square error; secondly, in multiple scales, carrying out forward estimation on a future capacity, and obtaining a forward prediction result through weighted integration; thirdly, adding a prediction result into a known sequence, and carrying out forward iterated prediction; and finally, judging whether a multi-step prediction result achieves a failure threshold value or not, and calculating remaining service life. By use of the generalized degradation model which is put forward by the invention, a complex combined degradation rule can be fit, and the model has the advantages of high universality, wide applicable range and small fitting error. On the basis of phase space reconstruction, multiscale prediction is carried out, a prediction result is obtained through the weighted integration, and the prediction result is unlikely to interfere by noise and is high in accuracy.

The invention discloses an abstract based time-series data index building method and belongs to the technical field of big data and databases. The core content lies in that time-series data are divided into small time segments, the time segments are then compressed with a linear fitting method to form abstracts of the segments, and a tree index structure is built for the compressed abstract data; when visualized query is needed, an incremental visualization technique on the basis of the index structure is adopted to ensure the visualization method to be accurate and efficient, and multiple times of queries and uses can be provided for building the index structure once. In the constructed tree index structure, fitting error of nodes top and down is lessened, display can be queried according to different query accuracies, the index structure is searched in querying results, the amount data is greatly reduced as compared with that in querying original data by scanning each time, the query results are ensured to be returned within time delay specified by a user, and the visualization method is ensured to be efficient.

The invention discloses a probability statistics-based method for modeling generalized node characteristics. The method comprises the steps of performing uniformly-spaced sampling on annual data of the voltage and active power at a root bus which is directly connected with a fan; performing compromised segmentation on active power data of annual root bus; performing numerical statement on probability distribution of the active power data of the annual root bus of a wind electric field; extracting a probability characteristic relation expression under each power segment by using a nerve network algorithm; establishing a final node characteristic model of the wind electric field; evaluating the fitting effect of each power segment and the fitting effect of an overall sample respectively. The method disclosed by the invention covers an annual sample rule and does not need to correct the model frequently, so that the method is superior to the conventional method in the aspects of self-description capability and generalization capability, and trend results of all power scenes under wind power fluctuation can be solved according to the segment probability model so as to provide a theoretical guidance for researching probability stability calculation, risk assessment, economic dispatch, wind power absorptive capability and the like.

The invention provides a high-speed machining stability domain prediction method based on cutter tooth cutting time course fine integration. The high-speed machining stability domain prediction methodcomprises: firstly, establishing a high-speed milling dynamics model considering regeneration flutter; secondly, dispersing the rotation period of the main shaft, and simplifying the Duhamel integralin the time delay differential equation into an integral of a state item and a time delay item at each dispersed small interval; secondly, using a piecewise cubic Hermite interpolation polynomial tofit a time lag item, and meanwhile, using a plurality of known time node responses and derivatives thereof to approach a required item, so that the fitting error of a calculation method is reduced, and the precision of the prediction method is improved. In addition, in the process of solving the stable lobe graph, a fine integration algorithm is applied, matrix inversion operation in the calculation process is avoided, the time consumed by the calculation method is greatly shortened, and the calculation efficiency is improved.

The invention provides an intelligent facility detection device comprising a shell, a controller, a voice storage module, a lens turning mechanism, a liquid crystal vision test chart, a touch screen, answering buttons and a printer. The shell is provided with a left sight hole and a right sight hole; the lens turning mechanism is provided with a left eye lens mechanism and a right eye lens mechanism corresponding to the left sight hole and the right side hole respectively. The intelligent facility detection device is characterized by intelligent fully-automatic one-button operations, increases the accuracy of detection results, prevents myopia test card printing error, detection distance error, detection time error, error caused by single-eye viewing instead of double-eye viewing, operation error, counting error, and subject cooperation error, increases the detection efficiency, standardizes detection result records, provides voice interaction, improves the enthusiasm of subjects, has complete functions, can be used for both detecting and training facility, and can monitor and guarantee the training effect and improve the training efficiency.

The invention discloses a composite structure of a radiator fan and a molding method thereof. The composite structure of the radiator fan is characterized by comprising a base plate, a stator insertion seat positioning part, a stator group, a plastic combined seat part, an axle center and a rotor blade group, wherein the base plate comprises an attached surface and a stator assembly surface; the stator insertion seat positioning part is arranged in the center of the stator assembly surface; the stator group comprises silicon steel sheets, coils and a plastic insulating frame seat; the plastic combined seat part is integrally formed on the stator group, and can be inserted, connected and matched with the stator insertion seat positioning part; the axle center is connected to the plastic combined seat part, and has a root fixedly connected to the center of the plastic combined seat part; and the rotor blade group comprises a hub, blades, a shaft sleeve, a bearing component and magnetic rings. The axle center is integrally assembled on the plastic structure of the stator group, and a composite design is adopted between the stator group and the base plate, so precise positioning between the axle shaft and the silicon steel sheets and coils of the stator group can be easily realized by the planning design of an extrusion mould. Therefore, the composite structure of the radiator fan can achieve great improvements in combination and position correction accuracy, is smoother to run and has the advantages of effectively reducing noises, along with practicability and progressiveness.

The invention discloses a north finding method of ARMA time sequence based on an optical fiber gyro. Before the process of data processing, the invention analyzes the output sequence of a plurality of optical fiber gyros with different models in a north finding system to determine that for the series of numbers in the original series of numbers, which goes through random stationarity processing, the fitting of ARMA (2, 1) model can be implemented with the best effect; therefore, the invention proposes the establishment of an output model of a north finding testing fiber gyro by using ARMA and offers a whole proposal of parameter solving and model building. First, the proposal reduces the error of the north finding model fitting, thus leading the fitted model to be closer to the property of original series of numbers. The model has higher accuracy and certain university in the north finding technology and provides a foundation for the application of subsequent filter methods such as kalman filter which needs relatively accurate north finding models and the like. In addition, in regard to the solving of model parameters, fixed solving steps and the establishment method of a model of the original series of numbers are proposed, thus improving the speed and efficiency of data processing in the north finding testing.

The invention discloses a thermal power plant time sequence variable abnormity monitoring method and system. The method comprises the steps of collecting time sequence data of thermal power plant analog signal monitoring variables; taking the whole time sequence data as an initial optimal PLR segment, calculating a decision coefficient of the optimal PLR segment, and taking the whole time sequencein the historical sample data as a first-layer original data segment; segmenting the original data segment to form two new data segments, and respectively calculating decision coefficients of the twonew data segments; judging whether the segments are allowed or not, and if yes, continuing to segment the segments; and judging whether the segmentation is finished or not to obtain the optimal PLR segmentation number. According to the method, the optimal segment number of the PLR is determined through a coefficient determination method, and the problems that an L method highly depends on a fitting error and left and right points in a time sequence are solved.

The invention requests to protect a yarn defect detection method and device based on a genetic algorithm. The method comprise sthe following steps of extracting a yarn image group by using a microscope lens; initializing characteristic parameters of the yarn image group; nitializing correction, performing standard particle swarm segmentation processing on the initialized and corrected yarn image swarm; extracting the hairiness image of the segmented yarn block image, carrying out morphological processing to obtain a full multi-frame sample image frame, carrying out defect detection of a genetic algorithm on the hairiness image, outputting the optimal defect detection of the yarn genetic algorithm, and storing the yarn image group with the detected defect. Yarn input from the input port endof the yarn textile machine serves as a research object, and a corresponding solution is provided for the application problem of the genetic algorithm in defect detection and diagnosis in the yarn textile detection work process. Through analysis of image processing parameters, an image processing method with good self-adaption is obtained. By combining the advantages of the two length calculationmethods, the length information of the hairiness can be accurately obtained.

The invention discloses a machine cover hinge high-precision centering and positioning structure and relates to the technical field of positioning structures. A Y-direction nylon position block facingthat middle is arrange at both ends of the Y-direction centering mechanism, the two nylon positioning blocks are respectively in close contact with the outer sides of the two front doors of the vehicle, An X-direction nylon position block facing that front side of the vehicle is arrange at both ends of the XZ-direction positioning mechanism, Two cam guides are symmetrically arranged in the middle, two X-direction nylon positioning blocks are respectively close to the rear sides of the two front doors of the vehicle, the two cam guides are close to the rear side of the engine room front uppercasing assembly, and the two ball-head bolts are positioned at the overlap edges directly above the engine room front upper casing assembly. As that front face and the side face of the left and rightvehicle door are directly taken as the positioning reference, when the mount hole on the engine compartment longitudinal beam is taken as the positioning reference of the tooling, the spatial dimension error between the engine compartment longitudinal beam itself and the left and right vehicle doors is reduce, thereby reducing the matching error between the engine cover hinge and the vehicle door,and further reducing the clearance error between the engine cover and the vehicle door.