60results about How to "Avoid the modeling process" patented technology

The invention discloses a multi-spectrum remote sensing image change detection method. The traditional remote sensing image change detection method comprises the following steps of forming a difference image on the two remote sensing images different in time phases through an algebraic method, then modeling the difference image, determining the change threshold value, and detecting the change of the image. However, the process has the problems that the difference image may not meet the designated model and the change threshold value is difficultly determined, so the remote sensing image change detection method based on an SVM (support vector machine) mixed kernel function is put forward for solving the problems. The implementation process comprises the following steps of firstly, constructing the difference image in a way of combining PCA (primary component analysis) conversion and a correlation coefficient integrating method; then, extracting the gray feature and texture feature of the difference image, and carrying out normalization and formatting; selecting a training area, and constructing the SVM mixed kernel function to train; finally, enabling an SVM mixed kernel classifier to classify the difference image, so the change detection result of the target image is obtained.

The invention provides a dynamic power system economic dispatch method based on deep reinforcement learning. The method converts dynamic economic dispatch into a multi-stage sequential decision model,takes a dispatch center for action decision as a decision-making main body and an actual power system as an environment, and improves economic dispatch efficiency by designing elements such as actions, states and rewards in reinforcement learning. and converts the economic dispatching model of the power system into a typical multi-stage sequential decision model. According to the model, modelingof an increasingly complex power system is avoided, an accurate thermal power generating unit output cost function is not required, and dynamic economic dispatching of the power system in any scene isrealized through continuous interaction of an intelligent agent and the environment, updating of a strategy, self-adaptive load and uncertainty of new energy output.

The invention relates to an Alzheimer's disease multi-classification diagnosis system based on deep study. The Alzheimer's disease multi-classification diagnosis system comprises an image characteristic extracting module, an index characteristic selecting module, a vector linear merging module and a disease classification and diagnosis module, wherein the image characteristic extracting module is used for extracting characteristic vectors of a cerebral three-orthogonal plane MRI image according to a neural network model; the index characteristic selecting module is used for selecting checking indexes according to medical pertinent literatures to form index characteristic vectors; the vector linear merging module is used for adopting a multivariate data linear merging method based on canonical correlation analysis to merge the characteristic vectors of the image and the index characteristic vectors; and the disease classification and diagnosis module is used for inputting the merged vectors to a multi-classification classifier to distinguish the three stages of the Alzheimer's disease. The Alzheimer's disease multi-classification diagnosis system disclosed by the invention can assist the multi-classification diagnosis of the Alzheimer's disease.

The invention discloses an intelligent algorithm based electricity-larceny-preventive monitoring method for distributed type photovoltaic power generation. The method is characterized by including step 1, creating a distributed type photovoltaic power generation database, wherein database information includes historical meteorological data of a photovoltaic power station and photovoltaic power generation power data in the corresponding period; step 2, taking the meteorological data as input, taking photovoltaic power generation power values as output, and creating an electricity-larceny prevention model according to an intelligent algorithm; step 3, predicting photovoltaic power by the aid of the electricity-larceny prevention model, and calculating theoretical power generation quantity of the photovoltaic power station by the aid of a calculation model of the theoretical power generation quantity of the photovoltaic power station; step 4, inputting collected real-time power generation quantity of the photovoltaic power station and the theoretical power generation quantity of the photovoltaic power station into a power quantity abnormity judgment module to acquire an power generation electricity-larceny suspicion judgment result of the photovoltaic power station. The intelligent algorithm based electricity-larceny-preventive monitoring method has the advantages that photovoltaic power quantity abnormity judgment accuracy is high, electricity-larceny suspicion coefficients are given, auditing work efficiency is improved, and the basis is provided for effective monitoring of distributed type photovoltaic power generation.

The invention relates to a device for extracting chip fingerprint and a method thereof, which belong to the technical field of information processing. The device comprises transmission delay passage, an excitation signal controller and an arbiter. The transmission delay passage is formed by N switches in series arrangement, wherein N is an integral multiple of 2, and each of the switches comprises a pair of 2-1 multiplexers and a pair of buffers. The excitation signal controller comprises an N-bit linear shift feedback register. The arbiter comprises a D trigger. The N-bit output of the excitation signal controller is connected with the control terminals of the N switches. The arbiter is connected with the output terminal of the last switch. The same one pulse signal competitively passes through the two paths. The D trigger in the arbiter judges that the output is 0 or 1 according to the sequence in which the rising edge of pulse signal of the upper path and the lower path arrives. The device is excellent in the 0/1 balance of output response.

The invention relates to a generator wide area damping control method based on a model-free adaptive control algorithm and belongs to the electric power system stabilization analysis technology field. The method comprises steps that control parameters of a traditional generating wide area damping control method are acquired according to a traditional transfer function expression and a parameter calculating formula; control parameter initial values of the generator wide area damping adaptive control method is calculated according to the acquired control parameters of the traditional generating wide area damping control method, an acquired pseudo gradient vector initial value is taken as an initial point, and an estimated value of a pseudo gradient vector is updated online by utilizing input output data actually measured by a controlled system; a generator wide area damping adaptive control signal is calculated according to the estimated value of the pseudo gradient vector which is updated online; repeated calculation is carried out till the generator wide area damping control signals reach to a maximum level. The method can guarantee a control effect of generator wide area damping control under multiple operation states of the system.

The invention relates to the field of hull deformation measurement, and provides a time delay compensation-based model-free hull deformation measurement method which is capable of estimating a deformation angle of a ship in real time and quickly under a condition of no deformation angle prior model and estimating and compensating a time delay existing among data. The method comprises the following steps: installing two laser gyro systems near a ship center inertial navigation system and shipborne equipment; building a deformation filter observation amount according to attitude information at installation points; deducing a mathematic relation between an ideal attitude matrix and an actual attitude matrix through introduction of a time delay amount according to a quaternion attitude matrix, extending the time delay amount into a system state variable, estimating the deformation angle of the ship through a neutral network, extending a connection weight coefficient of the neutral network into the system state variable, solving a built system state equation and an observation equation by utilizing a nonlinear filter, and estimating to obtain the deformation angle of the ship and the size of the time delay.

The invention discloses a mountain fire risk prediction modeling method based on stacking algorithm, which can improve prediction efficiency and prediction accuracy. The modeling method of mountain fire risk prediction is based on combustible data, geographic data, meteorological data, historical mountain fire data and other spatio-temporal data to predict the risk of mountain fire. The processingtechnology of multi-source, heterogeneous, massive spatio-temporal data is designed to form a rich set of characteristics of mountain fire prediction. Have the ability to deal with massive spatio-temporal data; data-driven modeling is realized to avoid tedious and complex Bayesian modeling process and improve the efficiency of spatio-temporal data modeling. At the same time, the modeling method of hill fire risk forecasting takes into account the characteristics of time, space, dynamic and static characteristics, and realizes the secondary processing generation of the characteristics by stacking method, which improves the overall effect of hill fire risk forecasting; the experimental results show that the AUC index reaches 0.85. It is suitable for popularizing and applying in the field ofdata processing technology.

The invention relates to a Chinese named entity recognition model based on reinforcement learning and a training method thereof. The model comprises a strategy network module, a word segmentation recombination network and a named entity recognition network module. The method comprises: firstly, a strategy network specifying an action sequence; then, the word segmentation recombination network executing actions in the action sequence one by one; obtaining a phrase through a 'termination' action, the phrase being used as auxiliary input information to perform lattice-LSTM modeling to obtain a hidden state sequence, inputting the hidden state into a named entity recognition network to obtain a label sequence of sentences, and a recognition result being used as update of a delay reward guidance strategy network module. According to the method, sentences are effectively divided through reinforcement learning, modeling of redundant interference words matched in the sentences is avoided, dependence on an external dictionary and influences on long texts are effectively avoided, correct word information can be better utilized, and the Chinese named entity recognition model is better helpedto improve the recognition effect.

The invention discloses a three-dimensional model change method and device. The method comprises the following steps: comparing a target three-dimensional model of a target object with a preset reference three-dimensional model to determine a model conversion parameter, wherein the model conversion parameter is used for representing a conversion mode used for converting the reference three-dimensional model into the target three-dimensional model; and according to the model conversion parameter, converting an original three-dimensional model which needs to be converted. A conversion mode of the original three-dimensional model is determined through comparison with the reference model so as to convert the original three-dimensional model, a problem that modeling is difficult to carry out since the original three-dimensional model is different from the target three-dimensional model can be effectively overcome, remodeling and repeated calculation can be avoided, and problems of low modeling precision and high cost caused by the remodeling and the repeated calculation can be avoided, and the method and the device exhibit high flexibility and operation efficiency and can effectively adapt to constantly changing demands.

The invention discloses a robust-regression-based distributed photovoltaic generating electricity-stealing identification method. The method comprises the following steps: (1), establishing a historical information database; (2), carrying out determination and filtering on abnormal data existing in historical data; (3), carrying out processing by using a robust regression model algorithm to obtain an irradiation power curve; (4), carrying out operation to obtain a corresponding photovoltaic generation power; (5), carrying out calculation to obtain a theoretic generating capacity; and (6), carrying out determination. The method has the following beneficial effects: (1), with the robust regression model algorithm, the influence on the model precision by the abnormal data can be reduced and concrete modeling of a photovoltaic system inversion model and a photovoltaic conversion model can be avoided; and (2), electricity-stealing suspicion determination is carried out based on three-layer screening architecture, so that accuracy of abnormal determination of the photovoltaic electric quantity and the electricity-stealing determination reliability are high and the electricity-stealing checking pertinency is improved.

The invention discloses a hill fire risk prediction method based on a stacking algorithm, which can improve prediction efficiency and prediction accuracy. The hill fire risk prediction method adopts combustible factor data, geographic data, meteorological data, historical hill fire data and other time-space data to predict the hill fire risk. The processing technology of multi-source, heterogeneous, massive time-space data is designed, and the abundant hill fire occurrence prediction feature set is formed. The hill fire risk prediction method has the ability to deal with massive spatio-temporal data; the data-driven modeling is realized to avoid tedious and complex Bayesian modeling process, and the efficiency of spatio-temporal data modeling is improved. At the same time, the hill fire risk prediction method takes into account the characteristics of time, space, dynamic and static characteristics, and realizes the secondary processing generation of the characteristics through the stacking method, which improves the overall effect of hill fire risk prediction. The experimental results show that the AUC index reaches 0.85. The hill fire risk prediction method is suitable for popularizing and applying in the field of data processing technology.

The invention discloses an integrated energy system weak point identification method based on a unified power flow model and a random matrix. The integrated energy system weak point identification method comprises the following steps of: (1) establishing an integrated energy system unified power flow model; (2) selecting power flow data and establishing a random matrix; (3) standardizing and preprocessing the random matrix; (4) identifying the abnormal state of the system; and (5) identifying weak points of the system. Methods of the invention may be provided. According to the invention, a random matrix theory is introduced into integrated energy system weak identification; the operation state of the system is analyzed in a data driving mode, the physical structure of the comprehensive energy system does not need to be identified, the modeling process of the comprehensive energy system is avoided, the complexity of the physical structure and the modeling process of the comprehensive energy system is overcome, and the accuracy and rapidity of system state identification and weak point identification are improved.

The invention discloses a finite element modeling method of a special-shaped honeycomb skin structure, comprising the following steps of: step1, building a section geometric model of the special-shaped honeycomb skin structure; step2, generating a honeycomb central point of a certain ring; step3, screening intra-domain honeycombs; step4, executing step15 if no intra-domain honeycomb exists in the current ring, otherwise, executing step5; step5, generating a node for the intra-domain honeycomb in the current ring; step6, processing the honeycomb node; step7, executing step8 if the number of the intra-domain nodes is more than 2, otherwise, executing step13; step8, dividing linear units of the honeycomb; step9, dividing the corresponding skin units; step10, executing step11 if the number of skin unit nodes is 0, otherwise, executing step13; step11, outputting information of the skin units; step12, outputting information of skin nodes; step13, executing step5 if honeycombs of the current ring are not completely processed, otherwise, executing step14; step14, executing step2 if honeycombs of all rings are not completely processed, otherwise, executing step15; step15, outputting information of the honeycomb units; step16, outputting information of the honeycomb nodes; step17, outputting information of a unit collection in bounding parts of the honeycomb.

The embodiment of the invention provides a tire wear degree identification method and device. The method comprises the steps that the circumferential acceleration sequence and the radial accelerationsequence of a tire in the driving process are acquired, and the sequences are normalized according to the vehicle speed square of the same moment; the normalized circumferential acceleration sequenceand radial acceleration sequence form a sample to be predicted in the form of a two-dimensional matrix; and the sample to be predicted is inputted to a preset convolutional neural network model, and the wear degree of the tire is determined according to the output result of the convolutional neural network model. According to the method, modeling of the complex mechanism of tire wear can be avoided, and wear degree identification can be realized by analyzing the circumferential and radial acceleration response of the tire so that the method is simpler and more convenient and higher in precision and online real-time prediction can be realized. End-to-end wear degree prediction is performed on the recombined acceleration signal matrix by using the powerful feature extraction function of theconvolutional neural network without too much early pretreatment so as to have robustness for the change of the driving condition.

The invention relates to the field of power system operation control technologies, and specifically relates to a cable conductor temperature determining method. The method comprises the following steps: (1) establishing a quarternary linear regression model of head and end voltage, current, active power, reactive power and resistance of a cable line; (2) determining the resistance of the cable line according to the quarternary linear regression model; (3) converting the resistance of the cable line into the resistance of a conductor of unit length; and (4) determining the temperature of a cable conductor. By adopting the method of the invention, the modeling of a complex heat transfer model or heat circuit model is avoided, measurement of the temperature of the outer sheath of the cable is not needed, and the impact of the external environment on the result of temperature calculation does not need to be considered. The method is based on a simple principle, is quick in calculation, and can provide a reference basis for overheat or overload risk identification and early warning.

The invention belongs to the field of automatic driving simulation, and particularly relates to a rapid lane drawing method and system, and the method comprises the steps: setting parameters of a lane; creating a corresponding number of lanes according to the parameters of the lanes; setting a Bezier curve parameter of a lane line center line in the driving line; determining a Bezier curve parameter of a corresponding lane line in the driving line according to the parameters of the lane and the Bezier curve of the center line of the lane line; drawing a lane according to the Bezier curve parameters of the lane; the lane lines are connected with different lanes. According to the lane line drawing method, the drawn lane lines are stored, the lane line center lines and the lane lines in the lane lines are rapidly drawn, mapping and modeling during traditional lane line drawing are avoided, and the lane line drawing method is very rapid and convenient.

The invention provides a microgrid reactive automatic control method based on a convolutional neural network. According to the microgrid reactive automatic control method, an SCADA is adopted to collect real-time operation data of a microgrid system to generate two-dimensional power matrix data; the optimal reactive power of a reactive device corresponding to the two-dimensional power matrix datais calculated by utilizing the optimal power flow, and the optimal reactive power of the reactive device is used as a label value; and a convolutional neural network model is trained, so that the optimal reactive power of each reactive device can be determined according to system operation data. The method is characterized in that the two-dimensional convolution operation sparse interaction, the weight sharing and equivariant representation are utilized, the convolutional neural network model is established and model training is carried out, so as to realize automatic feature extraction on theoperation state of the microgrid, so that the optimal reactive power of each reactive device is determined; and meanwhile, the voltage deviation and the network loss during microgrid operation are taken into consideration, and the method has high economical efficiency and safety.

The invention discloses a camera and galvanometer combined variable sight line system three-dimensional imaging model and a calibration method thereof. Hardware of a variable sight line imaging system comprises a computer, a two-dimensional galvanometer, a CCD or CMOS area-array camera and a group of optical lenses. The two-dimensional galvanometer comprises two optical total reflection mirrors, and the two optical total reflection mirrors can rapidly deflect around an axis under the control of a computer signal, so that the sight line direction and the corresponding imaging area of an image sensor composed of the camera and the lens are changed. According to the three-dimensional imaging model of the variable sight line imaging system, a mapping relation among pixel positions on an image plane, two angle control quantities of a two-dimensional galvanometer and spatial incident light is established. The invention further discloses a calibration method of the variable sight line imaging system three-dimensional imaging model. The model has the outstanding advantage that the camera and galvanometer combined variable sight line imaging systemcan meet the requirements of large-view-field three-dimensional vision application.

The invention provides an online prediction method for ash combustible material content after combustion of a coal ash boiler. The online prediction method comprises the following steps: analyzing factors influencing boiler ash combustibles; adjusting a test result through certain boiler combustion; analyzing in conjunction with a boiler combustion theory to obtain an influence factor influencing the boiler ash combustibles. On the premise of ensuring a certain prediction accuracy, the online prediction method can be implemented in a boiler combustion external control system in a logical way, and complicated computation and modeling in similar nervous network model prediction are avoided, so that the online prediction method has a high application value on the aspect of online combustion optimization control of the boiler.

The invention relates to the pure electric vehicle performance estimation field, to be specific, relates to an electric vehicle traction motor power verification method. The electric vehicle traction motor power verification method comprises a, highest speed power estimation; b, maximum gradient driving power estimation; c, 100km/h accelerating condition power estimation; d, extremum acquisition. The electric vehicle traction motor power verification method has advantages of simple process and reasonable result.

The invention provides a using method of an intelligent meal selling robot based on machine vision. The using method comprises the steps that a dining car slowly moves back and forth on a student dormitory building, the dining car automatically stops when it is detected that someone exists, and a student can select a needed meal on a touch screen; a mechanical arm obtains information and performs image recognition on the meal to obtain a color image and depth; the area where a target object is located is segmented, and the target object is identified; the pose of the identified target object is obtained; motion planning is carried out according to the pose; the target meal is grabbed according to the motion planning; the grabbed meal is placed on a dinner plate to be weighed, the price is calculated, and an external display screen displays the price and a two-dimensional code to wait for payment; and after payment is completed, a valve on the dining car is opened to send out the meal. According to the using method of the intelligent meal selling robot based on machine vision, the mechanical arm has high identification accuracy and pose information estimation precision, intelligent operation is achieved according to image recognition and remote control, the using method can further cope with a complex working environment, and the using method is very convenient in the using process.

The invention provides a business modeling method for a computing system. The business modeling method comprises the steps of managing meta-models of one or more model categories in a meta-model database; receiving first modeling application data from a client, wherein the first modeling application data comprises a first model category; obtaining a first meta-model for the first model category from the meta-model database, and sending the first meta-model to the client; receiving service configuration information about the first meta-model from the client, wherein the service configuration information comprises object configuration information and relationship configuration information; and based on the first meta-model and the service configuration information, constructing a service model belonging to the first model category in a service model database. The invention further provides a business modeling device for the computing system and a computer system.