340results about How to "Solve the imbalance" patented technology

The invention provides an MRI (Magnetic Resonance Imaging) brain tumor localization and intratumoral segmentation method based on a deep cascaded convolution network, which comprises the steps of building a deep cascaded convolution network segmentation model; performing model training and parameter optimization; and carrying out fast localization and intratumoral segmentation on a multi-modal MRIbrain tumor. According to the MRI brain tumor localization and intratumoral segmentation method provided by the invention based on the deep cascaded convolution network, a deep cascaded hybrid neuralnetwork formed by a full convolution neural network and a classified convolution neural network is constructed, the segmentation process is divided into a complete tumor region localization phase andan intratumoral sub-region localization phase, and hierarchical MRI brain tumor fast and accurate localization and intratumoral sub-region segmentation are realized. Firstly, the complete tumor region is localized from an MRI image by adopting a full convolution network method, and then the complete tumor is further divided into an edema region, a non-enhanced tumor region, an enhanced tumor region and a necrosis region by adopting an image classification method, and accurate localization for the multi-modal MRI brain tumor and fast and accurate segmentation for the intratumoral sub-regions are realized.

The invention discloses a deep residual network-based semantic mammary gland molybdenum target image lump segmentation method. The method comprises the following steps of: labelling pixel categories of lumps and normal tissues corresponding to a collected mammary gland molybdenum target image so as to generate label images, and dividing the mammary gland molybdenum target image and the corresponding label images into training samples and test samples; preprocessing the training samples to form a training data set; constructing a deep residual network, and training the network by utilizing thetraining data set, so as to obtain a deep residual network training model; after a to-be-segmented mammary gland molybdenum target image lump is preprocessed, carrying out binary classification and post-processing on a pixel of the to-be-segmented mammary gland molybdenum target image by utilizing the deep residual network training model, and outputting lump segmentation image to realize semanticsegmentation of the mammary gland molybdenum target image lump. The method is capable of effectively improving the automatic and intelligent levels of mammary gland molybdenum target image lump segmentation, and can be applied to the technical field of assisting radiologists to carry out medical diagnosis.

The invention discloses a disk fault detection method using multi-model prediction, which extracts multiple characteristics of disk SMART indexes through a sequential data processing means, and establishes a classification model to predict disk state. Step 1, data input: acquiring a data set composed of monitoring data of a plurality of disks in a period of time; step 2, SMART screening: adoptingmutation point detection mode to select SMART index; step 3, feature engineering: using SMART index as the input of the user-defined feature extraction module to extract the features of the SMART index, then extracting the corresponding parameter configuration, and transmitting the parameter configuration to the feature extraction module as a parameter, so as to extract the feature sets of the training set and the test set; step 4, data set balance: desampling the negative sample which occupies a large amount by adopting dimension reduction clustering; step 5, algorithm selection and modeling:on the basis of the step 4, training the classification model and testing whether the current disk belongs to the normal state or the fault state that needs to be replaced.

The invention discloses a PHS mobile phone which comprises an antenna, a front end filter, an antenna switch, a radio frequency power amplifier, a radio frequency transceiver chip, a baseband chip, a memory and a peripheral device. The antenna, the front end filter, the antenna switch, the radio frequency power amplifier, the radio frequency transceiver chip and the baseband chip are connected in turn, and the baseband chip is further connected with the memory and the peripheral device respectively. The invention further discloses a transmission power control method of the PHS mobile phone. The PHS mobile phone can adjust the transmission power of the PHS mobile phone at all times according to the strength range of received signals. The transmission power of the PHS mobile phone is increased in the weak signal area, and the transmission power of the PHS mobile phone is reduced in the stronger signal area.

The invention discloses an analog circuit fault diagnosis method and an implementation method of the analog circuit fault diagnosis method. The content includes the first part of analog circuit fault feature information extraction, the second part of fault classifier construction, and the third part of implementation of algorithm software. The analog circuit fault diagnosis method includes the following steps of constructing a fault feature information base, selecting an optimal mother wavelet through an information entropy maximizing principle, conducting wavelet decomposition on response nodes of a measured circuit, extracting the optimal feature of the measured circuit, conducting dimensionality reduction on the fault features through principal component analysis, conducting fault classification and intelligent diagnosis, constructing a fault diagnosis device according to the obtained fault feature information and through a multi-classifier cascade connection model and the classifier integration technology so as to recognize existing faults and causes of the faults, and conducting specific implementation on the algorithm through a C#.NET platform and through combination with the Weka software. The diagnosis method and the implementation method have the advantages of being high in fault diagnosis performance, wider in diagnosis range, higher in algorithm robustness and higher in interpretability.

A method of assigning a carrier for mitigating inter-cell interference in a fixed relay based communication system is provided. In a normal mode, the method includes defining at least one sector group by grouping cells at least two tiers away from each other; selecting a single sector group; and assigning, by a base station of the selected sector group, a first carrier to a relay station or a mobile station which is located in a coverage area of the base station. In a relay mode, the method includes determining whether a current mode is a normal mode or a relay mode; determining whether a base station is included in a selected sector group; and assigning carriers to a relay station or a mobile station belonging to the base station, when the current mode is the normal mode and the base station is included in the selected sector group.

The invention discloses a method for measuring the pressure of a lubricating liquid film of a liquid sliding bearing to achieve the real-time measurement of the all-around lubricating liquid film pressure of a liquid sliding journal bearing and a thrust bearing with different rotation speeds. A pressure sensor is mounted on a revolving shaft by different methods according to different shaft diameters; the sensor is connected with a wireless emission device mounted in the shaft or on the tail end of the shaft; the wireless emission device transmits the liquid film pressure signal of the bearing to the outside of a bearing rotor system while rotating together with the shaft; an external wireless receiving device receives the pressure signals so as to monitor the pressure distribution characteristics and the lubrication state of the lubricating film of the bearing in real time.

The invention discloses a water-saving and high-yield water and fertilizer management method for corn. The method comprises the first step of sowing and irrigating after seedling emergence, wherein the corn is sowed in the mode of alternating wide row spacing and narrow row spacing; the second step of irrigating and regulating of first irrigating after the seedling emergence; the third step of fertilization and irrigation regulation in a fertility interim, wherein from the small bell stage to the heading stage, the soil water content is kept to account for 70 percent-80 percent of the maximum water holding capacity of a field through drip irrigation; the fourth step of fertilization and irrigation regulation in the fertility middle and later periods, wherein from the flowering stage to the milk-ripe stage, the soil water content is kept to account for 80 percent-90 percent of the maximum water holding capacity of the field through the drip irrigation; the fifth step of water and fertilizer management before harvesting, and the sixth step of harvesting through a selected suitable harvesting machine, wherein corn stalks are smashed through a machine and returned to the field. According to the management method, the purposes of being high yield and quality, saving water and fertilizer and efficiently utilizing water and fertilizer resources in corn production are achieved under the condition of the drip irrigation and fertilizer application with the water.

The invention provides an edible blending oil with balanced nutrients and a production method thereof. The proportion of saturated fatty acid to monounsaturated fatty acid to polyunsaturated fatty acids of the blending oil is designed based on daily intake animal meat fat, so that the standard of and 1t; 1:1:1, omega-6 and omega-3 of 4-5:1) recommended by Chinese Society of Nutrition for the final intake of dietary fatty acid of people is met. The production method comprises the steps of accurately weighting various kinds of vegetable oil, vitamin E equivalent to 0.5 percent of the total mass of the vegetable oil in mass and/or tea polyphenol equivalent to 0.3 percent of the total mass of the vegetable oil in mass according to different formulas, adding the weighted materials into a dosing tank, keeping the temperature to be 20-40 DEG C, evenly stirring the mixture for 20-40 minutes at low speed, filtering the mixture and filling the obtained product to obtain the finished product, i.e. the edible blending oil with the balanced nutrients.

The invention provides a single-lead electrocardio abnormal signal identification method based on generative adversarial networks and convolution recurrent neural networks, and mainly solves the problem that data concentration samples are unbalanced. Categories being low in data concentration data volume are subjected to data enhancement, and then identification and classification of electrocardioabnormal signals are performed, so that reference is provided for assistance of doctors, the wrong diagnosis rate and missed diagnosis rate are reduced, and the workload of the doctor can be alleviated. The generative adversarial networks are used for enabling the data concentration samples to achieve the relative balance, so that training the convolution recurrent neural networks can be realized, and good classification effects can be achieved.

The invention discloses a VIENNA rectifier neutral point potential alternating-direct-current component balance control method which can reduce design capacity of a capacitor on a direct-current side and a filter on an alternating-current side and save cost of a tri-level VIENNA rectifier. On the basis of common mode component injection modulation, a common mode component direct-current compensation coefficient and an alternating-current compensation coefficient are modified dynamically, so that the problem of neutral point unbalance of direct-current and alternating-current components is solved; more importantly, the method guarantees uniformity between triphase alternating-current side harmonic distortion rate and SVPWM while substantially inhibiting neutral point potential low-frequency oscillation in conventional modulation methods and improving quality of output power and can be popularized and applied in the field of battery testing, charging facilities and driving systems.

The invention belongs to the technical field of aquaculture feed and biological engineering, and particularly relates to a novel biological feed for prawn and application thereof. The feed is a dry feed comprising the following components: by weight, 10 ~ 15% of fermented soybean meal, 15 ~ 25% of fermented raw meal, 15 ~ 20% of enzymolysis polypeptide, 5 ~ 7% of wall breaking yeast, 5% ~ 7% of fish meal, 3 %~ 5% of shrimp shell powder, 20% ~ 25% of bread flour, 1 ~ 2% of soybean phospholipids, 1 ~ 2% of calcium dihydrogen phosphate, 0.5 ~ 1% of compound vitamins and 0.5-1% of compound minerals. The novel biological feed uses small biological enzymolysis peptide, fermented plant proteins and the wall breaking yeast as main protein sources for production of low fish meal formula feed, the fish meal use amount of the biological feed for the prawn can be significantly reduced, the feed efficiency and utilization rate can be improved, resources can be effectively saved, environmental and water pollution can be reduced, animal immune resistance to diseases can be enhanced, and long term feeding experiment results show that the low protein feed for the prawn is used for feeding litopenaeus vannamei, has no adverse effects on the growth performance and feed coefficient, and can reduce nitrogen and phosphorus emissions in water.

The invention discloses a high-sugar and low-protein tilapia feed. Based on the total mass fraction of 100%, it consists of the following components: fish meal: 0-3%, soybean meal: 16-20%, rapeseed meal: 18-21%, cotton meal: 12-16%, secondary meal: 39- 42%, soybean oil: 2-3%, choline chloride: 0.2-0.5%, sodium chloride: 0.1-0.5%, calcium dihydrogen phosphate: 1-2%, vitamin B1: 0.004-0.006%, vitamin B6: 0.002-0.004%, zinc sulfate: 0.02-0.04%, chromium chloride: 0.0001-0.0003%, vitamin premix: 0.1-0.2% and mineral premix: 0.5-1%. The invention maximizes the amount of secondary powder in the feed, adds several plant protein sources to overcome the problem of amino acid imbalance of a single protein source, and at the same time significantly reduces the amount of fish meal. By optimizing and adjusting the feed formula, supplementing suitable vitamin and mineral premixes to adjust the sugar utilization ability of tilapia, so that the tilapia can make good use of the sugar compound of the present invention, avoiding the damage of fish liver caused by high sugar, Fatty liver is formed, and even the fish body becomes obese, and the symptoms of "big belly" appear.

The invention relates to a cotton cultivation method. The method comprises the following steps: sowing of cotton seeds and laying of drip irrigation pipes: sowing three rows of cotton seeds under a mulching film, laying a drip irrigation pipe on each row; drip irrigation of seedling emergence period water; drip irrigation of bud period water, wherein the maximum water holding amount of the soil in the bud period is 70%-80%, the drip irrigation liquid of the bud period water is clear water or a water and fertilizer mixed solution; drip irrigation of flowering and boll-setting period water, wherein the maximum water holding amount of the soil in the flowering and boll-setting period is 80%-90%, and the drip irrigation liquid of the flowering and boll-setting period water is a water and fertilizer mixed solution; drip irrigation of boll opening period water, wherein the maximum water holding amount of the soil in the boll opening period is 60%-70%, and the drip irrigation liquid of the boll opening period water is clear water or a water and fertilizer mixed solution. The method also comprises the steps: flowering and boll-setting period chemical topping, flowering and boll-setting period cotton spindling controlling, and ripening and defoliating. Water and fertilizer are saved, water, fertilizer and sunlight can be fully utilized by the cotton, the mechanical production is achieved, manpower and material resources are saved, and finally the cotton yield is improved.

The invention relates to a network attack type identification method based on multi-layer detection, belonging to the technical field of information security. The specific operation steps are as follows: Step 1, acquiring the original training data and preprocessing. 2, construct an integrated classification model. 3, train that ensemble classification model. 4, preprocess that test data. 5, classify that test data. Compared with the existing technology, the network attack type identification method based on multi-layer detection proposed in this patent has the following advantages: (1) adopting smart algorithm to upsample a small number of samples and downsample a large number of samples, so as to solve the problem of imbalance of data set samples. (2) Using the integrated model, the precision and recall rate of the detection are improved. (3) The Drosophila optimization algorithm FOA is combined with support vector machine SVM to realize the optimal and adaptive selection of parameters C and gamma in SVM.

The invention discloses a two-stage alternating-current photovoltaic module based on a DC-DC (direct-current/direct-current) converter with high voltage gain and a single-phase hybrid clamping type three-level inverter, which comprises the DC-DC converter at a front stage and the single-phase hybrid clamping type three-level inverter at a rear stage, wherein the DC-DC converter consists of a switch tube M1, a switch tube M2, an inductor L1 and an inductor L2, the input end of the DC-DC converter is connected with a photovoltaic cell, one output end of the DC-DC converter is connected with a diode D1, and a capacitor C0 is connected in parallel between one output end and the other output end of the DC-DC converter; the DC-DC converter adopts a multi-stage pulse sequence mode to control output voltage; the single-phase hybrid clamping type three-level inverter comprises a single-phase hybrid clamping type circuit and an inverter, the single-phase hybrid clamping type circuit consists of a diode D2, a diode D3 and a capacitor C3, wherein the diode D2 and the diode D3 are connected in series, the capacitor C3 is bridged at the two ends of the series circuit formed by the diode D2 and the diode D3, the inverter consists of an IGBT (insulated gate bipolar translator) Q1, an IGBT Q2, an IGBT Q3, an IGBT Q4, a capacitor C1 and a capacitor C2, and the output end of the inverter outputs alternating current via an inductor L3; and the single-phase hybrid clamping type circuit adopts a carrier phase shift control mode to control the inverter. The fluctuation of voltage can be reduced, the output power can be improved, the method is simple, and the module is easy to design.

The invention discloses a plant alga, zooplankton and fish automatic culture integrated device. Zooplankton, plant algae and fishes are separated through micro-filtration membranes for integrated culture, the food chain unbalance problem existing in polyculture is avoided, and convenience is brought to use; oxygen and carbon dioxide are mutually communicated to meet survival conditions required by internal animals and plants. Water resources are saved by means of a water circulation filtering mode, and a sterilizer in a water circulation module can be used for disinfection and sterilization to guarantee water quality. A magnetizer can convert large molecules into small molecules, water molecules can permeate the micro-filtration membranes more quickly, and the high content of dissolved oxygen in magnetized water makes the survival rate and quality of the animals and the plants improved. Excrement of the zooplankton and the fishes is collected and fermented and then serves as fertilizer of the plant algae for circulation, and the excrement is independently fermented to prevent water from being polluted and meanwhile can be absorbed and used by the plant algae more easily.

The invention provides a constant-backpressure direct drive type electro hydraulic servo system and a control method thereof. The constant-backpressure direct drive type direct electro hydraulic servo system comprises a controller, a driver, a servo motor, a bidirectional constant delivery pump and an asymmetric hydraulic cylinder, wherein the controller, the driver and the servo motor are sequentially connected, the controller controls the servo motor to operate through the driver, the servo motor, the bidirectional constant delivery pump and the asymmetric hydraulic cylinder are sequentially connected, and the servo motor controls the output flow and direction of the bidirectional constant delivery pump for realizing motion control over the asymmetric hydraulic cylinder. The invention further provides the control method of the constant-backpressure direct drive type electro hydraulic servo system. The control method is suitable for controlling the asymmetric hydraulic cylinder of a vertical or non-horizontal installation structural style, the problem of unbalanced flow of the asymmetric hydraulic cylinder and the problem of pressure jump generated in reversing are completely solved, the loop cost can be greatly lowered on the premise of meeting the requirements of the system control accuracy and the response characteristic, and the control scheme can be simplified.

The invention discloses an image classification method and system based on hierarchical multi-task learning. The method comprises the following steps of: firstly, carrying out hierarchical category labeling on images in a training sample set from coarse to fine by utilizing a relation between image categories through experience knowledge of experts in some professional fields; secondly, constructing a hierarchical deep convolutional neural network comprising a feature extraction module, a coarse classification layer and a fine classification layer; secondly, training the coarse classificationlayer and the fine classification layer by adopting a multi-task learning algorithm; and finally, classifying the input test images by using the trained hierarchical deep convolutional neural network.The hierarchical structure information between the image categories is combined with the convolutional neural network, the hierarchical deep convolutional neural network is designed, and a multi-tasklearning algorithm is utilized to realize coarse-to-fine classification of the images.

The invention relates to a management control system of transformer-station VRLA batteries. The system comprises a voltage and current detection unit, multiple voltage detecting and balancing units and a master control unit, wherein the voltage and current detection unit employs a Hall sensor to timely detect the total voltage, charging/discharging current and charging/discharging capacity of a battery group which is composed of all storage batteries, and is equipped with a CAN communication interface with the master control unit to upload data in real time; the voltage detecting and balancing units are connected with the batteries of the battery group, detect related parameters of the batteries in real time, actively balance voltages and communicate with the master control unit; and the master control unit manages and stores data and is communicated with a host computer. The invention also relates to a management and control method of the transformer-station VRLA batteries. According to the invention, an online intelligent battery management system which is capable of actively balancing the batteries and detecting battery health replaces a battery inspecting instrument which can only detect voltages of the batteries, thereby balancing battery voltages online in real time, accurately predicting health levels of the batteries, improving operation and maintenance levels, improving automation degree of the transformer station, and easing the burden of workers.

The invention discloses a natural scene image text detection method based on a cascaded convolutional neural network, which comprises the steps of firstly performing candidate character extraction onan input image by using a maximally stable extremal region method; secondly, connecting a pre-trained 16-net convolutional neural network and 32-net convolutional neural network in series to form a cascaded convolutional neural network, classifying the candidate characters by using the cascaded convolutional neural network, eliminating inauthentic characters in the candidate characters to obtain afinal authentic character detection result; then pairing every two adjacent characters in the character detection result, combining the characters having a shared character until all of the characters are combined, and obtaining a text line; and finally dividing the text line into a plurality of words or phrases according to the character spacing, and acquiring a final text detection result. Thenatural scene image text detection method effectively solves a problem that training samples are unbalance in category, and effectively improves the text detection efficiency and performance.

The invention discloses a liquid crystal display panel CF picture identification method based on statistical learning, and relates to the field of image target detection. The liquid crystal display panel CF picture identification method includes the steps: abstracting a question for identifying defect of a liquid crystal panel as a target detection question and a classification question in image processing; performing exploratory data analysis; annotating images, and deleting the images which are not beneficial to training to avoid the interference on a model; preprocessing the data; performing data enhancement on the images: performing data enhancement by transforming an original image, wherein the transformation mode comprises the steps of randomly overturning the image, changing the color of the image and the like; training a convolutional neural network model, improving the current specific question, and carrying out model evaluation; and outputting a final model, and detecting andclassifying the liquid crystal panel. For the liquid crystal display panel CF picture identification method, pictures are out of order before pre-reading, so as to avoid excessive memory space consumption caused by out-of-order during pre-reading while preventing only one type of data from being used for weight updating each time, and improving the stability of the model training process; and thelearning rate is dynamically adjusted according to image characteristics, so that the convergence process is accelerated, and the number of iterations is reduced.

The invention discloses an enhanced VSVPWM (virtual space vector pulse-width modulation) method. The method comprises steps as follows: space vectors are converted into virtual vectors through PWM, and the action time duty ratio of each virtual vector is calculated; duration per unit values of three-phase output voltage at states P and N are figured out according to the correspondence between the virtual vectors and actual space vectors; the duration per unit values of the three-phase output voltage at the states P and N are used as modulating waves respectively, and a three-phase switch state is obtained; small vector action time is adjusted according to the difference between voltage of two direct-current capacitors or the fluctuation magnitude of midpoint potential, and PWM output of an inverter is obtained. With the adoption of the method, the problem of virtual vector PWM midpoint potential cumulative deviation is solved, midpoint potential balance control can be realized within any modulation degrees, the low-frequency fluctuation of the midpoint potential can be completely eliminated, the anti-interference performance of a system is improved, sizes of the direct-current capacitors are reduced, and the output waveform quality is improved.

The invention discloses an automatic incident detection method based on under-sampling and used for an unbalanced data set. The automatic incident detection method comprises the steps of (1) using a maximum and minimum normalization method to carry out normalization processing on actually-measured traffic flow data, carrying out under-sampling processing on a majority class in a training set on the basis of a neighborhood cleaning rule to obtain a new training set which is relatively balanced, (2) selecting a radial basis function as a kernel function of a support vector machine, using an improved grid search algorithm to optimize a penalty factor C and a kernel parameter g of the support vector machine, and (3) training the support vector machine through the training set which is relatively balanced so as to obtain an automatic incident detection model used for the unbalanced data set. According to the automatic incident detection method based on under-sampling and used for the unbalanced data set, the problem that an existing traffic incident detection algorithm is not applicable to unbalanced traffic data in reality is solved, detection performance of the traffic incident detection algorithm is remarkably improved, the average detection time is shortened, and the requirement of traffic incident detection for real-time performance is met.