94results about How to "Address underutilization" patented technology

The invention discloses a legal consulting method and a legal consulting system based on big data analysis. The legal consulting method comprises the following steps of (S1) selecting legal fields required to be consulted; (S2) selecting corresponding problem description frames according to the legal fields; and (S3) analyzing a consulting result from a stored data unit according to the inputted problem description frames. By the legal consulting method and the legal consulting system, the problems that in the traditional legal consulting process, legal provisions and legal cases are required to be retrieved manually so as to answer legal consulting questions, and legal information is used insufficiently in a process of manually answering the legal consulting questions due to restrictions on professional ranges of workers are solved.

The invention provides an extracting method of an antioxidation active cordyceps polysaccharide, in particular to a method for extracting the antioxidation active cordyceps polysaccharide by taking the waste fermentation liquid of Paecilonyces hepialid chen and amp; Dai Cs-e as materials. The method includes the steos that: liquid containing the antioxidation active cordyceps polysaccharide is used as the materials which are centrifuged for obtaining supernatant; the pH of the supernatant is adjusted to be 4.5 to 5.0; ethanol of 50 to 100 percent (V/V) with a volume of 2 to 5 times is added, deposition is carried out for 20 to 60 min, then centrifugation is carried out; a precipitate is taken and washed in sequence by ethanol, acetone and ether, then dried for obtaining the rough polysaccharide powder containing the cordyceps polysaccharide.

The invention relates to an allocation and recovery method of database connections. According to the allocation and recovery method, a database connection pool is used; the database connection pool comprises a global connection pool and a logical connection pool; the global connection pool comprises queues; the logical connection pool comprises queue corresponding to all databases. According to the invention, the queue structure of two connection pools is used, so that the connection in the sequence corresponding to the least active database or the connection never used can be found rapidly through the global connection pool, and the full utilization of connection resources of all the databases can be ensured.

The invention provides a lazy learning-based self-adaptive robustness forecast control method of blast furnace molten iron quality, and relates to the technical field of automatic control of blast furnace melting. The lazy learning-based self-adaptive robustness forecast control method comprises the steps of determining controlled quality and control quantity; acquiring historical production input-output measurement data of a blast furnace to construct an initial database; constructing an inquiry regression vector, and determining abnormal data; inquiring a similar leaning subset from the database, selecting an optimal learning subset, and processing the abnormal data; building a forecast model by taking the optimal learning subset; calculating index reference track of the molten iron quality, constructing a forecast control performance index to obtain an optimal control vector; and sending the optimal control vector to a bottom-layer PLC system, adjusting an execution mechanism, acquiring a new group of blast furnace measurement data, pre-processing the data, and updating the database. According to the method provided by the invention, the influence of the input-output interference can be effectively suppressed, the influence of the abnormal data is overcome, the blast furnace molten iron quality is stabilized to a value near to an expected value, and stable running, good quality and high yield of the blast furnace are facilitated.

The invention discloses a small ball algae cultivation method with high density and high quality, which comprises the following steps: adding the culture medium with glucose and a little of CSL addictive in the gas culture bottle; proceeding rapid small ball algae culture in the heterotrophy condition. The invention simplifies the producing craft and reduces the cost, which is benefit to cultivate large-scale high-density small ball algae. When the substrate glucose density is 100 g per L, the obtaining biomass of small ball algae is 34.97 g per L.

The invention discloses an efficient nitrogen and phosphorus removal waste water treatment process and device. The device is formed by connecting a membrane and sludge coupling pond, a chemical phosphorus removal pond, an aerobiotic nitrifying biological filter and a clean water pond in series. A nitrifying liquid reflux system formed by a reflux pump and a reflux pipe is arranged between the membrane and sludge coupling pond and the clean water pond. The membrane and sludge coupling pond comprises a flowing membrane and sludge coupling pond body or a fixed membrane and sludge coupling pond body. The efficient nitrogen and phosphorus removal treatment process is achieved through the steps of hydrolysis, carbon removal and nitrogen removal; phosphorus removal, carbon removal and suspended solid (SS) removal; nitration and carbon removal. By means of the efficient nitrogen and phosphorus removal waste water treatment process and device, the objective of synchronous nitrogen, carbon and phosphorus removal can be achieved, and the advantages of low cost, few sludge deposits, multiplied organisms and the like are achieved at the same time.

The invention discloses an integrated refining system and process using an oil shale retorting gas to produce hydrogen and upgrade. The system comprises an oil shale destructive distillation unit, a shale oil hydrogenation and upgrading unit, a methane vapor reforming unit and a building material production unit, wherein the methane vapor reforming unit comprises a vapor reforming reactor, a CaO regenerative combustion furnace, a compression heat exchanger, a vapor reforming gas flash-evaporation tower and a transformation adsorption device. According to the system and process disclosed by the invention, methane reforming reaction is carried out onto the destructive distillation gas, a hydrogen source is provided for shale oil hydrogenation and upgrading, complementary utilization of the destructive distillation gas and the shale oil hydrogen element is realized, and waste fragment shale of the oil shale destructive distillation unit is combusted to provide heat for roasting CaCO3 in the CaO regenerative combustion furnace, so that utilization efficiency of resources and energy resources is improved. According to the system and process disclosed by the invention, economic benefits of the existing oil shale oil-refining and upgrading processes can be remarkably improved, resource utilization rate of the process is increased, and emission of solid waste and an industrial waste gas is reduced.

The invention discloses a multi-feature fusion-based online academic report classification method. The method is characterized by comprising the steps of (1) collecting and building an academic report database and coarsely classifying an academic report into certain subjects homogenous with an organizer according to organizer information of the academic report in the database; (2) building a database of researchers in various fields around the country, matching the database with reporter information in the academic report and determining classification; (3) extracting a keyword in a data header of the academic report; (4) carrying out synonym feature expansion on the extracted keyword; (5) carrying out text classification on obtained features after synonym expansion; and (6) synthesizing the classification result to obtain the final classification result of the academic report. The academic report is classified through a multi-factor method, so that the classification speed and accuracy are ensured.

The invention relates to a microblog user age prediction method based on LSTM and LeNet fusion, and belongs to the technical field of information prediction. The method comprises the following steps of crawling data, crawling the information of a microblog user, and storing the information in a local computer; carrying out word segmentation processing on the microblog text, after the text contentword segmentation and stop word filtering, vectorizing the word segmentation results; building an LSTM, adopting a long short-term memory model (LSTM) to model the vector, and predicting the age of the user; preprocessing the images, unifying the images into the same size; establishing the LeNet, establishing a LeNet model, enhancing the picture data of the data set, converting the picture data into the tensor, and testing to select a model with the highest hit rate; and integrating the results, integrating the trained text processing module model and the picture processing module model. Compared with the prior art, the problem that a previous model is difficult to follow the trend is solved, the recognition accuracy is improved, and the method has a wide application prospect in the fieldsof user operation, precise advertisement marketing, user analysis, data analysis, recommendation systems and the like in the future.

The invention discloses a method, a device and equipment for generating a target re-identification model and performing target re-identification, and a storage medium, and relates to the technical field of artificial intelligence, in particular to the technical field of computer vision and deep learning. The specific implementation scheme is as follows: obtaining a labeled sample set, an unlabeledsample set and an initialization model obtained by supervised training; performing feature extraction on each sample in the label-free sample set by using an initialization model; clustering the features extracted from the label-free sample set by using a clustering algorithm; for each sample in the label-free sample set, distributing a pseudo label to the sample according to the corresponding class cluster of the sample in the feature space; and mixing the sample set with the pseudo label and the sample set with the label to serve as a training sample set, and performing supervised trainingon the initialization model to obtain a target re-identification model. According to the embodiment, the labeled data is fully utilized to perform model training, and the model training speed and precision are improved, so that the re-identification accuracy is improved.

The invention discloses a Chinese electronic medical record entity extraction method based on character and word information fusion, and the method comprises the steps: obtaining a Chinese word vectorand a character vector in a Chinese electronic medical record, and obtaining a word vector table and a character vector table; segmenting a document into sentences and characters, and looking up thecharacter vector table for corresponding character vectors; obtaining a context word of each character and carrying out feature extraction on a word vector of the word to obtain context word information of each character; obtaining attention weight, and combining the weight with the character vector of each character and the context word information of the corresponding character to obtain new character information representation; extracting sequence information of each character; and labeling and identifying the sequence information by using a conditional random field to obtain a BIO categoryof each character in the sentence, and decoding an identification result to obtain a specific medical entity. According to the method, manual extraction is reduced, Chinese character and word information can be automatically learned and combined, character and word features are extracted, and the Chinese medical named entity identification result is effectively improved.

The present invention belongs to the forestry remote sensing image intelligent processing field technology, and especially relates to a forest three-dimensional canopy morphological structure extraction method on the basis of an active contour model. The method comprises the steps: obtaining corresponding image points through adoption of a SURF stereo matching algorithm based on a structural information consistency examination mechanism, resolving the three-dimensional space information through combination of a light beam method rigorous solution, constructing a three-dimensional shape model according to the significance area segmentation algorithm and the dynamic contour model based on a selectivity visual attention mechanism, and obtaining forest three-dimensional canopy morphological structure information. The forest three- dimensional canopy morphological structure extraction method on the basis of an active contour model is helpful to greatly improve the forest region tree species partition degrees and serves the comprehensive control and monitoring of forest resources.

The embodiment of the invention provides a deep convolutional neural network acceleration method and system suitable for NPU. According to the NPU calculation unit structure, the data arrangement format of the input parameters of the convolutional layer is adjusted in advance; therefore, the problem that the calculation units are not fully utilized in the convolution calculation process is solved, and the throughput of the matrix calculation units is fully utilized. and execution assembly line of the calculation instruction in the NPU is rearranged, the convolution calculation capability of the NPU is mined, and the calculation parallelism in the deep convolutional neural network is enhanced. According to the acceleration method provided by the invention, the arrangement mode of the data streams and the instructions in the deep convolutional neural network is modified, so that the efficient and rapid convolutional neural network can be realized on the NPU.

The invention discloses a self-supervised learning and skeleton information behavior recognition method, and relates to the technical field of computer vision. The method comprises the following steps: S1, constructing a configurable depth model; S2, in a network pre-training stage, obtaining a pre-training sample according to a preset optical flow prediction task, wherein the pre-training sample comprises a skeleton video and a label of an optical flow prediction task automatically generated by a machine; training the transformation network by using the pre-training sample to obtain an initial parameter theta' of the converter network; S3, in a network fine tuning stage, initializing the converter network according to the initial parameter theta ', and constructing a fine tuning depth model in combination with the initialized converter network and the randomly initialized fine tuning classification network; and S4, inputting a to-be-identified bone video into the trained fine adjustment depth model, and outputting a classification prediction result by the fine adjustment classification network. According to the invention, on the premise of ensuring high precision, human body behavior identification with better effect, robustness and generalization is realized.

The invention relates to a method and a controller capable of enhancing the management efficiency of a nonvolatile storage device, which only read data stored in a system management area of an original paging of each area block to establish a logic entity address mapping table in a buffer storage of the controller so that the management efficiency of the nonvolatile storage device is enhanced. Besides, the method and the controller can also integrate all or partial system management areas inside the original paging to more efficiently manage and use the storage space of all the system management areas inside the original paging.

The invention discloses a double-cropping cultivation method of spring wheat and summer millet in a natural year. The method comprises spring wheat cultivation and summer millet cultivation. The spring wheat cultivation method includes the steps of land preparing, seeding, irrigating and fertilization, prevention and treatment of disease and pest damage, harvesting and the like. The summer milletcultivation method includes the steps of land preparing, fertilization, seeding, thinning, weeding, harvesting and straw returning. A double-cropping mode in one natural year, namely, a spring wheat and summer millet one-year double-cropping cultivation technique, can be achieved in an eastern Hebei region (or regions with similar natural ecoclimate) under natural ecological conditions, the cultivation technical mode herein makes full use of light and heat resources, the problem of insufficient utilization of light and heat resources or insufficiency of light and heat resources of the easternHebei region is solved, and grain yield per unit land area is increased.

The invention discloses a heat recovery device for a yellow phosphorus tail gas boiler. The device comprises a heat inlet, a water tank and an outlet; a filter box is arranged on the right side of theheat inlet; a first filter plate is arranged in the filter box; the upper end of the first filter plate is connected with a rubber plug; a sealing pad is fixed at the lower end of the rubber plug; arecovery groove is formed under the first filter plate; a collecting box is arranged under a second filter plate; the water tank is positioned on the right side of a filter box; a steam inlet is formed in the upper end of the water tank; the outlet is positioned at the bottom end of the water tank; and the right side of the water tank is connected with a treatment box. The heat recovery device forthe yellow phosphorus tail gas boiler solves the problem that most heat recovery devices cannot filter hot air, the problem that general heat recovery devices are insufficient in heat utilization andthe problem that the general heat recovery devices directly exhaust gas without performing treatment.

The invention belongs to the technical field of refractory materials, and particularly relates to a hydrophobic magnesium/calcium synthetic clinker and a preparation technique thereof. The hydrophobic magnesium/calcium synthetic clinker is prepared from the following raw materials in percentage by weight: 25-35% of light-burned magnesia powder, 50-60% of light-burned dolomite powder, 3-10% of high-purity lime and 3-8% of additive. The additive is a combination of rare-earth oxide and fused quartz, or a combination of rare-earth oxide, fused quartz, metallic oxide and non-metallic oxide. The preparation technique comprises the following steps: respectively grinding the light-burned magnesia powder, light-burned dolomite powder and additive, evenly stirring, adding a forming binder, mixing in vacuum, ejecting, molding, carrying out oven drying, and burning to obtain the hydrophobic magnesium/calcium synthetic clinker. The hydrophobic magnesium/calcium synthetic clinker has the advantages of high volume density, low porosity, high thermal shock stability, no environment pollution and the like, and can not be hydrated.

The present invention discloses an absorption tower for producing enriched oxygen through vacuum pressure swing absorption. The absorption tower comprises a material inlet, a raw material gas inlet, a product gas outlet, an upper screen plate and a lower screen plate, wherein the absorption tower presents a tapered shape from big to small according to a direction from the gas inlet end to the gas production end. According to the novel absorption tower, the existing radial and axial absorption tower is changed into the tapered structure, such that it is ensured that the gas stream from the gas inlet end to the gas production end always maintains a uniform and consistent flow rate, an utilization rate of a molecular sieve in the absorption tower is increased, a yield per unit of the absorption agent is increased, and equipment efficiency is high.

The invention discloses a power equipment fault detection model based on combination of an attention mechanism and a GRU, the power equipment fault detection model comprises a classification neural network model, and training data of the classification neural network model comes from a preprocessing model. The preprocessing model converts input unbalanced power equipment data into balanced data and performs embedded representation, and outputs intermediate data: a historical state sequence based on power equipment representation, label data embedded representation and embedded representation of power equipment portrait features; time and space features of the power equipment are extracted from the historical state sequence through a GRU module; state sequence features are extracted from the output of the GRU module through an attention mechanism module; environment information of the power equipment is extracted from the embedded representation of the portrait features of the power equipment through a graph attention mechanism module; and the state sequence features, the label data embedding representation and the environment information are aligned and fused to serve as training data input of the classification neural network.

The invention discloses a shifting switching value input/output device. The shifting switching value input/output device comprises a direct-current step-down circuit which is simultaneously and respectively connected with a control interface circuit, an output signal processing circuit, a signal acquisition processing circuit and a cascade circuit, the cascade circuit is sequentially connected with the signal acquisition processing circuit and the control interface circuit, the control interface circuit is provided with a signal input end, two signal output ends of the control interface circuit are simultaneously connected with the output signal processing unit and the signal acquisition processing circuit respectively, the signal acquisition processing circuit is additionally provided with a signal input end, and the output signal processing circuit is additionally provided with a signal output end. The shift-chip-based IO (input/output) device of a serial mechanism only occupies a DO6 of a controller, functions of DO8 and DI8 can be achieved by a DI1, capacity can be expanded in a cascade mode, and control of multiple points can be achieved, so that input and output channels of the controller are saved, and wiring during remote acquisition is concise while use is flexible.

The invention discloses a unitized intelligent warehouse dynamic configuration method and a terminal, and the method comprises the steps: forming a virtual unitized warehouse which comprises a plurality of different types of performance units, including a manual performance unit, an automatic performance unit, a grid performance unit and a robot performance unit; forming an order merging unit among the multiple fulfillment units, wherein the order merging unit is used for order assembly and article packaging; according to different types of performance scenes, matching the performance units matched with the performance scenes for storage, sorting and confluence operation. According to the invention, the cross-unit cross operation and the in-library order walking distance are reduced; The problems of temporary warehouse renting and temporary equipment renting in the peak season of order fulfillment and insufficient and unbalanced logistics resource utilization and even resource mismatching in the slack season of order fulfillment are solved, and the total cost of warehouse order fulfillment is finally reduced by processing orders with large quantity fluctuation and heterogeneous and changeable goods combination in a plurality of fulfillment units through collaborative optimization.

The invention discloses an adaptive adjustment method of steel production. The adaptive adjustment method comprises: selecting specification and variety of steel, and calling historical production data of the steel with the selected specification and variety; obtaining a first data group from the historical production data, wherein the first data group comprises a plurality of continuous long cycles, each long cycle comprises a plurality of continuous short cycles, each short cycle in each long cycle is numbered, and the corresponding short cycles in different long cycles have related numbers; calculating a first production model based on the first data group; selecting a pointed time point; obtaining a second data group from the historical production data, wherein the second data group comprises a plurality of continuous short cycles before the pointed time point; calculating a second production model based on the second data group; calculating the forecasting production data of the pointed time point according to the first production model and the second production module, and performing adaptive adjusting on production of steel according to the forecasting production data.

The invention relates to a face attribute recognition method and system based on a parallel shared multi-task network. The method comprises the following steps: face attributes contained in a face attribute data set are divided into a partial attribute group and an overall attribute group; a parallel shared multi-task network is constructed, and the parallel shared multi-task network is composed of a shared sub-network and two specific task sub-networks, namely a partial attribute sub-network and an overall attribute sub-network; an attention mechanism is adopted in each specific task sub-network, so that the correlation between local and global features of the shared sub-network and the specific task sub-network is utilized; a loss function is set, and a self-adaptive penalty strategy is adopted to relieve the problem of class imbalance, so that the face attribute recognition rate is improved; the parallel shared multi-task network is trained through the face attribute data set; and a face image to be recognized is input into the trained parallel shared multi-task network model to realize face attribute recognition. The method and the system are beneficial to improving the accuracy of face attribute recognition.