144results about How to "Solve application limitations" patented technology

The invention discloses a model parameter training method and device based on federal learning, equipment and a medium. The method comprises the following steps: when a first terminal receives encrypted second data sent by a second terminal, obtaining a corresponding loss encryption value and a first gradient encryption value; randomly generating a random vector with the same dimension as the first gradient encryption value, performing fuzzy on the first gradient encryption value based on the random vector, and sending the fuzzy first gradient encryption value and the loss encryption value toa second terminal; when the decrypted first gradient value and the loss value returned by the second terminal are received, detecting whether the model to be trained is in a convergence state or not according to the decrypted loss value; and if yes, obtaining a second gradient value according to the random vector and the decrypted first gradient value, and determining the sample parameter corresponding to the second gradient value as the model parameter. According to the method, model training can be carried out only by using data of two federated parties without a trusted third party, so thatapplication limitation is avoided.

The invention relates to an engine waste heat recovery system based on organic rankine cycle, specially a waste heat recovery system for heavy engine. The whole system comprises a plurality of heat exchanger, an energy converting apparatus (turbine power generation apparatus) and corresponding link pipe and valve, wherein the waste heat recovery unit includes a boiling device unit comprising an engine waste gas recycling cooler, an overheating unit and an engine waste gas cooling unit or the like, a reheater of refrigerant, a precooler for pressed air. The working fluid of the whole system is refrigerant such as R245fa. The present waste heat recovery system is more efficient in energy recovery then the waste heat recovery systems in prior art, the operation is more stable, the structure is more compact, and the problem of aluminum material application limitation due to double stage boosting of the high boosting of the engine is solved.

The invention provides a deception jamming detection method based on multiple antennas, aiming to solve the problems about detection and recognition of deception jamming for a satellite navigation system. The deception jamming detection method includes utilizing observed quality carrier phase capable of being provided by a receiver itself and adopting a multi-antenna direction finding process to achieve estimation of direction of arrival of satellite navigation signals or deception jamming signals, comparing the estimation result with ephemeris messages solved from the receiver so as to achieve detection and recognition of deception jamming. The deception jamming detection method based on multiple antennas utilizes the observed quality information provided from the receiver of the existing satellite navigation system sufficiently, has the advantages of simple structure, low complexity and wide application, and is a practical deception jamming detection and recognition algorithm for the field of satellite navigation deception monitoring.

The invention provides a moulded coal raw material suitable for low rank coal upgrading technology, the moulded coal raw material includes a low rank coal moulded coal body formed by low rank coal and an additive, the additive is an organic binder, air isolation destructive distillation weight loss of the organic binder at 850 DEG C is less than 50%; the moulded coal raw material also includes a non low rank moulded coal layer externally coating the low rank coal moulded coal body, the non low rank moulded coal layer is prepared from one or a plurality of medium rank coal, high rank coal or medium-high rank coal, and the adding amount of the low rank coal is 60-80wt% of the moulded coal raw material. The moulded coal raw material not only solves the problems of low using degree of a stamped coke oven and excess production capacity of coking industry in the prior art, and solves the problems of restricted application due to no large-scale industrial equipment in view of the low rank coal, after the additive is mixed with the low rank coal, effective components in the additive contribute to degradation of polymer in low rank coal at high temperature and fracture of polymer in the low rank coal so as to improve the oil and gas yield in technical process.

The invention provides a machine learning recognition method based on deep learning. Multiple times of different learning training can be performed on a machine learning model f1 with a certain quantity of known classes of multimedia data samples in different contrast sample input arranging orders, the machine learning model f1 obtained by learning is used for recognizing multimedia data classes and selects a convolutional neural network model or a fully connected neural network model, dependence on a huge number of training samples is reduced substantially, the multimedia data classes not subjected to learning training can be expanded conveniently for class recognition, the problem that practical applicability and generality are limited due to dependence of the conventional machine learning method for multimedia data classification on the huge number of training samples and failure in direct classification recognition on the classes not subjected to learning training is solved well, and the machine learning recognition method can be applied to more specific multimedia data classification occasions more extensively and more effectively.

The invention discloses a Device-to-Device (D2D) method and device, and a D2D communication control device. In the abovementioned method, the position of a first resource is determined among D2D resources in a D2D resource period; according to the position of the first resource and a predefined hopping rule, the position of a second resource is determined; and D2D signals are sent on physical resources corresponding to the first resource and the second resource, or, a D2D signal is sent on the physical resource corresponding to the second resource. According to the technical scheme provided by the invention, a circumstance that different user equipment sends and monitors a D2D signal at the same time and thus the user equipment cannot receive the D2D signals from each other is avoided, thereby improving universality of the D2D communication method.

The invention relates to an electric operating mechanism of a high-voltage isolating switch, fundamentally solving the problems of potential safety hazards and the like existing in the field welding of an output shaft and an isolating switch operating rod. The electric operating mechanism of a high-voltage isolating switch comprises a tank body as well as a mechanical driving device, an electric appliance control device and a manual-electric electrical locking device which are assembled in the tank body. The electric operating mechanism has the technical essential that a gear sleeve and a large gear which are meshed with each other form primary speed change in the mechanical driving device, the gear sleeve is fixed on an output shaft of an electric motor, a worm shaft and a worm gear which are meshed with each other form secondary speed change, one end of the worm shaft arranged in parallel to the output shaft of the electric motor is fixed with the large gear, and the end of a mechanism output shaft fixed on a worm gear, extending out of the tank body, is connected with the isolating switch operating rod through a semi-subdivision type connecting sleeve. The electric operating mechanism has the advantages of reasonable design, labor saving of manual operation, compact structure, small volume and convenient maintenance, and the mechanism output shaft and the isolating switch operating rod do not need to be welded in a field so that the safety and the reliability of field installation operations are increased, thereby obviously increasing the integral beautiful appearance of the electric operating mechanism.

The invention discloses a method for molding a coaxial dielectric filter green body. The method comprises the following steps: preparing a ceramic-based diaphragm, and cutting; disassembling the device structure into a group of plane figures from the vertical plane of a laminating direction according to the thickness of the diaphragm, and performing pattern transfer on the cut diaphragm; laminating the pattern transferred diaphragm in sequence until the total thickness of the diaphragm reaches a set device thickness, and heating and pressurizing the layered diaphragms to form a molded ceramic blank; degreasing and sintering the molded ceramic blank body to obtain the required ceramic device green body. According to the process, the problem that a complex ceramic device is hard to mold and limited in application can be solved, the thought of integral molding assisted by later process, which is followed by a traditional method is bypassed, the problems of low yield, poor operability, low flexibility, high broken rate of filter green bodies, offset out of control of a coupling structure, reduced bandpass filtering capability and the like are solved, and preparation of the ceramic device green body with complicated structure can be effectively realized.

The invention discloses a high-temperature alloy wire rod and a high-speed wire rod rolling method thereof. The high-speed wire rod rolling method comprises the following steps that blanks are prepared, and the blanks are designed to be made of GH2132 high-temperature alloy; heat treatment is carried out, the blanks are treated in three stages of preheating, heating and soaking, wherein the temperature of the preheating stage is from 960 DEG C to 1000 DEG C, the temperature of the heating stage is from 1120 DEG C to 1170 DEG C, and the temperature of the soaking stage is from 1200 DEG C to 1230 DEG C; rolling is performed at high speed, rolling is carried out in a plurality of passes, wherein the pressing amount of each pass is less than or equal to 30%, and the maximum rolling speed is 30-60 m / s; and the cooling is controlled, wherein the cooling speed is 20-25 DEG C / s. The high-speed wire rod rolling of the GH2132 high-temperature alloy wire rod with the wire rod weight greater than 1.5t is achieved, a series of defects and safety problems caused by the fact that harmful phases are separated out and the rolling force is too large are avoided, rolling cracking is effectively prevented, the tensile strength of the prepared wire rod is 600-800 MPa, the shrinkage rate of the section is larger than or equal to 50%, the grain size can reach 8.3-9.3 grade, and the problem that the application of the GH2132 high-temperature alloy wire rod is limited due to the wire rod weight and the rolling efficiency is solved.

The invention discloses N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide and a preparation method and an application of the N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide and preparation method and application. The structural formula of the N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide is shown in the description. The preparation method comprises the following steps: enabling formaldehyde and an amino on the chitosan oligosaccharide to react to obtain N-methylene chitosan oligosaccharide, enabling 3-chlorine-2-hydroxypropyl trimethyl ammonium chloride and a hydroxyl on C6 site of the N-methylene chitosan oligosaccharide to react to obtain N-methylene-O-quaternary ammonium chitosan oligosaccharide, finally enabling thiosemicarbazide and a carbon-nitrogen double bond of the N-methylene-O-quaternary ammonium chitosan oligosaccharide to have an addition reaction to obtain the N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide. The N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide solves the problems that chitosan oligosaccharide and chitosan derivatives are poor in adsorbability and are easy to dissolve and corrode, and can be used in acid, neutral and alkaline solutions. Moreover, the antibacterial activity of the N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide is improved, and the N-thiosemicarbazide-O-quaternary ammonium chitosan oligosaccharide has excellent effects on adsorbing metal ions and inhibiting bacteria.

The invention belongs to the field of intelligent materials, and discloses a preparing method of a graphene aerogel/polyurethane foam conductive composite. The method includes the following steps of preparing and then mixing graphene oxide dispersing liquid and a chitosan solution, filling hole foam with a mixed solution in a vacuum state, and preparing the finished graphene aerogel/polyurethane hole foam composite product through the measures of freezing treatment, freeze-drying and thermal treatment. The method has the advantages that by introducing high-conductivity high-piezoresistance-sensitivity graphene/chitosan aerogel into the polyurethane hole foam with excellent mechanical property, the excellent conductivity and piezoresistance of the graphene/chitosan aerogel are realized, meanwhile it is ensured that the composite has the excellent mechanical property, the problem that application is limited due to the poor mechanical strength of the aerogel is solved, and the prepared composite has conductivity, high piezoresistance sensitivity, high compression resilience, high durability and the like.

The invention provides a storage method, device and system based on a block chain and a storage medium. The method comprises the steps that a data node stores storage information to a storage node, and then the data node obtains a first Merkle tree and a second Merkle tree; Wherein the first Merkle tree is obtained by the data node according to the storage information; wherein the second Merkle tree is obtained by the storage node according to the storage information; Therefore, the data node verifies the second Merkle tree by using the first Merkle tree, if the verification is passed, the data node broadcasts a first broadcast message on a block chain, and the first broadcast message is used for indicating the mining node to generate a new block according to the second Merkle tree. According to the method provided by the invention, the problem that the application of the block chain-based evidence storage system is limited due to the storage pressure can be solved to a certain extent.

The invention relates to a manufacturing method for a composite tailplane of small and medium size unmanned aerial vehicles. According to the method, the problem of limitation to the application of the present integrally molded tailplane structure is solved, the operability, airtightness and demolding property during the molding process of the product are promoted, and the complexity of integrallymolding assembling and the application scope of integral molding for the tailplane structure are effectively improved. The potential quality hazard in the present integral molding is effectively reduced, the quality of the integrally molded tailplane is more stable and the reliability of the products is effectively promoted; the integrally molding assembling is simplified and the molding technology is simplified; the performance stability and size stability of the integrally molded product are guaranteed, the design/manufacturing integration is realized, the technological difficulty is reduced and the production cycle is shortened.

The invention relates to the field of fresh keeping of meat products and in particular relates to a fresh-keeping method of a pre-conditioning meat product. The pre-conditioning meat product is treated through adopting a combination method of eugenol and tea polyphenol so that the quality guarantee period of the pre-conditioning meat product is remarkably prolonged and the umami of the pre-conditioning meat product is improved; the fresh-keeping method of the pre-conditioning meat product, which can be used for remarkably prolonging the fresh-keeping period of the pre-conditioning meat product and is beneficial for health of human bodies is obtained.

The invention belongs to the field of chemical synthesis and discloses soluble polyimides containing phenolic hydroxyl group and a preparation method and application thereof. The soluble polyimides containing phenolic hydroxyl group has a structure as shown in the formula (I). The invention also discloses a preparation method of the soluble polyimides containing phenolic hydroxyl group. By high temperature and by the use of a catalyst, cyclization efficiency is enhanced, and reaction time is shortened. As the obtained polyimides contain lots of flexible groups and large volume of phenyl ring side-group, the product has high molecular weight and is chemically soluble. The problem that polyimides have poor dissolvability and application of polyimides is limited is solved. On this basis, it is also easy to process a polyimide film, and the application range is widened. By introducing phenolic hydroxyl group into PI molecular chain, thermal property is not influenced, and compatibility or adhesive property of the polyimide material with other materials is improved. The product can be applied to electronic packaging, film materials and adhesives.

The invention provides a machine learning recognition method based embedded coding learning, and the method comprises the steps: carrying out the different learning training of a machine learning model f1 for many times through a certain number of multimedia data samples of a known type and different contrast sample input arrangement sequences, so as to carry out the type recognition of the multimedia data. The machine learning model f1 is designed to be a combined model framework of a coding function model and a convolutional neural network model or a full-connection neural network model, thereby greatly reducing the dependence on the mass training samples. Moreover, the method can be conveniently extended for the type recognition of the multimedia data which is not trained, solves problems that the actual application and universality are limited because a conventional multimedia data classification machine learning recognition method depends on the mass training samples and cannot directly achieve the classification recognition of the types which are not trained, and can be applied to more specific multimedia data classification application occasions more widely and effectively.

The invention discloses a novel biological membrane water purifying and fountain aeration coupling device. The device is characterized by comprising an organisms gathering tower, a fountain aeration system and a fixing system, wherein the organisms gathering tower comprises a round-cake-shaped hollow base, a circular-truncated-cone-shaped hollow tower body, a hydrophilic biofilm colonization carrier material and an anaerobic zone annular cover; the fountain aeration system comprises a stainless steel sleeve with screen holes, a submersible pump, a submersible pump fixing device, a plastic hose, a submersible pump power supply cable, a water conveying pipe and a fountain head; the fixing system comprises a stainless steel anchoring rod, a nylon soft rope and annular binding holes. The submersible pump arranged in a polluted river water body is used for lifting river water which is low in dissolved oxygen content, the river water makes full contact with air after being sprayed by the fountain head, the oxygen-enriched river water is sprayed to the organisms gathering tower, and the aerobic-anaerobic biological membrane purified river water is formed by means of microorganisms, algae, protozoa and the like.

An amphipathic luminescent molecule with an AIE (aggregation-induced emission) characteristic is characterized in that a chemical structural formula of the luminescent molecule is shown in the specification, wherein X and Z are integers from 0 to 100, Y is an integer from 5 to 500, and each of an R1 group and an R2 group is any one of a linear or branched alkyl group, a naphthenic base, a heterocyclic alkyl group, an alkoxy group, an alkylamino group and halogenated alkyl group. A synthetic method comprises the following steps: polyether amine which is taken as a reaction initiator has a heating reflux reaction with a grafting agent in nitrogen atmosphere, a brown liquid product is obtained, washed with an organic solvent, filtered and dried, and a target product is obtained. The luminescent molecule is dissolved in water, and an autofluorescence nano-micelle is formed. The invention further provides an application of the autofluorescence nano-micelle. The problem of limited application of luminescent materials with the AIE characteristic due to hydrophobicity and toxicity in the prior art is solved.

The invention provides a coking technology for briquettes containing low-rank coal, garbage and biomass. The technology comprises loading low-rank coal and garbage into a bundle-shaped material formed through packing of biomass, and then performing compacting treatment on the obtained material with a core-shell structure under specific conditions, so as to prepare the briquettes containing the low-rank coal, the garbage and the biomass. Therefore, the low-rank coal, garbage and biomass raw materials can be pushed into a stamp-charging coke oven for high-temperature dry distillation, and also, solid products generated after dry distillation of the low-rank coal, the garbage and the biomass can be easily pushed in the stamp-charging coke oven, and thus a considerable amount of raw gas, tar and carbon residues can be obtained. The technology of the invention not only can reduce the raw material cost of oil gas and coke products, but also can effectively realize resourceful utilization of garbage and biomass, and accordingly has both considerable economic benefits and certain social benefits.

The invention discloses a machine learning recognition method based on embedded coding and contrast learning, and the method comprises the steps: carrying out the different learning training of a machine learning model f1 for many times through a certain number of multimedia data samples of a known type and different contrast sample input arrangement sequences, so as to carry out the type recognition of the multimedia data. The machine learning model f1 is designed to be a combined model framework of a coding function model and a convolutional neural network model or a full-connection neural network model, thereby greatly reducing the dependence on the mass training samples. Moreover, the method can be conveniently extended for the type recognition of the multimedia data which is not trained, solves problems that the actual application and universality are limited because a conventional multimedia data classification machine learning recognition method depends on the mass training samples and cannot directly achieve the classification recognition of the types which are not trained, and can be applied to more specific multimedia data classification application occasions more widelyand effectively.