265 results about "SmallerThan" patented technology

The invention discloses a training device for a memristor-based neural network and a training method thereof. The neural network comprises N neuron layers which are connected one by one. The trainingmethod comprises the following steps of: inputting input data into a first neuron layer of the neural network so as to output an output result of the neural network at the Nth neuron layer, and calculating an output error of the Nth neuron layer; and counter-propagating the output error of the Nth neuron layer in a layer-by-layer manner so as to correct weight parameters between the neuron layers;and in the layer-by-layer counter-propagation process, three-valuing an output error of the mth neuron layer, and reversely inputting a voltage signal corresponding to an output result of the three-valuing operation to the mth neuron layer so as to correct a weight parameter of the mth neuron layer, wherein N is an integer greater than or equal to 3, and m is an integer greater than 1 and smallerthan N. According to the training method, the calculation ability of the memristor-based neural network is improved.

The invention discloses a transregional automatic radio station switching method for a vehicle-mounted radio, wherein a vehicle-mounted radio system consisting of a microprocessor module, a positioning module and a radio module is adopted. The transregional automatic radio station switching method comprises the steps of: locking preferential programs and establishing a preferential program file according to the habits of a user; acquiring notices of a new region in which a radio is located, in real time by the positioning module; when the signal intensity of a current radio station is smallerthan a preset threshold value, starting and searching radio stations again; recording all the radio stations with signal intensity values more than the threshold value in a frequency band into a listby the radio module, wherein the list comprises the signal intensity values, frequencies and radio station types determined according to information in a radio station database; and comparing the preference program file with the list one by one, and locking the frequency of the radio stations with the same types as that of preferential programs in the list, in the radio. According to the invention, the functions of automatic radio station searching and locking as well as transregional automatic channel recommendation based on vehicle owner habits are fulfilled for the vehicle-mounted radio.

The invention discloses a GEO and LEO cognitive satellite network and a dynamic frequency distribution method therefor. Efficient spectrum sharing can be realized, and the performances of a primary user and a secondary user are both considered. The GEO and LEO cognitive satellite network comprises at least one GEO satellite serving as a primary user and at least one LEO satellite serving as a secondary user; the GEO satellite adopts a multi-beam antenna and seven-color frequency reuse; the LEO satellite adopts a multi-beam antenna and frequency reuse, and the frequency reuse factor is smallerthan seven; downlink links of the GEO satellite and the LEO satellite share the same frequency band, so as to provide broadband fixed services for users; the beam coverage area of the GEO satellite is10 times or more the beam coverage area of the LEO satellite; and the LEO beam located in the GEO beam uses any frequency other than frequency of the GEO beam to perform the space division frequencyreuse of the LEO multi beams, and the frequency reuse factor is smaller than the frequency reuse factor of the GEO.

The invention discloses a circuit module, a dimmable LED driving circuit and a control method. According to the technical scheme in the embodiment, when a conducting angle of a triac dimmer is smallerthan a preset value, driving current is controlled to be reduced to a current value corresponding to a conducting angle signal, the smaller conducting angle is enabled to correspond to smaller driving current, so that current ripples caused by conducting angle jitter can be effectively reduced, and the problem of LED load flashing is solved.

The invention relates to a mechanical arm tail-end pose error correction method based on elliptic characteristics. The method comprises the steps of obtaining an image of a round-mouth target tank, extracting a target ellipse from the image, determining the relative position and axial relation between a tank mouth of the round-mouth target tank and an actuator according to the elliptic characteristics of the target ellipse, determining a current pose of the actuator, joint angles of all joints of a mechanical arm and the pose deviation value of the actuator according to the relative positions,the axial relation and configuration constraint of the mechanical arm, and adjusting the mechanical arm according to the pose deviation value till the pose deviation value of the actuator is smallerthan a preset value. By means of the method, the problems that single elliptic characteristic positioning has ambiguity and axial rotation poses cannot be determined are solved, accurate guidance canbe provided for the mechanical arm to collect samples and put the samples into the target tank, and control errors caused by local flexibility or joint clearance are corrected. The invention further provides a mechanical arm tail-end pose error correction system based on the elliptic characteristics.

The invention relates to an easy-to-weld container plate with the thickness ranging from 40 millimeters to 60 millimeters and the excellent center low-temperature toughness, and a making method thereof. The steel plate comprises the chemical ingredients, by mass, 0.04% to 0.09% of C, equal to or smaller than 0.40% of Si, 1.00% to 1.60% of Mn, equal to or smaller than 0.008% of P, equal to or smaller than 0.002% of S, 0.12% to 0.30% of Mo, equal to or smaller than 0.10% of the sum of Nb, V and Ti, 0.20% to 0.60% of Ni, 25 ppm to 55 ppm of N, equal to or smaller than 2ppm of H, the balance Fe and unavoidable impurity elements. Meanwhile, the steel plate meets the conditions that Pcm is equal to or smaller than 0.20%, w(Nb+V)/w(Ti) ranges from 10 to 30, the yield strength of the steel plate is larger than or equal to 490MPa, the tensile strength Rm of the steel plate is larger than or equal to 620MPa, the elongation A is larger than or equal to 21%, the yield ratio is equal to or smallerthan 0.88, the side knock power minus 50 DEG C KV2 of the position of one fourth of the thickness of the plate is larger than or equal to 220 J, and the side knock power minus 50 DEG C KV2 of the position of one second of the thickness of the plate is larger than or equal to 100 J. The method comprises the following working steps of molten iron pretreatment, converter smelting, ladle furnace refining, vacuum treatment, continuous casting, casting blank cover cooling, casting blank heating, controlled rolling, controlled cooling, stack retarded cooling, quenching, tempering, flaw detection andperformance examination. The easy-to-weld container plate with the thickness ranging from 40 millimeters to 60 millimeters and the excellent center low-temperature toughness not only has the excellenttoughness matching, but also has the excellent center low-temperature impact toughness, and thereby meeting the requirements on steel for a large spherical tank.

The embodiment of the invention discloses a neural network parameter adjustment method, device and equipment, and belongs to the technical field of computers. The method comprises the steps of acquiring parameters and a training sample of a trained neural network with the training sample comprising input data, and the parameters and the input data are floating-point numbers with the same precision; for the ith layer of network in the neural network, performing predetermined operation on the ith layer of input data and the ith layer of parameters to obtain a first operation result; Respectivelyconverting the ith layer of input data and the ith layer of parameters into fixed-point numbers conforming to decimal bit widths of the ith layer of network, and performing predetermined operation onthe converted ith layer of input data and ith layer of parameters to obtain a second operation result; And when an error between the second operation result and the first operation result is smallerthan an error threshold, determining the converted ith-layer parameter as a target parameter of the ith-layer network. According to the embodiment of the invention, the efficiency of parameter adjustment can be improved on the premise of ensuring the precision of the neural network.

The invention discloses an organic light emitting display panel and a display device. An irregular edge saw tooth phenomenon of an irregular display panel can be weakened, and the display effect nearan irregular edge of the display panel can be improved. The organic light emitting display panel comprises an array substrate and a plurality of pixel units, wherein the array substrate comprises a display zone and a non-display zone around the display zone; the display zone has an irregular edge; the pixel units are formed in the display zone of the array substrate; the pixel units comprise a plurality of sub-pixels; each sub-pixel comprises a first electrode layer, a light emitting pixel layer and a second electrode layer which are arranged in sequence; the display zone comprises an irregular display area close to the irregular edge and a central display area far away from the irregular edge; the area of the first electrode layer of the sub-pixel in the irregular display area is smallerthan that of the first electrode layer of the sub-pixel in the central display area; and/or the area of the second electrode layer of the sub-pixel in the irregular display area is smaller than that of the second electrode layer of the sub-pixel in the central display area.

The invention discloses an air conditioner and a control method and device thereof. The method comprises the following steps: after the air conditioner is started, obtaining a target temperature set by a user, and acquiring an indoor environmental temperature and an indoor environmental humidity in real time; determining whether the current indoor environmental humidity is smaller than the presethumidity; if the current indoor environmental humidity is greater than or equal to the preset humidity, controlling the air conditioner to operate in a preset maximum dehumidification output or maximum dehumidification mode; if the current indoor environmental humidity is smaller than the preset humidity, controlling the air conditioner to perform refrigeration according to the preset maximum output capability; during the refrigeration operation of the air conditioner, determining whether the difference between the current indoor environmental temperature and the target temperature is smallerthan a first preset temperature difference; and if the difference between the current indoor environmental temperature and the target temperature is smaller than the first preset temperature difference, performing down-conversion control on the air conditioner. According to the control method, a good cooling effect can be achieved in a short time, energy consumption can be reduced, and energy saving is achieved.

The invention relates to EH47 crack arrest steel for a large container ship. The chemical components of a steel plate comprise, by mass, 0.03%-0.06% of C, 0.15%-0.50% of Si, 1.50%-2.00% of Mn, smallerthan or equal to 0.008% of P, smaller than or equal to 0.0050% of S, 0.050%-0.080% of Nb, 0.020%-0.050% of V, 0.008%-0.020% of Ti, 0.80%-1.30% of Ni, 0.10%-0.30% of Cr and 0.10%-0.30% of Cu. The manufacture technology process of the steel plate comprises the steps of smelting, continuous casting, reheating, TMCP rolling and controlled cooling. The steel plate obtains fine and uniform acicular ferrite in the whole thickness, the strength is moderate, the toughness is excellent, the good brittle fracture resistance is achieved, namely, the CTOD characteristic value delta is larger than or equal to 0.15 mm at -60 DEG C, and the crack arrest performance is excellent, namely, Kca is larger than or equal to 6,500 N/mm<3/2>.