33results about How to "Easy and flexible implementation" patented technology

The invention discloses a method for online generating a tobacco leaf curing process curve for a bulk curing barn. The method comprises the following steps: firstly, setting a curing process curve in the initial stage, and starting the curing process curve parameter prediction in the next stage 5 hours before the stage is finished; then, collecting curing barn gas signals once every one hour, and extracting the odor characteristics, inputting the characteristics to a trained RBF neural network, and predicting and outputting a curing process curve parameter in the next stage to form the curing process curve in the next stage; and repeating the above steps until curing is completed so as to realize online generation of the tobacco leaf curing process curve for a bulk curing barn. According to the method, the process for artificially determining the tobacco leaf curing state is converted into intelligent identification by a machine, the curing process curve can be generated online in real time, and the curing process is accordingly conducted, the system does not need to preset a curing process curve, and the whole process does not have artificial participation, so that artificial cost for the curing system is reduced, and artificial subjective factor interference can be avoided.

The invention discloses an evaluation system for pressure regulation and leakage reduction of urban water supply pipe networks. The system of the present invention comprises a monitoring module of a lower computer, a wireless communication module and a data receiving and evaluating module of an upper computer. The monitoring module of the lower computer is signal-connected with the data receiving and evaluating module of the upper computer through the wireless communication module. The lower computer adopts the operation scheme of alternating pressure regulation, and the evaluation of the upper computer comprehensively considers the characteristics of the average value, median value, and cumulative probability distribution of flow and pressure. Compared with the prior art, the present invention has the advantages of unattended automatic operation, convenient remote setting of pressure control schemes, elimination of interference from work and rest days, weather, etc. on users' water consumption, and improved comprehensiveness and scientificity of effect analysis.

The invention discloses a wood defect recognition and classification method based on multi-features. Firstly, the experimental sample data is selected; then the stress wave propagation velocity data is collected; then the resistance value is collected; and then the stress wave propagation velocity data and the resistance value are data processed; Then, several groups of sample data of each type of wood defects are divided into training group data and classification group data, and finally the support vector machine method is used for training and classification. Using the stress wave wood non-destructive tester to collect the data characteristics of the stress wave propagation velocity and the micro-drilling resistance meter to collect the characteristics of the resistance value can better characterize the global characteristics of the wood cross section. Using these two wood data features and the support vector machine method Identifying and classifying the defects inside the wood can accurately classify the types of wood defects, and has high reliability, and the method is convenient and flexible to implement, and can save instrument costs.

The invention provides a data prefetching method based on MapReduce, which belongs to the technical field of computers. The method of the present invention predicts the data block processing capacity of each computing node through performance evaluation, and evaluates which computing nodes will have non-localized tasks according to a series of calculations. When applying for processing the task, it will be prefetched to the local computing node in advance, so that the computing node will not generate computing waiting. The purpose of the present invention is to improve the execution efficiency of the MapReduce job and the overall throughput of the system, so that the computing nodes do not need to wait for the remote transmission of data blocks, and improve the utilization rate of the computing nodes. The prefetching method proposed by the present invention can work in homogeneous and heterogeneous MapReduce cluster environments at the same time, and the idea of ​​prefetching is not only for MapReduce, but all distributed computing frameworks can be used for reference and improved.

The invention relates to a chip package structure and a packaging method thereof. The structure comprises at least one chip, wherein each chip is provided with a plurality of first contact pads and second contact pads, a lead frame, a group of first bonding wires and a group of second bonding wires, wherein the lead frame comprises a plurality of pins for external connection; the chip is arrangedon the lead frame; the group of the first bonding wires are used for directly and electrically connecting the first contact pads to the lead frame; and the group of the second bonding wires are used for electrically connecting the second contact pads to the plurality of pins of the lead frame. The structure and the method can conveniently realize the chip with low consumption and easy heat dissipation, reduce packaging size of the chip, and contribute to the function expansion of the chip.

The invention proposes a MapReduce task scheduling algorithm that can work simultaneously in homogeneous and heterogeneous cluster environments, belonging to the field of computer technology. The scheduling algorithm can comprehensively consider the processing performance of each computing node in the cluster, abstract the computing nodes and computing tasks into a bipartite graph, and form the final global task scheduling scheme by appropriately expanding the bipartite graph and combining the KM weighted optimal matching algorithm. Experimental data show that the scheduling algorithm can improve the data localization degree in the Map stage to nearly 100%, and the overall execution time of the MapReduce job can be reduced by 67.1%.