7results about How to "Increase data dimension" patented technology

The present disclosure provides an object recommendation method, an object recommendation device, an electronic device and a storage medium, which can be applied to the technical field of big data. The object recommendation method comprises: determining M first objective indexes, N second objective indexes and a target variable relationship matched with an application scene, the target variable relationship being used to determine a to-be-recommended object, the first objective indexes being used to represent transaction characteristics of a user, the second objective indexes being used to represent social relationship characteristics of the user, M being greater than or equal to 1, and N being greater than or equal to 1; obtaining a recommendation data set of a target user, the recommendation data set comprising first objective data corresponding to the first objective indexes and second objective data corresponding to the second objective indexes; and determining the to-be-recommended object according to the M first objective data, the N second objective data and the target variable relationship.

The invention is suitable for the technical field of laser additive manufacturing, and provides a laser additive multi-mode monitoring device which comprises a first mounting plate assembly, a second mounting plate assembly, a powder bed monitoring module, an OT camera monitoring module, a structured light camera module, an infrared monitoring module, a printing substrate and a supporting shell. The supporting shell is of an external bearing structure, and the first mounting plate assembly and the second mounting plate assembly are transversely arranged along preset mounting positions on the top of the supporting shell and are fixedly connected with the supporting shell through detachable fixing pieces. According to the method, multi-modal data is fully covered, and a complete process data link is constructed; acquisition precision and stability are obviously improved, and requirements of severe working conditions are met. And the process optimization of data closed-loop driving is realized, and the forming quality and the large-scale production capacity are improved.

The present application relates to a distribution line protection method, device and electronic equipment, and to the field of circuit protection technology. The method includes: obtaining a voltage waveform data array on both sides of the distribution line, and a current waveform data array on both sides of the distribution line; filtering the current waveform data array to obtain a filtered current waveform data array; determining the voltage qualification rate corresponding to the voltage waveform data array, and determining the waveform similarity corresponding to the filtered current waveform data array when the voltage qualification rate at both ends of the distribution line meets the discrimination range; determining the section discrimination criterion of the distribution line based on the waveform similarity and the discrimination threshold, and determining the protection strategy of the distribution line based on the section discrimination criterion and the overcurrent protection criterion of the distribution line; the discrimination threshold is determined after simulating the waveform similarity of the distribution line under different fault conditions based on the distribution network simulation model. The use of this method can improve the safety of the distribution line.

This invention discloses an environmental health analysis method and system based on egg-laying hen farming, relating to the field of livestock environmental monitoring technology, including methods based on NH3 concentration and VOCs. S Concentration and NO X Two-dimensional empirical wavelet transform is performed on the concentration to calculate texture intensity, gradient direction, and local entropy. Spatial texture features are obtained by horizontal stacking based on NH3 concentration, HRV standard deviation, and VOC. S Concentration and NO X Concentration is calculated, time-domain features are calculated, and frequency-domain features are calculated based on heart rate and HRV standard deviation. Weighted summation yields the SPS joint features. This invention improves data dimensionality and monitoring comprehensiveness by collecting multimodal data and calculating spatial texture features, thereby enhancing the accuracy of environmental dynamic analysis. Through SPCA dimensionality reduction and weighted risk scoring models, it provides accurate quantitative evidence for the environmental health status of laying hens, improving the health management level and production efficiency of laying hen farming.

The invention relates to the technical field of enterprise management, in particular to an enterprise portrait classification system based on a block chain technology, and the system comprises a distribution module which is used for creating block chain distributed digital identities and distributing the digital identities to enterprise users; the interaction platform is used for verifying the digital identity information of the enterprise user, allowing the enterprise user to log in and requesting the state information of the enterprise user; the request module is used for controlling the interaction platform to send an enterprise user state information request to an enterprise user logged in the platform; and the encryption transmission module is used for monitoring the enterprise user state information in the request module to solve whether the enterprise user state information is agreed by a corresponding online enterprise user in an interaction platform, and under the agreed state of the enterprise user, the block chain technology is utilized to create a block chain distributed digital identity for the user, and asymmetric encryption and data chaining modes are utilized to improve the encryption efficiency of the user. And data flow records are solidified, and credible and reliable data traceability is realized. And a solid foundation is laid for improving the data quality of the system.

The utility model discloses a magnesia carbon brick performance test bench, which is characterized in that a magnesia carbon brick is a magnesia carbon brick body; comprising a first testing box, a second fixing block arranged above the first testing box, moving mechanisms fixedly connected to the two sides of the second fixing block, a lifting mechanism fixedly connected to the upper end of the second fixing block, a bearing mechanism fixedly connected to the lifting end of the lifting mechanism and a magnesia carbon brick body placed in the bearing mechanism. The acid-alkali liquid soaking mechanism is arranged outside the first testing box, the pressure mechanism is arranged on the side wall of the acid-alkali liquid soaking mechanism, and a camera is fixedly connected to the side wall of the second fixing block. The utility model has the advantages of integration and high efficiency.

The invention relates to the technical field of artificial intelligence, and provides a multi-scale correlation time sequence prediction method, system, device, product and medium, and the method comprises the steps: obtaining an input sequence, splitting the input sequence to obtain a plurality of single-dimensional input sequences, and carrying out the calculation to obtain a stationary time sequence; obtaining a position coding enhancement vector, performing cutting to obtain a plurality of slice time sequences, and obtaining a time sequence vector through the slice time sequences and the position coding enhancement vector; constructing a time encoder of a multi-head attention mechanism to obtain a plurality of time encoding sequences; constructing a stage conversion layer to obtain a plurality of feature time sequences; constructing a variable encoder of a multi-head attention mechanism, and calculating to obtain a plurality of variable encoding sequences; and obtaining a plurality of single-dimensional prediction results through recombination and expansion projection processing, obtaining a plurality of target single-dimensional prediction results through the plurality of single-dimensional prediction results, and fusing the target single-dimensional prediction results to obtain a target prediction result. According to the invention, the precision of time sequence prediction is effectively improved.