40 results about "Algorithm transformation" patented technology

The invention relates to a deep defect quantifying method based on an eddy rotating probe signal. The method comprises the steps of 1, acquiring a calibrated defect impedance variation vector signal; 2, performing differential treatment for the defect signal; 3, normalizing the signal; 4, forming a calibration quantifying curved line; 5, acquiring an actual signal; 6, performing deep defect quantifying for the actual signal through the calibration quantifying curved line. With the adoption of the method, the problem of failure of quantifying of a rotating probe can be solved; the algorithm transformation is performed to solve the drifting problem of an absolute signal.

The invention relates to a key storage method. The method comprises the following steps: generating a true random number by using a hardware random number generator, wherein the true random number isused as an encryption factor of the present storage; generating a seed by using the true random number, and substituting the seed into a linear feedback shift register to obtain a pseudo random number; performing an xor operation on a master key and the pseudo random number to obtain a random master key, and then encoding the random master key by using an error correcting code algorithm to obtainencoded data; performing private algorithm transformation on the true random number to obtain intermediate data, wherein the intermediate data are used as error locations of data to be stored to tamper the encoded data to obtain error code data; and performing encrypted storage on the error code data. According to the key storage method provided by the invention, the generated random number is hidden in the data to be protected, thereby improving the intensity of data randomization. In the case of limited storage resources, in order to prevent brute force attacks after ciphertext leakage, thedata are mixed with error codes, and the data are encrypted by using a private algorithm, so that the key storage is more secure.

The invention discloses a mixed type scheduling method and system for a cloud platform wide node scientific workflow. The method comprises the following steps that: when each scheduling stage is initialized, utilizing bond energy algorithm transformation to construct a task into a task multilayer clustering tree structure; in a global initialization stage, utilizing the bond energy algorithm transformation and basic dichotomy multi-layer division to construct scheduling resources into a cloud resource multilayer tree structure; according to the mapping deadline of a task group and the each-dimension resource distribution of scheduled resources, carrying out a mapping operation on the node of a task subtree in the multilayer clustering tree structure and the node in a cloud resource multilayer tree structure; at the scheduling tail of the stage, judging whether the amount of tasks which finish the mapping operation exceeds a set threshold value or not; and if the amount of tasks which finish the mapping operation exceeds the set threshold value, classifying tasks which do not finish mapping to a next scheduling stage, and re-distributing the deadline of residual tasks. By use of themethod or the system, the scheduling method of the dynamic scheduling and the static scheduling of dual features is fused, and the energy cost optimization of a cloud resource provider is realized onthe basis of meeting user execution time requirements.

The invention provides a spherical display image processing method and system, and the method comprises the steps: enabling a to-be-displayed image to serve as a texture map and to be pasted on a model or a curved surface, and rendering the model or curved surface till a cubic texture or directly carrying out rendering; building a target image of a plane for spherical display according to the cubic texture, wherein the corresponding relation between the target image and the cubic texture is that the target image is located in a polar coordinate system and the pixel coordinates corresponding to each pixel of the target image are expressed through polar coordinates; enabling the polar coordinates of each pixel to be correspondingly converted into a three-dimensional coordinate point on the a spherical surface in a three-dimensional coordinate system, and enabling the color value, obtained through the sampling of the cubic texture at each three-dimensional coordinate point, of each point to be given to a corresponding pixel of the target image; carrying out the algorithm transformation in an actual implementation process with the help of a top point shader of a graphic platform directx or opengl, and achieving the high-efficiency and high-quality conversion output through a graphic display card, thereby enabling display to have no blind region, enabling display to be flexible, and enabling the operation efficiency to be high.

The invention mainly relates to the field of non-contact large rotary mechanical high-speed rotating blade parameter measurement technologies and intelligent fault diagnosis systems, in particular to a method for automatically and synchronously positioning the high speed blades \by directly transforming and matching signals output by the conventional non-contact sensor. The method is suitable for a parameter non-contact online measurement system for most of rotary mechanical rotor blades. In order to fulfill the aim, the technical scheme provided by the invention is that: a device for automatically and synchronously positioning the high-speed rotating blades comprises a non-contact sensor for measuring a blade arrival signal, a high-precision blade tip timing module for receiving the signal output by the non-contact sensor to obtain a blade arrival time sequence {t}, and a signal processing module for performing algorithm transformation to obtain an included angle sequence {phi}. The method and the device are mainly applied to large rotary mechanical high-speed rotating blade parameter measurement occasions.

The invention belongs to the information safety field, which relates to an encryption method based on algorithm transformation. The password (encryption) algorithm is uncertain and changeable, some calculation components in the password algorithm are random and one of n functions or calculation methods, some structures of the algorithm are uncertain, or the order of some calculation components is changeable, and all the uncertainties are determined by one part of data in the key. The encryption method has the following advantages: because the uncertain algorithm, the password analysis is very difficult, which caused by leakage of the algorithm information; the current password analysis aims at the certain algorithm, so when the algorithm is uncertain, the password analysis is difficult to execute, simultaneously the types of the algorithm are diversified; and the calculation quantity is not greatly increased, so the key space is greatly improved under the condition that the calculation quantity is not obviously increased, and the uncertainty and the complexity of the password system are improved to improve the safety.

The embodiment of the invention provides an algorithm transplantation method and device, relates to the field of artificial intelligence, the method specifically comprises the following steps: obtaining a first algorithm used for processing the image into a format meeting the input requirement of the deep learning model in a cloud or a server, a first deep learning model used for outputting an identification result by utilizing the processed image; converting the first algorithm into a target algorithm supported by the target embedded device; converting the first deep learning model into a target deep learning model supported by target embedded equipment, and then the target algorithm and the target deep learning model are transplanted to the target embedded device, so that the target embedded device can realize convenient image recognition based on the target algorithm and the target deep learning model without depending on a network and a server.

The invention discloses a compiling method and device suitable for a storage and calculation integrated computer device, computing equipment and a storage medium. The compiling method comprises the steps of obtaining operation information of a to-be-compiled algorithm; converting the to-be-compiled algorithm into a first intermediate representation according to the operation information, the firstintermediate representation comprising operator operations executed by the storage and calculation integrated computer device; mapping the first intermediate representation into a second intermediaterepresentation, wherein the second intermediate representation comprises a corresponding relationship between hardware information and operator operation of the storage and calculation integrated computer device; and according to the hardware information, compiling the to-be-compiled algorithm into instruction information identified by the storage and calculation integrated computer device, so that the storage and calculation integrated computer device executes the instruction information. According to the compiling method, the operation information can be compiled into an instruction which can be directly executed by the storage and calculation integrated computer device, so that the effect of performing accelerated operation on various algorithms by utilizing the storage and calculationintegrated computer device can be realized.

The invention relates to a machine learning algorithm-oriented prediction and training process unification method and apparatus. The method comprises the steps of obtaining a first machine learning algorithm, wherein the action object of the basic primitive in the first machine learning algorithm is a tree structure which cannot be accessed in an index mode; carrying out calculation flow transformation on the basic primitive, so that the action object of the transformed basic primitive is a tree structure accessed in an index mode; based on the transformed basic primitive, converting the first machine learning algorithm into a second machine learning algorithm satisfying a data privacy protection protocol, wherein the action object of the second machine learning algorithm is a tree structure accessed in an index mode, and the tree structure comprises internal nodes, leaf nodes and position indexes of the nodes; the second machine learning algorithm is used for the prediction task based on the tree structure. According to the invention, unification of a prediction process and a training process can be realized in a non-intrusive manner.

The invention discloses an Internet of Things signal anti-interference transmission system for breeding. The Internet of Things signal anti-interference transmission system is characterized by comprising a signal processing module, a data processing unit and a protocol processing unit which are connected in sequence, wherein the signal processing module is used for reading analogue quantity signals or communication protocols with different types, which are input by each agricultural environment sensor, carrying out digital-to-analogue conversion or communication protocol conversion and then transmitting converted digital quantity information to the data processing unit; the data processing unit is used for carrying out algorithm transformation on the received digital quantity information.According to the Internet of Things signal anti-interference transmission system for breeding, the wire arrangement cost and the post-period maintenance coat can be greatly reduced; furthermore, the deployment cost and difficulty are simplified and a lightning stroke problem is avoided; therefore, the Internet of Things signal anti-interference transmission system has very good operability and practicability, and an application level of an agricultural Internet of Things sensor can be improved.

The invention provides a question and answer method, system and device based on a water conservancy project knowledge graph and a medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: converting water conservancy project unstructured data into structured data through an algorithm to construct an atlas triple, and synchronously storing the atlas triple to ES and Neo4j; establishing an MCP service to access a custom atlas service, and configuring a plug-in to connect Neo4j and a retrieval service; matching fuzzy knowledge of the question in the ES is obtained, a Neo4j mode is retrieved to filter irrelevant nodes, and query nodes and relations are judged; a Cypher statement is generated and verified; iteratively calling a service to obtain node relation data, identifying a relation type, judging a dynamic relation according to a calling rule, and filtering to generate the node relation data; and finally listing entities, relationships and attributes, and feeding back a natural language result of the user. According to the method, the utilization rate, query accuracy and reuse efficiency of water conservancy knowledge are effectively improved, and efficient support is provided for water conservancy decision making.

The invention discloses a nonlinear underwater image gain algorithm for hyperbolic tangent deformation function transformation. The problems of obvious noise pollution, low contrast ratio, low brightness, low saturation, color distortion and the like of underwater optical images due to water transparency, light and the like can be obviously improved through algorithm transformation of the invention. Firstly, an underwater grating image file is read as matrix data, three layers of matrix data of RGB three channels are extracted respectively, a hyperbolic tangent deformation-based function with an expansion coefficient k is constructed, the gain degree and the key gain color gradation range can be controlled through the appropriate expansion coefficient k, and a hyperbolic tangent deformation function is constructed; the function is used to carry out gain transformation on three layers of matrix data of three channels of RGB, and then the three color channels are merged and converted into a new image. The underwater optical image is processed through the algorithm, the contrast and brightness of the underwater image are improved, important color gradation is reasonably redistributed, and the dark area and the color gradation are improved while detail information of the image is reserved and even enhanced.

The invention discloses a mixed type scheduling method and system for a cloud platform wide node scientific workflow. The method comprises the following steps that: when each scheduling stage is initialized, utilizing bond energy algorithm transformation to construct a task into a task multilayer clustering tree structure; in a global initialization stage, utilizing the bond energy algorithm transformation and basic dichotomy multi-layer division to construct scheduling resources into a cloud resource multilayer tree structure; according to the mapping deadline of a task group and the each-dimension resource distribution of scheduled resources, carrying out a mapping operation on the node of a task subtree in the multilayer clustering tree structure and the node in a cloud resource multilayer tree structure; at the scheduling tail of the stage, judging whether the amount of tasks which finish the mapping operation exceeds a set threshold value or not; and if the amount of tasks which finish the mapping operation exceeds the set threshold value, classifying tasks which do not finish mapping to a next scheduling stage, and re-distributing the deadline of residual tasks. By use of themethod or the system, the scheduling method of the dynamic scheduling and the static scheduling of dual features is fused, and the energy cost optimization of a cloud resource provider is realized onthe basis of meeting user execution time requirements.

The invention discloses a clinical medicine practical skill multi-dimensional evaluation and analysis system, and relates to the technical field of clinical medicine education and skill evaluation, and the system comprises a multi-source data collection module which obtains operation and other multi-dimensional data; the data standardization preprocessing module unifies formats, corrects exceptions, encrypts and desensitizes; the data intelligent quantitative calculation module converts a quantitative result by using an algorithm; the multi-dimensional dynamic evaluation module constructs a three-dimensional model, sets three stages of periods, and combines AI and manual output scores; the personalized feedback module generates a report and a scheme and supports multi-scene data linkage; according to the method, multi-source data is integrated, the limitation of single data of traditional evaluation is broken through, the difficult-to-quantify ability is converted into objective indexes through an intelligent algorithm, a three-dimensional evaluation model and a three-level evaluation period are constructed, skill development is dynamically tracked, AI and manual rechecking are fused, efficiency and accuracy are balanced, and a personalized scheme and a multi-scene report are generated. And clinical medicine evaluation upgrading is promoted.

The invention discloses an artificial intelligence model training method and system based on a block chain, and relates to the technical field of block chains, and the method comprises the steps: collecting physical position coordinate data, and converting the data into a position coding character string through a position coding algorithm; generating an adjacent target training node list according to a prefix matching rule of the position coding character string, and triggering the target training nodes through block chain transaction to create a local training group; the target training node initiates real-time training gradient exchange in the local training group; when the exchange fails, requesting historical gradient data associated with the position coded character string from a gradient bank; the artificial intelligence model is updated by using historical gradient data, and a new training gradient is generated after updating is completed; and storing the new training gradient to a partition position corresponding to the position coding character string. According to the method, the artificial intelligence model is updated through the historical gradient data associated with the position coding character string to generate the new training gradient, and self-recovery of model iteration in a communication interruption scene is realized.

The invention is adaptable to the technical field of automation of electronic design, and provides a method for algorithm transformation. The method includes converting a C-language program or flowchart of the C-language program into a state graph, simplifying the state graph to obtain the simplified state graph including control logic and arithmetic logic, separating the arithmetic logic from the simplified state graph, implementing the arithmetic logic through hardware and determining the implemented result as a data path, separating the control logic from the simplified state graph, implementing the control logic through hardware and determining the implemented result as a controller, and implementing the data path and the controller by hardware description language HDL. Since the C-language is converted into the state graph to construct the data path and the controller which are implemented by the HDL, poor universality in conversion from the C-language to the HDL hardware implementation is overcome.

The disclosure provides a machine learning algorithm conversion system, method, and electronic equipment, which relate to the field of computer technology. The system includes: a programming interface layer, which is used to construct a data flow diagram of an original machine learning algorithm based on a preset data flow generation tool; The operator placement evaluation module is used to calculate the corresponding placement cost of each operator in the data flow graph when executed by different participants; the data flow graph split scheduling module is used to divide the data flow graph into multiple sub-parts according to the placement cost graph, and dispatch the subgraph to the target participant for execution; the compilation execution layer is used to compile the subgraph into a new data flow graph based on the greedy algorithm strategy, and generate the instructions of each operator in the new data flow graph to obtain the distributed Privacy Preserving Machine Learning Algorithms. The disclosure can effectively improve the problems of poor compatibility, high coupling degree and poor performance existing in the algorithm conversion process.

The invention provides a system for saving and analyzing log data, and the system comprises a log data compression and conversion part which is used for compressing high-frequency data appearing in the log data or using a conversion function to compress the high-frequency data to obtain a new log data structure and storing the new log data structure; a log analysis part used for rearranging, iteratively accessing and sequencing the converted log data by using a specified algorithm; and a log data decompression and restoration part that restores the original log data using inverse transformation of the algorithm transformation used by the log data compression and conversion section. Logs are stored by utilizing the hash table through an optimized storage structure, so that the processing time of the business logs can be greatly shortened, and the optimization problem of query and analysis of big data is solved.

The invention belongs to the technical field of FPGA hardware algorithm acceleration, and discloses an algorithm acceleration method and system for large number modular exponentiation, a medium, a device and application. The system comprises a Montgomery algorithm module, an R-L mode modular exponentiation algorithm module, a top layer module, a serial port communication module, a modular exponentiation state control module, a modular multiplication state control module and a modular multiplication operation module. According to the present invention, various operations are completed through various functional modules and the corresponding design schemes, the form of converting an algorithm into modules is solidified into the functional modules on various circuits, and higher stability and compatibility are achieved; by utilizing the parallelism of an FPGA, the large number modular exponentiation algorithm is greatly accelerated; a special low-complexity Montgomery operational circuit is used for real-time processing, so that the characteristics of high density, high reliability and low power consumption are realized; a IP core is packaged, and the algorithm has the very good transportability and high flexibility.

The invention provides an algorithm conversion method and system in a heterogeneous many-core processor, and the technical scheme of the method comprises the steps: predefining a data structure of a C + + algorithm recognition library in a main core of the heterogeneous many-core processor, predefining a statement recognition rule and a grammar mapping rule for converting a C + + algorithm into a C language, and carrying out the grammar analysis of a C + + source code through an ANTLR technology, forming an abstract syntax tree; traversing the abstract syntax tree, analyzing a C + + algorithm definition file, and constructing an algorithm recognition library according to a data structure of a predefined C + + algorithm recognition library; traversing the abstract syntax tree again, and mapping a C + + source code algorithm statement to a libcstl based on a C language according to a syntax mapping rule; and reconstructing the C + + source code algorithm statement according to the grammar rule of the libcstl, and converting the C + + algorithm into the libcstl of the C language. The problems that an existing STL algorithm is complex in implementation and complex and time-consuming in reconstruction, and a manual writing mode is low in efficiency and high in cost are solved.

The invention relates to a parameter issuing method, device and equipment for ultrasonic equipment and a computer storage medium, and the method comprises the steps: receiving an adjustment instruction, and converting the adjustment instruction into parameter data according to a preset algorithm; putting the parameter data into a pre-registered linked list, and triggering an adjustment identifier corresponding to the parameter data in the linked list; and traversing the adjustment identifier in the linked list, and issuing the parameter data corresponding to the triggered adjustment identifier to a scanning controller of the ultrasonic equipment. According to the invention, the algorithm conversion of the adjustment instruction is separated from the input and output control, so that the parameter data issuing efficiency is improved.

The embodiment of the invention provides a forage grass knowledge graph construction method and device based on cross-document information fusion and a storage medium, and belongs to the technical field of knowledge graph construction. The construction method comprises the following steps: converting paragraphs of each document in acquired multi-source forage grass document data into paragraph vectors in combination with a TF-IDF algorithm, and screening out highly related candidate paragraphs according to the paragraph vectors and given query vectors; according to the forage grass related information extracted from the documents to which the candidate paragraphs belong, a reasoning chain is constructed to guide cross-document forage grass related information fusion; and based on a thinking chain prompt template, a knowledge extraction model is guided to perform entity relationship extraction on the fused forage grass text data, and then a forage grass knowledge graph is constructed. According to the knowledge graph construction method provided by the invention, information extracted from a plurality of documents can be effectively associated and integrated, so that the accuracy of entity and relationship extraction is improved.

The invention relates to the technical field of antenna signal detection, and provides a low-complexity signal detection method and system, and the low-complexity signal detection method employs a layered updating mode, and is more beneficial to the design and implementation of a high-throughput hardware architecture. And meanwhile, through algorithm transformation or approximation, the algorithm and implementation complexity is effectively reduced. In the interference elimination process, the constellation point characteristics are utilized, operation is carried out on an operand with a smaller bit width, a critical path during hardware implementation is effectively shortened, and implementation complexity is reduced; in the constellation matching process, a new reliability measurement method is provided by utilizing the characteristics of a quadratic function, so that the number of required multiplication and addition operations is reduced by half respectively; a signal detection method is reconstructed into initialization and iteration steps so as to reduce the processing period, and multiplication, addition and sorting operations required in the initialization process are completely eliminated through an approximation method.

The invention discloses a method for optimizing the cognitive user energy efficiency in a cooperative cognitive network. The method comprises the following steps that 1, an energy efficiency optimization question model of a cooperative cognitive network model under the constraint condition is built; 2, the energy efficiency optimization question model is converted into minus-form question to be solved through a Dinkelbach algorithm; and 3, the minus-form question is converted into a convex optimization question to be solved by introducing auxiliary variables. According to the method, the optimal transmission power and time allocation are calculated mainly by combining and using fractional programming and introducing the auxiliary variables and a Lagrangian duality theory, and therefore the cognitive user energy efficiency can be highest. Accordingly, not only are problems existing in the prior art solved, but also the method is simple, and the result is accurate.

The invention relates to the technical field of digital image acquisition, and discloses an image sensing behavior simulation model generation system, including a public model storage module for storing common parts of each model, and a special model design for designing heterogeneous parts of finger-to-shape models module, a program language conversion module that converts the programming algorithm in the design module into the program language required by the model, and a model combination generation module that combines the common model and the heterogeneity model into a new required model. The image sensing behavior simulation model generation system and method have the advantages of being convenient for rapid modeling and improving work efficiency.

The invention is adaptable to the technical field of automation of electronic design, and provides a method for algorithm transformation. The method includes converting a C-language program or flowchart of the C-language program into a state graph, simplifying the state graph to obtain the simplified state graph including control logic and arithmetic logic, separating the arithmetic logic from the simplified state graph, implementing the arithmetic logic through hardware and determining the implemented result as a data path, separating the control logic from the simplified state graph, implementing the control logic through hardware and determining the implemented result as a controller, and implementing the data path and the controller by hardware description language HDL. Since the C-language is converted into the state graph to construct the data path and the controller which are implemented by the HDL, poor universality in conversion from the C-language to the HDL hardware implementation is overcome.

The invention provides a method for measuring the rotation speed based on fusion of several kinds of characteristic data. The method comprises: collecting more than one kind of motor rotation speed characteristic data; processing the rotation speed characteristic data by using an autocorrelation function to obtain a periodic signal of a data waveform; carrying out algorithm transformation to obtain a frequency corresponding to the data waveform; and inputting the corresponding frequency into a data fusion prediction model for fusion to obtain a rotation speed of the motor. In addition, the invention also provides an apparatus for measuring the rotation speed based on fusion of several kinds of characteristic data. The apparatus includes a sensor module, a display module and a main controlmodule. According to the invention, the rotation speed is calculated based on fusion of several kinds of characteristic data; the stability and accuracy of the rotation speed measurement are improved;and a problem of the mounting trouble of the existing device is solved.

The application discloses a neural network-based data processing method, device, platform, medium and equipment, and belongs to the field of data processing. The method comprises the following steps: data definition is performed on to-be-processed data, and an object instance is created; a parameter interface pre-encapsulated by the object instance is called to set parameters of the object instance, and object parameters are obtained; the object parameters and the object instance are simultaneously input into a neural network by calling a function processing interface, the to-be-processed data is preprocessed in the neural network by using an algorithm reasoning library, preprocessed data is obtained, and the preprocessed data is output by calling a data output interface; the neural network performs optimization processing on the preprocessed data by using an algorithm configuration file and the algorithm reasoning library, and processing data is obtained; and the processing data is post-processed according to user pre-defined parameters, and standard processing data is obtained. The application does not need algorithm conversion, is simple and efficient, and saves the actual deployment time of a vehicle.

The present application relates to the technical field of antenna signal detection, and provides a low-complexity signal detection method and system. The low-complexity signal detection method adopts a layered update method, which is more conducive to the design of a high-throughput hardware architecture. with realization. At the same time, through algorithm transformation or approximation, the algorithm and implementation complexity are effectively reduced. In the process of interference elimination, using the characteristics of constellation points, it is proposed to perform operations on operands with smaller bit widths, which can effectively shorten the critical path of hardware implementation and reduce the implementation complexity; in the process of constellation matching, the characteristics of quadratic functions are used. , a new reliability metric method is proposed, which reduces the number of required multiplication and addition operations by half; the signal detection method is refactored into initialization and iterative steps to reduce the processing cycle, and the initialization process is approximated by an approximation method. The multiplication, addition and sorting operations required in the

The invention discloses a multi-dimensional data sorting method based on dynamic weight fusion, and relates to the technical field of computer science and data processing. Comprising the following steps of: converting a data source set and weight configuration into a dimension space through a data preprocessing method; converting the dimension space into a dimension priority hierarchical vector set through a dimension priority hierarchical recursive algorithm; and step 3, converting the dimension space and the dimension priority hierarchical vector set into a priority queue after multi-dimensional weight fusion sorting through a multi-dimensional weight fusion algorithm. According to the method, the sorting method is packaged, the weight configuration is opened for the user to use, the multi-dimensional data sorting efficiency is remarkably improved, the calculation complexity is remarkably reduced, multi-modal data processing, multi-dimensional weight fusion and real-time rule updating are supported, and the system expansibility and the service adaptability are improved.