30 results about "Data driven algorithms" patented technology

The invention relates to a multi-shaft motion control method based on data driving and parameter mixing optimization. According to the method, a data driving algorithm and a parameter mixing optimization algorithm are established and debugged in a PC host computer, and motion control codes which can be identified by a multi-shaft motion controller end of a lower computer are generated by a compiling module and an operation module of the host computer; an operation state of a motor side is detected by a photoelectric encoder, a detection result is taken as a feedback signal which is transmitted to a multi-shaft motion controller, the feedback signal is compared with an expected position input signal to acquire a position error signal, the position error signal is taken as input of an MFAC control algorithm, and an outer-most layer position control ring of a motor driving system is constructed. Through a multi-shaft motion control method based on a data driving theory, no specific controlled system mathematic model is required, the input and output data information is only required to design a controller of a controlled system, and influence of unmodeled dynamics on a multi-motor driving system and dependence on the system are solved.

The invention provides a health monitoring method for effective loads of a space station based on a data-driven algorithm. In the design stage, after historical data of the effective loads are subjected to state vector construction, parameter standardization and weight processing, training samples are obtained; then, clustering learning is performed on the training samples, and different working condition data classifications can be obtained. In the running stage, after real-time downlink test data of the effective loads are processed, through the working conditions obtained through clustering learning, the downlink data are monitored in real time, if abnormal data occur, it shows that new working conditions happen to the loads, a fault may happen or is about to happen probably, finally, the abnormal data are detected in combination with a fault diagnosis tree method, and the position of the fault is determined. Through machine learning of the historical data, a system health knowledge base is formed, the abnormal state of the loads is found through calculation of the distance value of outliers, real-time monitoring on the health state of the loads is achieved, fault detection and positioning of the loads can be supported, and prediction to a certain extent is achieved.

The present solution provides a new tool to facilitate the tracking of experiments and development of data-driven algorithms. The tool may obtain algorithms, parameters and data sets to execute one or more experiments to produce an outcome. The tool may identify differences between two or more experiments, such as differences in parameters, algorithms, or data sets to facilitate identifying the optimal combination of algorithms or parameters.

The invention provides a data driving algorithm for a spontaneous combustion coal-mine fire big data platform, and relates to a probability estimation method for the spontaneous combustion coal-mine fire mark gas occurrence rate. The method comprises the following steps that 1, coal-mine mark gas concentration data is collected; 2, data is pre-processed; 3, the probability distribution type to which the mark gas growth rate belongs is judged through hypothesis test; 4, parameters of probability distribution are calculated through all mark gas occurrence rate data by means of a Maximum Likelihood Estimation method; 5, by means of the obtained parameters of the probability distribution, real probability distribution of the mark gas change rate is simulated through a large amount of data points by means of a Monte Carlo Simulation method; 6, by means of the real mark gas change rate distribution obtained in the last step, mark gas concentration distribution in future prediction time is obtained by multiplying time wanting to predict; 7 a mark gas concentration alarm limit is set, and the probability of occurrence is obtained.

The invention discloses a method for estimating the SOC of a lithium battery by a data driving algorithm considering internal resistance. The method comprises the following steps: firstly, carrying out a charge-discharge test on the lithium ion battery by using test equipment, measuring voltage, current, temperature and internal resistance data of the battery in different working states, and preprocessing the obtained data; then, building a bidirectional GRU network, wherein one part of processed data serves as a training set to train the network, and the other part of the processed data serves as a test set to evaluate network performance; and finally, in order to improve the performance of the constructed network, optimizing the bidirectional GRU network by using an NAG algorithm. The input of the constructed bidirectional GRU-NAG network is the voltage, the current, the temperature and the internal resistance of the battery, and the output of the constructed bidirectional GRU-NAG network is the residual electric quantity of the battery, so that the method has the advantages of high estimation speed and simple process, and is a data-driven battery residual electric quantity estimation model.

The invention relates to the technical field of heating and air conditioning equipment, in particular to an air source heat pump variable return difference water temperature control method and system based on supply and demand matching. A building model capable of reflecting room temperature response and fluctuation is established according to building thermal design and air tightness conditions; according to air source heat pump operation measured data, state recognition parameters are introduced to distinguish a starting stage and a stable operation stage, and a unit model capable of reflecting the start and stop dynamic processes and losses is established through a data driving algorithm; an end model considering the dynamic thermal process and a transmission and distribution model considering waterway inertia are established; an air source heat pump system dynamic model is formed according to the dynamic coupling relation among the unit heating capacity, the end heat supply capacity and the building heat dissipation capacity; and a variable return difference water temperature control strategy is established according to the indoor thermal comfort demand of the target system, and a unit is controlled to operate according to the variable return difference water temperature control strategy during operation. The problem of frequent start and stop of the air source heat pump unit in the low-load state is solved, the system performance is improved, the system power consumption is reduced, the noise influence in the start and stop process is reduced, and the method and system contribute to prolonging the service life of a compressor.

The invention belongs to the technical field of convexity prediction, and particularly relates to a plate convexity prediction method based on kernel partial least squares (KPLS) and a support vectormachine combined. The plate convexity prediction method comprises the following steps: S1, field data are collected through a high-precision monitoring device; S2, the collected data are preprocessed;S3, a KPLS regression prediction model is established; and S4, a KPLS-SVM plate convexity prediction model is established. By taking a data-driven algorithm as a mathematical tool, abnormal values can be removed from a large quantity of collected field rolling process data, a convexity prediction model for a strip steel continuous-rolling plate based on a KPLS method and the support vector machine combined is established, the convexity of the strip steel continuous-rolling plate is predicted, the established model is optimized through a particle swarm optimization algorithm, and the prediction precision of the convexity of the strip steel continuous-rolling plate is further improved. The plate convexity prediction method is used for predicting the convexity of the strip steel continuous-rolling plate.

The invention provides a power grid weak link identification method based on Bayesian reasoning, which comprises the following steps: fusing historical load conditions of elements in a power grid withactual operation data, constructing a distribution probability table considering power transmission distribution factors by utilizing data-driven thinking, and establishing a node-branch Bayesian network through tide topology and a grid structure; and initiating a bifurcation attack to the network according to different fault modes, updating the distribution probability table, and calculating theweakness of all branches by using a maximum possible interpretation algorithm. The technical method provided by the invention mainly depends on a network structure and an operation state, and is lessinfluenced by interference of a fault type; the proposed data driving algorithm can meet the calculation requirement of weak identification of a large power grid, and modeling errors are reduced; multiple types of fault events in an actual power grid are covered, and the requirements for power grid operation reliability and planning directivity can be met.

The invention discloses a data-driven electric quantity sensor error online evaluation closed-loop improvement method and system and a medium. The method comprises the steps of creating an electric quantity sensor data correction model set of the nth iteration; acquiring electric quantity sensing data to generate a sensing data set; generating an electric quantity sensing data correction coefficient set after the nth iteration according to the electric quantity sensing data correction model set, and correcting the electric quantity sensing data set to generate a correction data set; carrying out electric quantity and electric quantity sensor error deduction on the correction data set according to an open-loop data driving algorithm to form a deduction result; if the end condition is not met, updating the number n of iterations and the electric quantity sensing data correction model, accumulating an electric quantity sensing data correction coefficient set, and continuing iteration; otherwise, outputting an iterative result. The invention does not need field tests, does not need to change the topology of a physical system, does not affect physical operation, can effectively improvethe result deduction accuracy, improves the generalization of a deduction model, achieves high efficiency and safety, and does not affect the physical system.

The invention discloses a time sequence missing value filling method based on variable window pattern recognition. The method comprises the following steps: selecting an active power correlation variable of a wind turbine generator based on an operation mechanism; selecting a wind turbine generator active power correlation variable based on data feature selection; carrying out variable window mode matching on the multi-dimensional correlation variable of the active power of the wind turbine generator; performing wind turbine generator active power continuous missing data block multi-filling based on a similar mode; and evaluating and confirming multiple filling results. Aiming at wide existence of the industrial Internet of Things and common data continuous missing conditions, high-proportion missing data can be efficiently and accurately filled, the effective data volume is greatly increased, and an important data foundation is laid for implementation and application of data driving algorithms such as machine learning and artificial intelligence.

The invention provides an expressway green channel vehicle transportation planning system based on big data driving. Comprising a data acquisition layer, a data analysis layer, a data storage layer and a data display layer, according to the method, a green channel vehicle driver is taken as a research object, typical disqualification reasons and types in highway green channel vehicle checking business data are mined, and a data-driven algorithm and text analysis are taken as technical means, so that the green channel vehicle driver timely knows green channel vehicle policy dynamics and easily confused directories of disqualified products; and plans of schemes such as cargo transportation, loading requirements, transportation routes and the like are provided for drivers.

The invention provides a method and a device for adaptively calibrating transmitting power in a radio frequency front-end module based on a data driving algorithm for a wireless communication system.By converting a transmitting power calibration problem into a dynamic linearized model, the design of a control system is simplified. Compared with other power calibration methods, the calibration algorithm provided by the invention eliminates a modeling step, does not need to establish an accurate system model, and only needs input/output (I/O) measurement data of the system. Related experimentalresults show that by using the proposed calibration algorithm and device based on data driving, the transmitting power error of the device can be reduced from 1.5 dB to 0.25 dB or less.

The invention discloses a data-driven engineering material ultra-high cycle fatigue life prediction method. The method comprises the following steps: firstly, collecting engineering material information and ultra-high cycle fatigue data to form initial sample data, and dividing the data into a test set and a training set; secondly, evaluating the contribution degree of each input characteristic variable to an output variable according to an existing physical model, sorting the importance of the characteristic variables, then, screening out key characteristic variables, and forming a target function of a data-driven model; embedding the target function into a machine learning algorithm to obtain an intermediate calculated value Z, and evaluating the prediction precision of the data-driven model by adopting a decision coefficient R2; and finally, associating the training set Z value with the fatigue life to realize the ultra-high cycle fatigue life prediction of the engineering material. According to the method, main influence factors of the ultra-high cycle fatigue life of the engineering material are combined with the data driving algorithm, the ultra-high cycle fatigue life of the engineering material can be quickly and effectively predicted, and a good implementation effect is achieved in defect-containing materials such as weld joints.