9 results about "Real time algorithm" patented technology

The invention belongs to the technical field of image processing, and particularly relates to a stereoscopic vision parallax prediction method based on an energy function and an attention fusion network. The objective of the invention is to solve the problem that a real-time algorithm is difficult to obtain high precision in stereoscopic vision parallax prediction. The method comprises the following specific steps: acquiring a data set: training a network model by adopting two public virtual and real stereo matching data sets; the method comprises the following steps of: constructing a network model: constructing an Entry Function and Attention Interference network, and optimizing cost body construction and cost aggregation by utilizing an attention mechanism; designing a minimum loss function; training a network model: inputting the acquired data set into the network model for training; finely adjusting model parameters; and storing model parameters. According to the method, an end-to-end supervised method is adopted, a UNet-like network architecture is integrally used, cost body construction and cost body aggregation are improved, and the efficiency of the model is improved by using an attention mechanism, so that the real-time stereoscopic vision parallax prediction method based on deep learning is realized.

This invention discloses an automatic identification method for the polarity of high-frequency pulse current waveforms based on deep learning, belonging to the field of power equipment insulation fault detection technology. This method deeply utilizes the characteristics of the pulse signal waveform to identify the first wave and its polarity. Firstly, under laboratory conditions, the propagation characteristics of the first wave at different typical discharge locations and types are obtained by injecting steep pulses. The response signal waveforms at each outgoing line coupling end are measured. Then, using the waveform sequence as input vectors, a deep learning network is constructed. Considering convolutional neural networks, a sample library of typical response waveforms for each injection method and location is established using a digital image matrix as input. The sample library is continuously expanded through adversarial learning. An artificial neural network is used to train the first wave waveform and polarity on the input waveform sequence, and then uses the artificial neural network to identify waveform details. This achieves fully automatic first wave polarity identification, suitable for real-time algorithm applications in online monitoring.

The invention discloses a display support dynamic trajectory compensation adjusting system and method based on real-time feedback, in the technical scheme of the invention, the system comprises a sensing module adopting a six-axis MEMS inertial measurement unit and a joint encoder, a calculation module for data processing and an execution module, the sensing module collects six-dimensional disturbance data and joint state data in real time through a six-axis MEMS inertial measurement unit and a joint encoder; the calculation module adopts a composite control calculation strategy of feedforward adjustment calculation based on a coefficient matrix and feedback adjustment calculation based on set parameters to calculate an optimal compensation adjustment control instruction through a real-time algorithm; according to the invention, a complete sensing, calculating, decision-making and execution closed-loop control system is established, the display can be effectively prevented from being disturbed and shaking in an unsteady state environment, and the system has the advantages of fast response, high precision and strong adaptability.

This invention proposes a high-performance collaborative processing method for ultra-high precision pointing measuring instruments. It consists of a master-slave processor. The master processor ensures periodic update rates through external communication, while the slave processor handles complex algorithms and transmits results to the master to improve its real-time algorithm performance. The master software is designed with a multi-task architecture to ensure real-time performance. A "master-slave interaction task" is periodically initiated, triggering a data reception interrupt on the slave and completing master-slave data interaction. Other tasks handle real-time algorithms and communication. The master-slave interaction task has a timeout exit mechanism, and its execution is determined by the timeout status of other tasks at the timed start. To ensure data transmission reliability, the master-slave communication bus uses unidirectional transmission and reception to avoid transmission conflicts caused by duplex mode. Data transmission uses checksums, and data to be sent in multiple periods is packetized and assigned sequence numbers to avoid duplicate reception or data loss. This invention can be widely applied to embedded platforms using a master-slave dual-processor architecture.

The invention discloses a modular design method for a bidirectional driving device, which comprises the following steps: defining a transmission system as three functional modules, respectively configuring a one-way transmission unit for a first power input module and a second power input module, and configuring a two-way transmission unit for the first power input module and the second power input module on the basis of the physical characteristics of the one-way transmission units; defining a transmission logic for realizing automatic selection and isolation of a power path; defining a power switching logic based on working condition prediction: designing a control unit which is configured to receive a vehicle state signal, and pre-judging and triggering preparation or switching of a power source based on a pre-stored working condition map and a real-time algorithm; defining a mechanical interface and an electrical interface among the first power input module, the second power input module and the power output module; the three function modules and the control unit are integrated through defined interfaces, and the transmission system with the predictive power switching function is constructed. According to the method, the overall operation efficiency in a multi-power-source driving scene is optimized.

The invention relates to a humidity calibration device, calibration system and calibration method, and belongs to the technical field of humidity sensor calibration detection. The temperature of liquid and the temperature of gas in a closed cavity are monitored synchronously, and real-time algorithm compensation is carried out by using an original humidity correction table, so that measurement errors caused by gas-liquid temperature difference are effectively overcome; the system integrates a drying agent, various saturated salt solutions and water, realizes 0-100% full-range humidity coverage, drives fan blades to accelerate balance through an external motor, simulates a real ventilation environment, and finally improves the humidity calibration precision under the condition that external temperature control is not needed.

An advanced multi-agent pathfinding and optimization system for logistic networks, revolutionizing fourth-party logistics (4PL) management through Multi-Agent Reinforcement Learning (MARL). It aims to optimize package transportation across multi-hub networks, shifting from a volume-driven approach to a rate-driven model. This system adapts dynamically to business fluctuations, redefining risk dynamics in logistics. It prioritizes shipments based on rate tiers, altering risk distribution and incentivizing efficiency over volume. The system addresses current logistical challenges by integrating multi-modal perspectives and anticipating real-time disruptions, overcoming scalability, adaptability, and dynamic change limitations. Important features include an advanced “GeoEngine” for accurate geocoding, a first / last mile engine with VRP solutions, real-time A* algorithm implementation for hub assignment, and a comprehensive RL model utilizing Q-learning with lightGBM. This system provides a more balanced, market-driven approach to logistics, offering a paradigm shift in efficiency, adaptability, and strategic rate management.

This invention relates to an adaptive time-frequency joint filtering method for electrostatic data from solid rocket motor exhaust. The method includes the following steps: constructing a main channel filtering algorithm to suppress the fundamental combustion oscillation, employing an adaptive IIR notch filter, with the core frequency detected by a real-time FFT algorithm and the bandwidth dynamically adjusted according to the combustion state; constructing an auxiliary channel to suppress harmonics, consisting of multiple parallel notch filters, each corresponding to a harmonic frequency, with the harmonic order adaptively determined based on the fundamental attenuation; addressing the phase distortion problem of traditional IIR filters by using forward-backward filtering to offset phase shift; particle pulse detection is the core of solid rocket motor exhaust electrostatic signal processing, aiming to accurately identify transient pulses (1-5 μs) generated by Al2O3 particle collisions; this invention uses a wavelet domain adaptive algorithm based on multi-scale noise separation, a time-varying threshold mechanism, and the SURE optimization criterion to suppress combustion oscillations and noise in the exhaust electrostatic signal while preserving particle pulse information.

This invention discloses a gas leak monitoring system for environmentally friendly ring main units (RNBs), belonging to the field of environmentally friendly RNB monitoring technology. The invention includes a real-time data acquisition and monitoring platform, which is communicatively connected to a joint real-time data acquisition and storage module, an external environmental gas dynamic analysis module, an internal environmental gas risk analysis module, and an emergency control decision processing module. The invention achieves multi-source data fusion monitoring of the internal and external environments of the RNB through the joint real-time data acquisition and storage module, solving the problem of the one-sidedness of single-parameter monitoring and improving the accuracy of leak risk identification. It utilizes real-time algorithms to achieve second-level data processing and graded risk warning, shortening fault response time. Based on a comprehensive risk index, a graded emergency response mechanism is established to accurately handle different risks, ensuring equipment safety while reducing operation and maintenance costs.