33results about How to "Realize online optimization" patented technology

The invention discloses a small-thrust transfer trajectory optimization method for asteroid detection based on successive convex programming, which belongs to the field of aerospace technology. The realization method of the invention is as follows: establishing an improved vernal equinox dynamics model for the small thrust transfer of the asteroid probe. According to the dynamic characteristics of low-thrust transfer, the constraints and optimization performance indexes of the low-thrust trajectory optimization problem are given. The specific form of the low-thrust interstellar transfer trajectory optimization problem is given. The nonlinear small thrust transfer problem is convexized by dynamic linearization and nonlinear equality constraint relaxation. The continuous optimal control problem after convexization is transformed into a convex optimization problem by numerical integration. Taking the sub-problem after convexization as the inner link of each iteration step, the optimal small-thrust transfer trajectory for asteroid detection can be obtained quickly by using the successive approximation strategy in a limited number of steps, that is, the optimality and accuracy of the low-thrust trajectory can be guaranteed. Under this condition, the online optimization of the low-thrust trajectory is realized. The invention has strong robustness, high repeatability and high flexibility.

The invention provides a distributed coordinated control optimization method for an urban rail transit ground supercapacitor energy storage system, and belongs to the technical field of urban rail transit supercapacitor energy storage system control. This method obtains the SOC state of the supercapacitor, the state of the substation and the running state of the train, and constructs a state set; takes the supercapacitor of each substation as an energy storage agent, and analyzes the energy flow theory and multi-objective coordination optimization of the traction power supply system The function determines the income function of each energy storage agent; constructs a multi-agent dynamic game model, based on this model, combines the distributed Q-learning algorithm to solve and optimize the distributed coordinated control of the ground supercapacitor energy storage system. The invention performs distributed and independent control on each energy storage system, effectively reduces the computational complexity, and improves the reliability of the control strategy; combines the dynamic cooperative game theory and the reinforcement learning theory, and the efficiency of the game equilibrium point is effectively improved. Therefore, it can Realize the online optimization of the overall energy-saving effect of the full-line energy storage system.

The invention provides a distributed subgradient algorithm-based optimal control method for a reactive power of a power system. Reactive power distribution of the power system is optimized and determined by a distributed subgradient algorithm; and distribution control on the reactive power is executed according to an optimal solution. By the distributed subgradient algorithm-based optimal control method for the reactive power of the power system, the probability of a single-point failure under centralized control and distributed control can be reduced; the loss of the power system is reduced when the running state of the power system is improved; and the optimal control method can be flexibly applied to the power systems with different scales and topological structures and is not disturbed by certain specific errors.

The invention discloses a method for controlling the elongation of a hot-dip galvanized four-roller skin pass mill, which belongs to the production control method in the field of cold-rolled strip steel. The present invention first establishes PID model parameter data storage variables, and calls PID parameter data and randomly fluctuates in the range of 5% when the head of a new coil of strip steel is close to the tempering machine. When the head of the strip passes through the tempering machine and the deviation between the elongation rate and the target value is greater than 0.2, execute the deviation statistics program, record the maximum value of the elongation rate deviation and the PID adjustment amount index within 30 seconds, and then determine the maximum elongation rate deviation Whether the value is greater than the original storage data and whether the adjustment index is smaller than the original storage data, if both conditions are met at the same time, the process parameters after random fluctuations are overwritten and stored in the data storage variable. Realize the automatic update process of PID parameters and achieve better elongation control effect.

The invention discloses an adaptive cruise control method based on approximate policy iteration. The adaptive cruise control method comprises the steps of (1), collecting samples; (2), learning on the samples by using an approximate policy iteration algorithm to obtain an approximately optimal policy; (3), optimizing PI controller parameters online in the cruise control, namely, optimizing the PI controller parameters online in a data driving mode by using the approximately optimal policy, so that the cruise control can achieve expected performances. The adaptive cruise control method has the advantages that the principle is simple, online optimization can be achieved, control performances can be improved, and the like.

The invention discloses an expert control system based on a batch type coating machine. The system comprises a coating machine, an image detection system and an expert system. The expert system consists of a control cabinet and a display screen; a human-computer interaction interface, a knowledge base module, an inference engine module, a feedback self-learning module and a control center are arranged in the control cabinet; the control center is an industrial personal computer electrically and bidirectionally connected with the human-computer interaction interface, the feedback self-learningmodule and a PLC control module of the coating machine respectively; the human-computer interaction interface is in electrical output connection with the knowledge base module; the feedback self-learning module is in electrical output connection with the inference engine module; and a signal input terminal of the feedback self-learning module is further in signal connection with a signal output terminal of an image detection system and is used for receiving the coating qualification rate fed back by the image detection system. According to the invention, the processing parameters are automatically optimized and combined for different seeds; the efficiency is improved, the dosage of a coating agent is saved, and the effect is remarkable on the premise of meeting the coating quality and qualification rate.

A processing deformation control method based on meta-reinforcement learning, which is characterized in that the processing deformation control process optimization of each part in different groups of source data is regarded as a task, a reinforcement learning model is established for each task, and the workpiece processing is divided into For several processing steps, the processing state of the workpiece is taken as the state, the process selection of the next processing step is used as the action, and the next processing state and the subsequent processing state are used as the basis for designing the reward function; based on the meta-learning method, each reinforcement learning model is used as The base model, through the collaborative training of the base model and the meta-model through the source data; when faced with a new processing task, fine-tuning the meta-model through a small amount of sample data of the new task to obtain a reinforcement learning model adapted to the optimization of the machining deformation control process for the new task. The invention improves the effect of deformation control, can realize online optimization of processing technology, and reduces the demand for actual processing technology data.