637 results about "Control variable" patented technology

The invention discloses an air valve on-line self-adaptive adjusting method and system fusing digital twinning, and belongs to the field of air valve intelligent adjusting.The adjusting method comprises the following specific steps that firstly, edge devices collect and preprocess multi-source data of a corresponding air valve system, and based on the preprocessed multi-source data, the multi-source data of the corresponding air valve system are obtained; constructing a virtual simulation environment corresponding to the current air valve system; iI, in a virtual simulation environment, pre-training an air valve control model, identifying a causal relationship between each sensing parameter and an air valve adjustment result, and dynamically optimizing each control variable; according to the method, model prediction precision and simulation credibility are remarkably improved, dependence on a large amount of new data is reduced, cross-system rapid adaptive adjustment is realized, misjudgment and interference of confounding factors can be avoided during strategy optimization, interpretability and credibility of model decision are enhanced, data privacy is protected, convergence of a global strategy is accelerated, and the method is suitable for large-scale popularization and application. The overall intelligent level of the system is improved.

The invention relates to the technical field of life cycle management, in particular to a catalyst industry chain-oriented full life cycle traceability method and system, and the method comprises the steps: obtaining the state attribute data of the same batch of recovered products, and carrying out the analysis to obtain an efficiency feature vector; inputting the efficiency feature vector into a preset quality evaluation model for matching, and identifying quality deviation data of the recycled product; the batch identity codes are analyzed, and resume data of the production full life cycle are obtained; a multi-dimensional data association model is constructed, quality deviation data is taken as a target variable, resume data is taken as an input feature, an influence weight coefficient and an attribution confidence index of each production control variable on the quality deviation are analyzed, and key control elements are screened out; and performing production regulation and control optimization operation based on the key control elements, calculating target parameter offset for correcting quality deviation, dynamically calculating a dynamic tolerance constraint range of a production control variable in combination with an attribution confidence index, and packaging into an execution management and control protocol.

The invention discloses a carbon dioxide mineralization storage dynamic intelligent regulation and control and permeation enhancement optimization method and system. The optimization method comprises the following steps: collecting field monitoring injection parameters and related data of reaction products in a mineralization storage process in real time; according to injection parameters monitored on site and related data of reaction products, two optimization objective functions of mineralization rate and free CO2 volume are formed; constructing a mineralization sequestration multi-objective optimization model, and screening out an optimal injection parameter set value from the Pareto solution set to obtain an optimal condition parameter; optimal injection parameters in the Pareto optimal solution set are input into the constructed field enhancement regulation and control module, and control variables are adjusted in real time according to real-time changes of reservoir response, mineralization reaction process and injection working conditions; and fracturing transformation is conducted on the target storage rock mass, the seepage enhancement effect of the target storage rock mass is quantitatively evaluated, an injection scheme is dynamically updated based on the transformed reservoir parameters, and the mineralization regulation and control system is enhanced.

A valve timing control device includes: a unit for calculating a target valve timing for one of an intake valve and an exhaust valve in accordance with an operational state; a unit for detecting an actual valve timing of one of the intake valve and the exhaust valve; an actual valve timing control unit for generating a controlled variable for the VVT based on a difference between the target valve timing and the actual valve timing; a unit for learning a learning value of the controlled variable required for maintenance of the actual valve timing based on an integral correction value; and a unit for detecting a malfunction in a VVT from an amount of a shift in the actual valve timing when learning is not completed yet, and for detecting a malfunction in the VVT from the difference when learning is already completed.

The invention discloses a mine semi-autogenous grinding process analog simulation method and system based on discrete elements, and relates to the field of analog simulation, and the method comprises the steps: according to a crushing characteristic label and a parameter mapping matrix, fusing Internet of Things sensor data, constructing a particle size distribution prediction model and an energy consumption optimization scheduling strategy, and obtaining an energy-saving operation plan; comparing the simulation data set with field production data, adjusting multi-phase coupling particle model parameters, generating a calibration parameter set, researching the influence law of semi-autogenous mill operation parameters by adopting a control variable method based on the calibration parameter set, obtaining a performance index database and an analysis report, and fusing multi-sensor data and laser radar data to obtain a semi-autogenous mill operation parameter analysis result. An ore grinding process sensing mechanism and optimization control logic are constructed, and a visual interface is generated; through multiphase modeling and data driving optimization, the simulation precision and the process suitability are comprehensively improved, and an efficient and economical solution is provided for the mine semi-autogenous grinding process.

The invention discloses a photovoltaic photo-thermal coupling hydrogen production system control method for a hydrogen-burning heavy truck traffic network. The method comprises the following steps: S01, collecting operation data of a photovoltaic array, an electrolytic cell and a hydrogen storage tank; s02, based on a model prediction control algorithm, establishing a multi-objective optimization model containing an objective function and constraint conditions; s03, merging the sub-objective functions into a comprehensive objective function through a linear weighting method; s04, applying constraint conditions; S05, performing rolling optimization calculation by adopting a model prediction control algorithm; s06, outputting an optimal control variable; s07, adjusting corresponding parameters in the system; s08, monitoring the running state of the system; s09, whether the system meets constraint conditions is judged, according to the control method, through multi-target collaborative optimization and precise constraint control, the performance, efficiency, economical efficiency and safety of the photovoltaic photo-thermal coupling hydrogen production system can be remarkably improved, and clean, efficient and stable hydrogen energy supply is provided for a hydrogen combustion heavy truck traffic network.

The invention relates to the technical field of machine vision and food quality detection, in particular to a food defect real-time detection method and system based on image processing. The method specifically comprises the following steps: integrating multi-modal data, and adaptively adjusting a weight coefficient to improve data quality; multi-stage noise reduction is carried out, key frames are extracted, and a key frame verification mechanism is enhanced through a 2FA control variable; constructing a YOLO-BioNet model, and optimizing the detection efficiency and precision in a complex scene through dynamic convolution kernel allocation and a spiking neural network; integrating Kalman filtering, LSTM (Long Short Term Memory) time sequence modeling and a deep reinforcement learning model, analyzing a food surface change track and predicting a defect state; real-time data and historical information are fused, and a dynamic threshold value is calculated through a Bayesian network; performing linkage response between an automatic sorting instruction and a production line; a distributed storage and graph neural network optimization model is adopted, and defect tracing is supported. The method has remarkable advantages in the aspects of real-time performance, detection accuracy and self-adaptive capability.

The invention discloses a power distribution network adaptive topological structure optimization method, system and device and a medium, and the method comprises the steps: obtaining related parameters of a power distribution network system, dividing the power distribution network system into different layers and different regions based on the related parameters of the power distribution network system, and constructing a layering and partitioning model; coordinating and solving the layering and partitioning model through an intelligent algorithm, and obtaining control variables of different regions; constructing an objective function based on the control variable, and solving the objective function by adopting an improved particle swarm optimization method to obtain a topological structure; and optimizing the topological structure by adopting a double-layer optimization strategy to obtain an optimization result, evaluating the optimization result, and feeding back the evaluation result to the target function. According to the method, the randomness and volatility challenges caused by multi-source resource access can be effectively dealt with, and the limitation of the traditional separation design is broken through the collaborative optimization of topology and control, so that the operation efficiency, the stability and the anti-interference capability of the power distribution network are improved, and the collaborative evolution of a topological structure and a control strategy is realized.

The invention discloses a standard system applicability evaluation method and system based on multi-source data, and relates to the field of standardization, and the method comprises the steps: collecting the historical operation data of equipment, and dividing the equipment into an experiment group equipment group and a control group equipment group; the method comprises the following steps: preprocessing historical operation data of equipment, and respectively constructing a performance index value sequence for quantitatively reflecting the achievement condition of a target standard for a technical target for an experimental group equipment group and a control group equipment group; and on the basis of the performance index numerical value sequences of the experimental group equipment group and the control group equipment group, a net effect value and statistical confidence of a target standard are calculated by adopting a double difference model, and a unit model and a geographical climate region are added into the double difference model as fixed effect control variables. According to the method, the corrected net effect value and the intermediary centrality score are fused into the comprehensive index, so that multi-dimensional dynamic evaluation combining data driving, causal identification and knowledge value of standard applicability is realized.

The invention relates to a trajectory tracking control method for a four-wheel unmanned vehicle. The method comprises the following steps: S1, acquiring the position and course angle of the vehicle; acquiring a vehicle reference position and a reference course angle corresponding to the vehicle reference track point; calculating a transverse error and a course angle error of the vehicle; establishing a vehicle trajectory tracking error model based on the vehicle lateral error and the course angle error; s2, obtaining a tire cornering stiffness estimation value at the current moment; s3, solving an optimization problem to obtain an optimization control quantity; s4, outputting a feed-forward compensation amount; s5, adding the optimal control quantity and the feed-forward compensation quantity to obtain a final control quantity; and S6, executing a control instruction based on the final control quantity, and performing trajectory tracking control on the four-wheel unmanned vehicle. Compared with the prior art, the method has the advantages that the number of control variables of MPC is reduced, the calculation complexity is remarkably reduced, and the trajectory tracking precision and the control real-time performance of the four-wheel unmanned vehicle are improved.

The invention discloses a reinforcement learning-based slurry balance shield construction attitude settlement collaborative optimization method and system, electronic equipment and a storage medium. The method comprises the following steps of: constructing state and action characteristics; a control strategy modeling step: constructing a composite reward function which takes incision pressure, partition propulsion pressure, tunneling speed and the like as control variables and fuses ground surface settlement and attitude control performance; designing a control strategy model based on reinforcement learning; a control strategy training step; a control parameter output step; in the shield tunneling process, control parameter setting is dynamically generated through current state information, and cooperative control over settlement and posture is achieved. According to the slurry balance shield construction posture settlement collaborative optimization method and system based on reinforcement learning, the electronic equipment and the storage medium, the adaptive capacity and self-learning capacity of the system can be improved, and the automation, intelligence and construction safety level of the slurry balance shield construction process can be improved.

The invention discloses an electric linear servo system control method based on cascade LESO-DISMC. The method comprises the following steps: firstly, establishing a complete mechanism model of a permanent magnet synchronous motor servo loading system; then designing DOB interference observer real-time compensation by adopting a LuGre friction model and a gap dead zone model; a feedforward control quantity is constructed by utilizing a measurable displacement signal, and redundant torque is inhibited; designing a dynamic integral sliding mode controller to ensure global stability; a high-low bandwidth LESO is constructed, and both fast tracking and smooth filtering are considered; and finally, trajectory optimization is carried out, a smooth trajectory is generated by adopting polynomial interpolation for the step and triangular wave signals, and mechanical vibration is inhibited. According to the method, a LuGre friction model is adopted to compensate nonlinear friction, cascade linear LESO and sliding mode control are combined, a track is optimized through an S-shaped curve, and the problem of'differential explosion 'is solved; according to the method, instruction smoothing, core control and friction compensation are fused, and the anti-interference capability and the loading precision of the system are improved.

The invention discloses a fan frequency control method and device, electronic equipment and a computer medium, and belongs to the technical field of automatic incineration. According to the specific scheme, the method comprises the steps of obtaining a parameter sequence value of a control parameter of the automatic combustion control system in a historical period, a current flow value of main steam flow, a flow set value of the main steam flow, a real-time value of steam pocket pressure, a hearth material layer pressure difference value and a control variable sequence value in the historical period; calculating a flow predicted value of the main steam flow; calculating flow deviation; calculating the frequency of an intermediate fan; the steam drum fan frequency is calculated; the flow fan frequency is obtained; calculating a primary frequency given value; and the primary frequency set value is dynamically corrected, and the target primary fan frequency of the primary fan is obtained and adopted to control the primary fan. The response speed of the automatic combustion control system can be effectively improved, so that the control of the frequency of the extension set can be closer to the actual production condition.

The invention provides a dimethyl sulfoxide production parameter control method and system based on reinforcement learning, and relates to the technical field of production parameter control, and the method comprises the steps: obtaining historical process operation data in a dimethyl sulfoxide production process; performing correlation analysis on the historical process operation data, and screening key control variables influencing the yield and the energy consumption; based on the historical process operation data and the key control variables, constructing a digital twinborn simulation model for providing an interaction environment for reinforcement learning; defining a state space, an action space and a reward function, and constructing a dimethyl sulfoxide production parameter control model based on reinforcement learning; based on the reward function and the digital twinborn simulation model, training and optimizing the dimethyl sulfoxide production parameter control model through a PPO algorithm, and outputting an optimal control strategy; and controlling the production parameters of dimethyl sulfoxide according to the optimal control strategy.

The invention provides an output feedback robust control method, system and device for a humanoid robot, and belongs to the technical field of humanoid robot control. The method comprises the following steps: establishing a kinetic model of a robot limb end system, and defining a state variable, a control variable and an output variable; determining a first robust control equation based on a state feedback optimal control principle, the kinetic model and an output feedback controller constructed by the output variables; converting the first robust control equation into a second robust control equation based on a mapping relation between a state variable and an output variable in the kinetic model, determining a to-be-estimated parameter vector according to the second robust control equation, and performing online estimation based on a preset adaptive law to obtain a control gain matrix; a second robust control equation is constructed based on the output variables; and based on the control gain matrix and the output variable obtained in real time, generating a corresponding control variable to be applied to the robot acra joint so as to carry out robust control on the robot acra system.

The invention discloses a helicopter prediction control method based on a physical mechanism neural network and a helicopter, and the method comprises the steps: building a helicopter prediction model, so as to describe a relation between a future state of the helicopter and a current state quantity and a control variable; constructing a predictive control model based on a physical mechanism neural network; generating a control quantity sequence by adopting a forward reasoning process of the neural network; according to the control quantity sequence and the state quantity, adopting the helicopter prediction model to perform multi-step iterative evolution in an autoregression form to obtain a predicted helicopter state trajectory; calculating a loss value by adopting a loss function of an embedded physical mechanism according to the expected instruction and the predicted helicopter state trajectory, and performing back propagation on the neural network to update neural network parameters; and inputting the state quantity obtained in real time and an expected instruction into the trained prediction control model to obtain an optimal control sequence, and applying the first control quantity to realize rolling time domain closed-loop control.

The invention discloses a guide bearing bush automatic grinding system and method based on multi-parameter cooperative control, and relates to the field of intelligent grinding of high-precision bearing parts. The system comprises a state sensing module, a rhythm analysis module, a stage identification module, a target weight adjustment module, a scheduling coordination module, a curved surface contact equalization module, a clamp thermal deformation compensation module, a regional energy integration module and an end point intelligent discrimination module. The self-adaptive frequency adjustment and priority scheduling of each control variable are realized by collecting pressure, temperature rise, displacement, vibration and microchip information in real time and calculating the change rate of the information; in combination with grinding stage identification and task target weight reconstruction, control strategy switching is optimized; and high-consistency morphology control is realized through pressure spectrum correction, thermal drift compensation and regional energy adjustment. The method has the advantages of being fast in dynamic response, high in control precision, intelligent in abnormal intervention and the like, and the grinding quality and production efficiency of the guide bearing bush are remarkably improved.

The invention provides a self-adaptive deep learning prediction control method and device suitable for the process industry. Comprising the following steps: constructing an all-condition dynamic model of a target control variable based on a deep learning technology according to historical control information of the process industry, and performing segmentation processing on a change gain curve of the all-condition dynamic model based on a self-adaptive segmentation algorithm of a threshold value to obtain a segmented linear gain model; a piecewise linear variable constraint of the piecewise linear gain model is determined, and an optimization target of a target control variable is determined based on the piecewise linear variable constraint; and obtaining a step response sequence of the target control variable at the current working condition point based on the piecewise linear gain model, and predicting a future control execution sequence of the target control variable according to the step response sequence and the optimization target. The problem that an existing predictive control method generally constructs a mapping relation between a control variable and a controlled variable based on a linear model, the nonlinear characteristic of the control variable cannot be accurately reflected, and then the predictive control accuracy is low is solved.

The invention relates to a PTFE extrusion model prediction control method and system based on a time convolution network. The method comprises the steps of collecting multi-dimensional time sequence data of an operation variable, a state variable and a controlled variable of a pushing production line in real time; performing data preprocessing; constructing a supervised learning sample by adopting a sliding time window; training the time convolution network; constructing a cost function, and setting constraint conditions; acquiring the actual die orifice outer diameter, the traction speed, the pushing speed and the pushing pressure at the current moment, and inputting the actual die orifice outer diameter, the traction speed, the pushing speed and the pushing pressure into the trained time convolutional network to obtain a prediction result of the future die orifice outer diameter; based on the prediction result, solving the cost function to obtain an optimal control sequence used for offsetting the die orifice outer diameter deviation caused by abnormal fluctuation of the pushing pressure; real-time data are collected again, the prediction and optimization process is repeated, and rolling prediction and optimization control are achieved. The problem that the outer diameter is out of tolerance due to fluctuation of raw materials is solved, and unmanned and intelligent production is achieved.

The invention discloses an intelligent intercepting well optimization control method based on overflow pollution load prediction, and the method comprises the following steps: S1, building a drainage pipe network model according to the data of a pipe network communicated with all intercepting wells, and completing the parameter calibration of hydrology and water quality in rainy days; s2, forming a control strategy scheme library containing all the intercepting wells; s3, a machine learning model is used for training, and the machine learning model of overflow pollution loads and decision variables under different rainfall conditions is established; s4, constructing an overflow pollution load value machine learning model under different rainfall conditions, and predicting an overflow pollution load minimum value under different rainfall sessions; and S5, based on the minimum value of the overflow pollution load, predicting the optimal control strategy of the control variables of all the intercepting wells by using the machine learning model of the overflow pollution load and the decision variable under different rainfall conditions. By optimizing the control strategy of the catch basin, overflow pollution control of the drainage system is achieved, and the operation efficiency of a pipe network in rainy days is improved.

The invention relates to an automobile thermal management system control method based on multi-target model predictive control, and belongs to the technical field of automobile thermal management systems. The method comprises the steps of establishing a discrete state space model of the thermal management system, setting a state variable, a control variable, a disturbance variable and a state space equation of the discrete state space model of the thermal management system, and then establishing a multi-target cost function and constraint conditions with stability, safety and low energy consumption of the thermal management system as targets; and then, based on fuzzy control logic, combining with an actual working condition, dynamically calculating a weight of each cost in the multi-target cost function, and finally solving an optimization problem according to the multi-target cost function under the determined weight in a discrete state space model of the thermal management system to obtain an optimal control input. And the optimal control input is transmitted to an actuator to regulate and control the automobile thermal management system. The system can adapt to various working conditions and environments, battery safety and comfort of a passenger compartment can be guaranteed, and energy efficiency and the service life of the actuator can be both considered.

The invention relates to the technical field of heat pump drying control, and discloses a dynamic identification control method of a closed heat pump drying system based on waste heat recovery. The method comprises the following steps: establishing a multi-level parameter monitoring network by arranging a sensor array, and continuously collecting system operation parameters to form time sequence data; constructing a waste heat distribution characteristic spectrum, and mapping the collected parameters to a thermodynamic state space to identify a coupling relationship among the parameters; analyzing a heat transfer path to determine a key control variable, and establishing a dynamic weight adjustment mechanism to generate a parameter priority sequence; and a multi-target coordination control strategy is made based on the sequence and the real-time operation state, and an actuator control instruction is output. According to the method, through global energy state recognition and dynamic priority adjustment, adaptive optimization operation of the system under different working conditions is realized, and the energy utilization efficiency and the control quality are improved.

The invention relates to the technical field of automatic control theory application and data center thermal management, and discloses a data center emergency refrigeration control method based on state space model prediction, which comprises the following steps: establishing a thermodynamic cycle state space equation; the refrigeration flow set value change rate and the water return side pressure change amount are obtained, and the dynamic heat exchange sensitivity is determined; monitoring feedback current characteristics of the driving unit, inverting cooling path resistance distribution and updating gain matrix elements; calculating a state prediction information sequence according to the temperature vector prediction deviation, and adjusting a time domain step length; the dynamic heat exchange sensitivity is introduced into prediction logic, controlled variable dynamic pressure retardation is compensated, an emergency refrigeration frequency control instruction is output, fluid fluctuation is converted into sensitivity compensation of a logic level, non-linear dynamic pressure damping of the emergency working condition is effectively impacted, the problems of control overshoot and logic instability caused by physical lag are solved, and the reliability of the system is improved. And the refrigeration response precision at the emergency conversion moment is improved.

The invention relates to a multi-target control method for a fuel cell hydrogen supply system based on switching type model prediction. The method comprises the following steps: S1, constructing a high-order nonlinear model of the fuel cell hydrogen supply system; s2, obtaining a high-load linear sub-model and a low-load linear sub-model; s3, converting the high-load linear sub-model and the low-load linear sub-model into a high-load model prediction controller and a low-load model prediction controller respectively; and S4, acquiring the actual load current of the electric pile, judging the adopted model prediction controller based on a switching strategy judged by a load current threshold, outputting a control variable by the adopted model prediction controller, controlling the fuel cell hydrogen supply system based on the control variable, and repeating the step S4 until the control is finished. Compared with the prior art, the hydrogen supply system has the advantages that the combined precise adjustment of the anode pressure and the hydrogen passing ratio is realized, so that the dynamic response performance of the hydrogen supply system and the overall operation stability of the system are improved, and the like.

The invention relates to the technical field of automobiles, and particularly provides a heavy-duty automobile cruise control method, system and device and a storage medium, and the method comprises the steps: obtaining real-time road condition information and vehicle operation data through the fusion of a vehicle-mounted sensor and a driving assistance map; predicting a future vehicle weight, and predicting a future road adhesion coefficient based on the vehicle operation data and the historical road adhesion coefficient data; constructing a state variable based on the real-time road condition information, the vehicle operation data and the road adhesion coefficient, and defining a control variable; establishing a cost function including a safety item, a fuel saving item and a stability item, and setting constraint conditions of a vehicle state variable and a control variable; based on a dynamic programming algorithm, solving a minimum value of a cost function under the condition of satisfying constraint conditions, and obtaining an optimal control variable sequence; and according to the optimal control variable sequence, the gear, the driving torque and the braking torque of the vehicle are adjusted. Through multi-source information fusion and layered dynamic planning, the cruise control performance of the heavy-duty car is remarkably improved.

Systems and methods may analyze a simulation model that includes model elements configured to operate as variants. The variants may be controlled by variant conditions that may include variant control variables. A group of variant configurations may be defined for the model where each variant configuration may include a different set of the variant control variables. For each variant configuration, the systems and methods may determine which model elements configured to operate as variants are active and which model elements are inactive. The systems and methods may also determine which model elements, if any, are always active, are never active, or are sometimes active. A visualization tool may present the results of the variant configuration analysis in the form of a heatmap, a textual report, an annotated version of the simulation model, or graph, among other forms.

A system for managing well site operations comprising a well site operations module, a chain of blocks of a distributed network, and a sensor bank and control module. The operations module generates earth model variables using a physics model, well log variables or seismic variables, or both, and a trained AI / ML algorithmic model. The chain of blocks comprises a plurality of subsequent blocks. Each subsequent block comprises a well site entry and a hash value of a previous well site entry. A well site entry comprises transacted operation control variables. The well site operations module generates production operation control variables or development operation control variables from earth model variables. The well site entry can also include transacted earth model variables and sensor variables. The sensor bank and control module provides well log variables and the operations module couples control variables to the control module to control well site equipment.

The invention provides a combustion control system for an industrial kiln. The combustion control system comprises a data acquisition and preprocessing module; the BP neural network module is used for operating three parallel BP neural network models which are respectively used for flame temperature control, flame length control and pollutant emission control, and each BP neural network model is provided with an input layer, a hidden layer and an output layer; the PID parameter adjusting module is used for dynamically adjusting the proportionality coefficient Kp, the integral coefficient Ki and the differential coefficient Kd of the PID controller module according to the output of the BP neural network module; the PID controller module is used for calculating a control quantity according to the parameters provided by the PID parameter adjusting module and outputting a control signal; and the execution mechanism module is used for receiving the control signal of the PID controller module and adjusting the mixing proportion and the flow of the first fuel and the second fuel. According to the invention, the PID control loop is optimized through the three parallel BP neural networks, and cooperative control based on flame temperature, flame length and pollutant discharge is realized.

A method for solving an integer optimization problem, comprising: defining a local representation of integer variables of the integer optimization problem as a set of qudits,; setting an state ansatz for the state of the qudits; setting a cost function as a function of the state representative of technical objectives and constraints; setting an optimization algorithm to minimize the cost function, and executing a quantum-inspired optimization method including: initializing control variables, evaluating the expression for the cost function, executing the defined optimization algorithm, conditionally rounding the current state of the qudits to a classical form, evaluating the cost function on the current state of the qudits, returning the solution found if the optimization method has converged or predetermined ending criteria have been met or repeating the optimization method with updated control variables to yield a tangible technical solution. Also, a computer program, a data processing apparatus or system, and a computer-readable medium.

The invention discloses a 1Ds-MPC control method and device for MMC multi-target step-by-step optimization, and a medium, which decouple the traditional highly-coupled MMC multi-target control problem into three independent control of alternating current tracking, circulating current suppression and capacitor voltage, and construct a single-peak one-dimensional cost function by taking the number of upper bridge arm input sub-modules as a unique search variable, so as to realize the multi-target step-by-step optimization of the MMC. Gradually shrinking the search interval by adopting a golden section proportion in the current control period, and determining a temporary input quantity; by taking the temporary input quantity as an initial value and keeping the circulating current predicted value to be zero as a target, applying error compensation sub-modules with the same quantity to the upper and lower bridge arms at the same time to obtain a final input quantity; and by taking the final input quantity as a reference, only performing local sorting and switching decision on the sub-module groups exceeding a set voltage fluctuation allowable range, and generating a switching pulse. The technical problems of complex weight factor design, control variable coupling and the like in the traditional model predictive control of the modular multilevel converter are solved.