78 results about "Offline optimization" patented technology

The invention discloses a vehicle detection method based on a convolutional neural network. The method includes the step S1 of collecting vehicle samples and non-vehicle samples and classifying the vehicle samples, the step S2 of preprocessing the samples, the step S3 of training a CNN vehicle detector, the step S4 calculating an average similarity table of a characteristic pattern, the step S5 of constructing a similarity characteristic pattern set, the step S6 of obtaining a CNN-OP vehicle detector, the step S7 of obtaining detection images, the step S8 of preprocessing the obtained detection images, the step S9 of constructing an image pyramid for the detection images, the step S10 of extracting characteristics, the step S11 of scanning characteristic patterns, the step S12 of classifying the characteristics, and the step S13 of combining detection windows and conducting output. An offline optimization scheme is put forward, the convolutional neural network which is completely trained is optimized, the strategy of scanning the windows after extracting the characteristics is adopted at the detection stage, and therefore the characteristics are prevented from being repeatedly calculated, and the detection speed of the system is increased.

The invention discloses a method for generating a double-deck multi-objective locomotive optimized manipulating sequence. The method is characterized by comprising steps of 1. collecting current vehicle parameters of locomotives, route data and historical data of driving of drivers and preprocessing the data; 2. defining all routes as different types of child segments based on the equivalent summing gradient obtained in route data preprocessing; 3. confirming a manipulating method based on the segmentation result in the step 2, and building a base layer multi-objective calculation model comprising a traction calculation model; 4. building an upper-layer multi-objective calculation optimized model based on the output provided by the base-layer multi-objective calculation model; 5. finishing the optimized locomotive manipulating by combining the upper-layer optimized calculation model and the base-layer optimized calculation model. The method greatly shortens running time, and can be used in offline optimization and short-distance online optimization.

A method of revenue management of an inventory of items in order to optimize the availability of one or several item(s) requested by a user, method comprising the steps of calculating an online estimate of a first element associated with the request; calculating an online estimate of a second element associated with the request; comparing the estimate of the first and second element to determine whether or not requested items can be granted to the requestor. The optimized calculations take place in real time and are handled by a single on-line system that does not need to rely on uploads of recommendations from an external off-line optimization system.

The invention discloses a motivation trajectory optimization method of space robot kinetic parameter identification; and the method can realize motivation trajectory optimization of space robot kinetic parameter identification, and can satisfy PE conditions of parameter identification and the mechanical arm joint motion restraint to improve the convergence rate and accuracy of space robot parameter identification. The incomplete characteristics of a free floating space robot determine a parameter identification model regression matrix A (k) to not only include related positions and related speed related to the mechanical arm joint motion trajectory and but also include the position, the gesture, the speed and the angular velocity of a base indirectly related to kinetic parameters to be identified; and the numbers are solved according to a system kinetic model, so that prior information of the kinetic parameters to be identified is used in offline optimization of a motivation trajectory.

The invention discloses a method for optimizing gasoline blending in an offline manner. The method includes building a gasoline blending model; optimally computing the built model by the aid of a particle swarm optimization algorithm. The minimum blending cost, the minimum product mass surplus and the minimum manganese-additive dosage are comprehensively taken into consideration in the model on the premise that requirements of product index constraints, component inventory constraints, manganese-additive dosage limit and material balance constraints are met. The method has the advantages that equality constraints in the blending model are processed by the aid of the particle swarm optimization algorithm in a novel processing mode, so that a result obtained after particle swarms are updated at every turn can assuredly meet conditions of the equality constraints, the scale of population can be guaranteed, and the arithmetic speed can be increased.

The invention discloses a vehicle speed tracking method of a radial basis function neural network based on particle swarm optimization. The invention constructs an automobile dynamic model through anengine model, a transmission system model, a vehicle model and a brake model. The parameters of radial basis function neural network model are calculated by gradient descent method, and the PID controller adjusts the parameters adaptively by radial basis function neural network model. Parameters of particle swarm optimization are obtained by off-line optimization of particle swarm optimization algorithm. The PSO parameters are initialized and assigned to the radial basis function neural network PID controller. The initial throttle opening or the initial brake pedal position is obtained by theinitialized radial basis function neural network PID controller and input to the vehicle dynamics model to calculate the actual tracking speed. The actual tracking speed and the output of PID controller are inputted into the neural network, and the parameters of RBF neural network and PID controller are adjusted according to the feedback error of the speed. The invention realizes safe and stable tracking target speed.

The invention discloses an offline-optimization and online-predication torque distribution method. The offline-optimization and online-predication torque distribution method comprises the steps of when a driver is going to drive a vehicle to one destination, manually selecting a driver model first and then inputting the destination; enabling a vehicle controller to optimize a travel for a learningresult of the driver according to the selected driver model; planning a path according to a starting point and a terminal point of the travel; using the selected terminal point of the travel to obtain road condition information by using a vehicle-mounted navigation system; combining the obtained road condition information, the selected driver model and a battery SOC value to perform preliminary torque distribution optimization by using a dynamic planning method; and then using model predication control to perform dynamic real-time optimal control on the basis of a preliminary torque distribution optimization result. By carrying out further real-time optimal control on the basis of existing preliminary optimization control, the calculation amount of dynamic optimization is reduced, and meanwhile, real-time online control of fuel economy of a hybrid automobile is realized.

The invention discloses a mixed intelligent boiler comprehensive combustion optimization method. A simple and practical index positively related with boiler efficiency is established aiming at the problems in boiler combustion efficiency and coal mill power consumption optimization, the index is combined with a coal mill power consumption index, the boiler comprehensive combustion optimization method with high learning capacity is provided, and economical efficiency is optimized. According to the technical scheme, through data acquisition of a boiler, a model is established aiming at the index of boiler combustion efficiency and the coal mill power consumption index, parallel optimization algorithm optimizing and other means are applied, and the boiler comprehensive combustion optimization method is determined; by means of the method, the efficiency of boiler comprehensive combustion optimization can be effectively improved, and offline optimization and online real-time combustion optimization can be carried out.

The invention discloses a DPF carbon loading computing method, and relates to the technical field of diesel engines. According to the technical scheme, the DPF carbon loading computing method comprises the steps that 1, actual engine operation data is acquired; 2, the data acquired in the step 1 is compared with a weighing result of a DPF; and 3, the data acquired in the step 1 is put into a computing model for simulation computing. The DPF carbon loading computing method has the advantages that only a few actual bench tests and road tests need to be conducted, and the needed data is recordedand computed during the tests; by means of the computing model in the step 3, offline simulation computing is achieved, and meanwhile offline optimization without needing an engine system entity is conducted on DPF carbon loading calibration parameters; and by means of the model and the calibration method, as much as 95% of road test time, staff cost and fuel consumption can be saved, and the research and development cost and the calibration cost of the DPF product are significantly reduced.

The invention discloses a city area traffic emergency plan method for a severe haze weather. Targeting at a haze weather which occurs recently, the invention establishes an emergency plan system which is formed by a database module, a mesoscopic simulator module, an offline optimization platform module and a preferred plan module. The system simulates a pollutant exhausting condition and a pollutant concentration diffusion condition of city area motor vehicles and on the basis of the conditions, the plan system obtains a reasonable and effective emergency plan method through computation and optimization, that is, a novel scheme for limiting a proportion of motor vehicles which enter city areas and timing of signal lamps in the areas. Under a severe weather, the method guarantees that the weather of city areas does not further deteriorate because of pollutant exhaust of motor vehicles and also prevents threats on human health by pollutants.

The invention provides a method and system for achieving railway locomotive operation control from an off-line mode to an on-line mode. A sequence pattern excavation method is used for obtaining a control gear sequence from original operation data of a locomotive in an off-line mode, an off-line optimization algorithm is used for optimizing the time distribution proportions of specific gears in the control gear sequence, a neural network model is constructed according to the time distribution proportion sequence with the optimal energy consumption, association rules of the specific locomotive, line parameters and the control gear sequence are obtained according to the neural network model, finally, the control sequence with the optimal energy consumption is used for guiding the locomotive to operate in an on-line mode. Due to the fact that off-line calculation is not affected by time factors so that an off-line part can own better optimized space, and in the operation process of the locomotive, a good energy-saving effect can be achieved by utilizing the control gear sequence of operation of the locomotive, wherein the control gear sequence is obtained in the off-line mode. In addition, an on-line control operation result of the locomotive serves as the data input of the off-line sequence pattern excavation method and the optimization algorithm, and therefore off-line learning can be adjusted and optimized continuously.

The embodiment of the invention provides an auxiliary traction control system for a train. The auxiliary traction control system for the train comprises vehicle-mounted processing equipment and a human-computer interaction terminal, wherein the vehicle-mounted processing equipment carries out measurement through various sensors and a speed measuring radar, calculates an actual running state of the train, calculates a train traction force, and sends auxiliary train traction control information consisting of the actual running state of the train and the train traction force to the human-computer interaction terminal; the human-computer interaction terminal displays the auxiliary train traction control information. According to the auxiliary traction control system for the train, provided by the embodiment of the invention, the running state information of the train can be automatically acquired in real time, the auxiliary traction control information is generated and is compared with a pre-stored offline optimization traction control scheme, and auxiliary information for traction control operation is provided for a driver and is sent to a ground center through a wireless communication network.

The invention discloses an ultra-short-term wind power prediction method according to offline track characteristic optimization and real-time extrapolation model matching, and belongs to the fields of development and utilization of renewable energy sources. The method comprises the following two steps: (1) establishing a model in an off-line mode and optimizing parameters, namely dividing a track formed by historical data of a wind power time sequence or a wind speed time sequence into different forms according to the given characteristic quantity, respectively establishing a prediction model for each form, and optimizing the parameters; (2) performing real-time prediction, namely calling a corresponding prediction model according to the forms of the track of latest measured data. According to the method, the time-varying characteristics of the wind power sequence are fully measured, and the statistical characteristics and change rules of a wind power sequence at different time intervals are reflected. The defect that dynamic change of the wind power sequence and statistical characteristics of the wind power sequence at the different time intervals can not be comprehensively reflected in a traditional wind power prediction method is overcome. The coordinative optimization among prediction models (or algorithms) is realized. Therefore, the prediction accuracy is improved, and the prediction efficiency is also improved.

The invention relates to a single-point offline optimization system and method for traffic signals, and the system and method can formulate an appropriate optimization strategy according with a current intersection based on the data condition of the current intersection, and optimize the signal timing through the data information. Single-point intersections are classified according to intersectiondata conditions, different offline optimization strategies and methods are adopted for different classifications, and two single-point offline optimization strategies are defined, so that data resources can be utilized more effectively, and efficient optimization of the single-point intersections is realized.

The invention provides a hybrid and intelligent updating method for a boiler efficiency combustion optimization model. At present, online real-time optimization of the boiler efficiency is always a difficult problem for industrial research personnel, because the change of coal types and the time-varying characteristics of the combustion law can cause the model prediction errors to be increased, and then boiler combustion optimization can be caused to lose effect. According to the hybrid and intelligent updating method for the boiler efficiency combustion optimization model, an updating method capable of judging whether the model needs to be updated or not and correcting the re-modeled model in the updating process is specifically constructed, the model can be updated effectively and timely, and therefore the prediction precision of the model is improved; and both online optimization and offline optimization can be carried out.

The invention relates to an implementation method for controlling to separate dynamic control of an insulin pump from computing. The method comprises the following steps of adopting an offline optimization method to store blood glucose values through a front-end insulin pump, and when building wireless communication between the front-end insulin pump and a rear-end computer, immediately uploading blood glucose data by the front-end insulin pump, and immediately emptying a storage space after data transmission is completed; after the rear-end computer receives the blood glucose values transmitted by the front-end insulin pump, carrying out data analysis and generating new control parameters, sending the new control parameters to the front-end insulin pump when wireless connection is built for the next time, and using the front-end insulin pump to carry out short-term data acquisition based on the new control parameters to verify the new control parameters. According to the implementation method for controlling to separate dynamic control of the insulin pump from computing, on one hand, the offline data storage capacity requirement of the front-end insulin pump is reduced, and on the other hand, the insulin control has better robustness.

The invention discloses an offline optimization / online switching reactive compensation method, and belongs to the technical field of electric power system electric energy quality control. The reactive compensation method is a static reactive compensation control method that a control strategy combining a chaotic PSO algorithm and a PI controller is adopted based on the conventional PI controller to perform offline optimization and adjustment on the parameters of the PI controller, and selection of different control strategies is performed in SVC system control by aiming at different load state. According to the reactive compensation method, the response speed and the compensation effect of a dynamic reactive compensation device can be enhanced so that the online implementation efficiency of the SVC system and the feasibility and the stability of the system can be enhanced.

Provided is an online optimization method of an industrial unit vapour system based on GPU acceleration. The method is based on a mathematical model of vapour system operational costs, and takes into account the constraint conditions such as conservation of mass and energy, turbine power capability, vapour requirements at each level and the like during an actual industrial process, regards the steam extracting quantity of an extraction-condensing steam turbine and the switching value of an electric pump\ a pump turbine and the like as operational variables, real-timely collects operating data of industrial devices, utilizes a parallel cooperative particle optimization algorithm of GPU acceleration to figure out optimized solutions, realizes online optimization of a vapour system, and decreases operational costs of the vapour system. The online optimization method has the advantages of being capable of providing guidance for off-line optimization of devices, combining a technology of APC to realize closed-loop real-time optimization of industrial unit vapour systems and reducing operational costs of devices, and being applicable to online optimization of all kinds of industrial unit vapour systems, therefore the online optimization method has an extensive adaptability.

The invention provides a sagging control method, and the method comprises the steps: proposing sagging control-decoupling sagging control; proposing an improved PSO (Particle Swarm Optimization) algorithm is proposed, which is an improved MMPSO with a multi-group and multi-speed updating mode; building an offline optimization model based on the improved PSO algorithm, wherein each inversion power supply in the model is in series connection with a reactor; supplying power to a load through parallel connection. The method can be used in the fields of new energy microgrids and uninterrupted power supply, can meet the parallel connection requirements of a plurality of inversion power supplies, can effectively reduce the parallel connection loop current, and improves the stability and reliability of a parallel connection system.

The invention relates to the technical field of injection molding control, in particular to a real-time optimal control method for a deep neural network of an injection molding machine. The real-time optimal control method comprises the following steps of S10, establishing a dynamic mathematical model of an injection molding filling process of the injection molding machine, and converting a flow rate control problem of the injection molding machine into solving an optimal control problem with constraints; S20, carrying out iterative offline optimization solution on the dynamic mathematical model to generate an optimal state-control data set based on different initial state starting points; S30, training the deep neural network by using the optimal state-control data set, and the deep neural network learns a mathematical relationship of nonlinear mapping between an input state and an output optimal action; and S40, collecting current state data of the injection molding machine, inputting the current state data into the trained deep neural network, and outputting a control signal of the injection molding machine. According to the real-time optimal control method for the deep neural network of the injection molding machine, the optimal control is combined with the deep neural network, so that the current system state of the injection molding machine quickly responds to a current optimal input control signal of a servo valve motor of the injection molding machine in the next step.

The invention discloses a method for calculating a self-optimizing controlled variable during forced circulation and evaporation control in the process of alkali liquid concentration and production. The method is characterized by including: through constraint control on inlet steam pressure and product alkali liquid concentration and stable control on separator liquid level, performing offline optimization on multiple disturbance circumstances acquired by sampling in disturbance space through a numerical value optimization algorithm to acquire corresponding optimal values of multiple groups of output variables; performing feature value decomposition on a matrix structured by the optimal values of the output variables; acquiring the self-optimizing controlled variable through a feature vector corresponding to a minimum feature value and a group of the output variables so as to quickly and effectively determine a self-optimizing controlled variable of a bottom-layer control loop. Online realtime optimization of the process of forced circulating and evaporation control can be indirectly realized by only online tracking a constant set value of the self-optimizing controlled variable.

The invention belongs to the field of software transplantation, and particularly relates to a dynamic and static fusion binary translation method and system based on a dynamic link library. The methodincludes: dividing a program by taking a function as a unit, and if the function is a third-party library function, executing the program in a local library replacement mode; if the indirect jump branch instruction exists in the function, placing the function in a dynamic translator part for translation execution, and if the indirect jump branch instruction does not exist in the function, statically translating the function by taking a basic block as a unit, recording relocation information translated by the function, and generating a function relocation information table; analyzing and optimizing the translated target code according to the static analysis information and the relocation information, and generating a dynamic link library for calling the target program in the dynamic execution process; and during dynamic execution, preferentially executing the optimized function according to the relocation information table and the dynamic link library. According to the method, the advantages of static binary translation offline optimization are fully utilized, codes needing to be translated and optimized in the dynamic execution period are statically executed, translation expenditure is reduced, and execution efficiency is improved.

The invention relates to an online virtual resource allocation method based on virtual machine pre-deployment, and belongs to the field of cloud computing and data center resource management. An online virtual resource allocation framework based on virtual machine pre-deployment is constructed, and the framework realizes organic combination of virtual machine deployment offline optimization and virtual machine deployment online decision making through a virtual machine pre-deployment mode; secondly, designing an online virtual resource allocation process based on the framework so as to realizerapid virtual resource allocation, and meanwhile, improving the physical resource utilization rate of the data center after resource allocation by preferentially using a pre-deployed virtual machineto complete resource allocation; quick response to a virtual machine request can be realized through virtual machine online deployment, and offline resource optimization can be performed on virtual machine deployment, so that the physical resource utilization rate of the cloud data center is improved.

The invention proposes an LCL filtering-based RBFNN segmentation online optimization passive control system and method. Grid-side three-phase voltage and current signals are acquired through three-phase voltage and current signal sensors respectively, and coordinate transformation is performed; a passive control Hamiltonian model based on an IDA-PBC algorithm is built according to coordinate axisvoltage and current, and improved d,q-axis switching function is built; particles containing parameters such as an RBFNN learning rate and a momentum factor under different load resistances are subjected to offline optimization through PSO to obtain an optimal particle set; the load resistances calculated by DC-side voltage and current sensor signals are used as segmentation triggering conditions,an RBF-PID model is built through the optimal particle, and a controller model is used to realize segmentation optimization control; the RBF-PID after parameter optimization is used for optimal solution for stably-operating Im; and according to the optimized Im and in combination of the d,q-axis switching function, control is carried out, IGBT control signals are generated by SVPWM, and rectification control is realized. The control precision is higher, and the robustness is better.

The invention relates to a rendering method, in particular to a method for rendering terrain through offline optimization, belonging to the technical field of computer graphics. The method comprises a step of: (4) filtering useless grids in a map. The method for rendering the terrain through the offline optimization is beneficial to further optimizing a map resource through the arts and reducing the size of the map resource; simultaneously, more collision detection about the terrain is reduced, and the running efficiency of a CPU (Central Processing Unit) is improved, therefore, the fixity of grid information is utilized to the maximum extent, and visible storage is performed in an offline manner; therefore, drawing of the grid is reduced to the maximum extent, and a purpose of increasing the number of frames is achieved.

The invention relates to a cold rolled strip steel galvanizing line process control method and belongs to the technical field of cold rolled strip steel processing line automation. The technical scheme comprises the following steps of classifying to-be-produced strip steel through utilization of a steel type classification algorithm; carrying out data matching based on steel type groups and stripsteel specification data, and correcting computed values through adoption of values if the same type of strip steel is produced previously; solving corresponding parameter set values through adoptionof a linear difference value algorithm; carrying out production if the set values are confirmed to be correct; and storing all history data, carrying out offline optimization and fitting a set value curve. The method has the beneficial effects that a galvanizing line process control level is improved, use convenience of production line workers is improved, misoperation and a defective rate are greatly reduced, and powerful support is provided for an information system by relatively precise production data and product information.

The invention discloses a hot rolling and rough rolling load distribution method considering convexity of an intermediate billet, and belongs to the technical field of control. According to the method, a mathematical model of a hot continuous rolling rough rolling process is established, a target optimization function is determined on the premise of considering the convexity of an intermediate billet, and under the optimal setting constraint condition, rough rolling load distribution is optimally designed by using a three-population differential evolution particle swarm algorithm. According tothe method, distribution method combining offline optimization and online control and a three-population difference algorithm are adopted, the calculation speed is high, the online calculation requirement is met, and load distribution optimization setting and intermediate billet convexity prediction can be achieved.

The invention provides an offline image optimization method and system. The offline image optimization method comprises the following steps: an offline user sends out a request of visiting a first page; the request is positioned into a network server storing the first page; the network server monitors the request and judges whether a first image exists in the first page; if yes, the first image is compressed so as to generate a second image; and the second image is used for replacing the first image in the network server. According to the method of the embodiment of the invention, on the one hand, the offline optimization can be carried out by fully using offline resources so that the reduction of the performance of an online server caused by the image optimization and the security risk are avoided, a developer does not need to manually optimize the image, thereby the development efficiency is improved; and on the other hand, the image is compressed and optimized so that the size of a paging file can be reduced, the flow expenditure is reduced, the cost of network bandwidth is saved, the transmission time of the image is shortened, the waiting time of the user is further reduced, and the user experience is promoted.

The invention discloses a dynamic reactive power compensation device control strategy verification system. The system comprises a main circuit simulation module, a new energy resource power station and power grid simulation module, an optical fiber conversion interface, a level conversion interface and a human-machine interaction interface. Various complicated power grid operation conditions are simulated by the dynamic reactive power compensation device control strategy verification system, and the feedback signal of a device link system can be given out according to requirements, so that thewiring debugging is simple and convenient; various control strategies of a dynamic reactive power compensation device can be verified, and offline optimization can be carried out on the control performance of the dynamic reactive power compensation device, so that the debugging and departure period before field operation is greatly shortened, offline verification and offline parameter optimization of the control strategies of the dynamic reactive power compensation device are realized, and the problem that operation performance optimization is needed for a part of new energy resource power stations in which the dynamic reactive power compensation device is put into operation but parameter online optimization cannot be carried out because of the restricted condition of site operation is solved.