134 results about "Discrete variable" patented technology

A method and apparatus for optimizing the operation of a power generating plant using artificial intelligence techniques. One or more decisions D are determined for at least one consecutive time increment, where at least one of the decisions D is associated with a discrete variable for the operation of a power plant device in the power generating plant. In an illustrated embodiment, the power plant device is a soot cleaning device associated with a boiler.

A light modulator comprises a plurality of discrete variable transmission electro-optic layers arranged so that light will pass successively through the plurality of layers; the light modulator has a higher transmission range than any of the individual electro-optic layers separately.

The invention provides a resource allocation method based on non-cooperative gambling in a super dense network. A double-layer network in the super dense network is analyzed to propose a shared and orthogonal hybrid spectrum allocation method based on perception; a multi-dimensional resource allocation model of base station connection, a user channel and power allocation is obtained by describing the base station connection, a channel model and system capacity; the resource allocation method based on the non-cooperative gambling is proposed to solve the multi-dimensional resource allocation problem, the algorithm describes a non-cooperative gambling model, an allowed domain is introduced to solve optimal power allocation, the 0-1 discrete variable of the base station connection is relaxed to a variable of a (0,1) section, an allocation channel is judged by normative punishment, and an algorithm for mutual iteration of the base station connection, the user channel and the power allocation is formed. The cross-layer and co-layer interference and the multi-dimensional radio resource allocation problem between a macro cell and a micro cell are solved, and the resource allocation method has certain superiority of inhibiting the interference and improving the throughput of the entire system.

The invention discloses a discrete variable structure observer-based lithium battery SOC (state of charge) estimation method. The method includes the following steps that: a rapid calibration test is performed on a lithium battery, a relation curve of SOC and open-circuit voltage (OCV) is obtained; a lithium battery discrete state space model for SOC estimation is established; a pulse discharge test is performed on the lithium battery, the parameters of the lithium battery model are identified; the terminal voltage and charge-discharge current of the lithium battery under a real-time collection condition are obtained; and a discrete variable structure observer is constructed to estimate the SOC of the lithium battery accurately. The method not only has a good SOC estimation effect, but also can ensure convergence strictly, and exhibits strong robustness to lithium battery modeling error, internal parameter perturbation and external disturbance.

The present invention leverages machine learning techniques to provide automatic generation of conditioning variables for constructing a data perspective for a given target variable. The present invention determines and analyzes the best target variable predictors for a given target variable, employing them to facilitate the conveying of information about the target variable to a user. It automatically discretizes continuous and discrete variables utilized as target variable predictors to establish their granularity. In other instances of the present invention, a complexity and/or utility parameter can be specified to facilitate generation of the data perspective via analyzing a best target variable predictor versus the complexity of the conditioning variable(s) and/or utility. The present invention can also adjust the conditioning variables (i.e., target variable predictors) of the data perspective to provide an optimum view and/or accept control inputs from a user to guide/control the generation of the data perspective.

The present invention utilizes a cross-prediction scheme to predict values of discrete and continuous time observation data, wherein conditional variance of each continuous time tube variable is fixed to a small positive value. By allowing cross-predictions in an ARMA based model, values of continuous and discrete observations in a time series are accurately predicted. The present invention accomplishes this by extending an ARMA model such that a first time series “tube” is utilized to facilitate or “cross-predict” values in a second time series tube to form an “ARMAxp” model. In general, in the ARMAxp model, the distribution of each continuous variable is a decision graph having splits only on discrete variables and having linear regressions with continuous regressors at all leaves, and the distribution of each discrete variable is a decision graph having splits only on discrete variables and having additional distributions at all leaves.

The present invention provides a coordinated voltage control method based on the coordinating and restricting a real-time updating power plant and substation, belonging to power system automatic voltage control technical field. The invention selects the substation bus as a coordination variable between plants, firstly, each substation is processed with instation automatic voltage control according to the method of ''Nine zone pictures'', and according to the control result, the constraint condition of the coordination variable of the substation is real-time updated by adopting expert rules method; a constraint condition is added to the coordinated secondary voltage control mode of the region to unify solution, reactive power control strategy of the generator is solved. The CSVC mode has no discrete variable, belonging to conventional secondary planning mode, avoiding the resolving of the problem of the complex mixed integer programming, improving the credibility of the real-time control.

The invention discloses a power transmission and distribution network integrated reactive power optimization method based on distributed computation. A generalized master-slave method is used for decomposing a power transmission and distribution network problem to three parts, namely power transmission network reactive power optimization, power distribution network reactive power optimization and boundary information exchange. A quadratic penalty function and a primal-dual interior point method are utilized for alternately computing a power transmission reactive power optimization sub-problem and a power distribution reactive power optimization sub-problem. Through exchanging boundary influence factors (boundary connecting node voltage, equivalent power and target function boundary Lagrange multiplier information), coordination participation is realized, and furthermore distributed computation for whole network reactive power optimization is realized, wherein discrete variables which commonly exist in reactive power optimization cause high difficulty in computing the boundary multiplier. The quadratic penalty function is introduced for regulating the discrete variables so that processing of the discrete variables is realized based on a differentiable power transmission and distribution global optimization target function. According to the method of the invention, a distributed computation method is utilized for settling an integrated reactive power optimization problem of the power transmission and distribution network, thereby keeping an existing computation mode of the power transmission and distribution network and realizing relatively high computation precision.

The invention discloses a coordinated planning method for energy storage, distributed power supply and distribution network based on Benders decoupling. On the basis of establishing a coordinated planning model of distribution network considering energy storage and distributed power supply, according to the idea of ​​Benders decoupling, Decompose the model into the main problem and sub-problems, and decompose the sub-problems again to form decomposition sub-problems and relaxation sub-problems, solve the decomposed sub-problems to obtain the values ​​of the continuous variables of the sub-problems, and solve the relaxed sub-problems to obtain the values ​​of the discrete variables of the sub-problems value, and form an effective cut set to feed back to the main problem. After iterative solution, the overall optimal solution that takes into account both the main problem and the sub-problem is obtained. The invention decomposes and re-decomposes the mixed integer nonlinear optimization problem containing a large number of integer variables and continuous variables, reduces the scale of the problem, and solves the difficult problem of large-scale planning considering multi-stages and multi-variables in distribution network planning. At the same time, coordinated planning is carried out for the optimal configuration of energy storage and distributed power sources and the reconstruction of the grid structure to achieve efficient solutions.

A discrete frequency conversion start-up system of the alternating current impression electromotor and its start-up method. It includes mutual inductor, amplifier, crystal brake, processor, liquid crystal screen, chip machine and keyboard. The steps are as follows: program initialization, detect the initial follow current angle, set the trigger angle of the discrete frequency conversion frequency and make the minimal trigger angle larger than initial follow current angle; frequency division trigger subprogram, frequency division trigger subprogram, frequency division trigger subprogram, frequency division trigger subprogram, frequency division trigger subprogram and flexible start-up subprogram. The invention can make the three-phase alternating current induction motor start up normally when it is heavily loaded and fully loaded.

The invention discloses a rail transit passenger flow distribution predicting model building method and a predicting method. The model building method includes the following steps: introducing the discrete variable for describing site property and scale, building a utility function of a passenger flow distribution predicting model by combining rail site generation attraction quantity, a network topological structure and relevant operation parameters, utilizing an individual representing method to convert the site generation attraction quantity belonging to aggregate data into disaggregate data and utilizing the maximum likelihood estimation to demarcate a passenger flow distribution predicting model. A rail transit passenger flow distribution predicting method is further disclosed. The two methods have the advantages of being low in data acquisition difficulty, high in practicability, accurate in prediction and the like.

The invention provides a reactive compensation optimization method of 400V distribution network areas per transformer. The characteristic parameters of the 400V distribution network areas per transformer are calculated, and distribution network areas per transformer with the power supply radius and the peak load rate higher than preset thresholds serve as the distribution network areas per transformer requiring reactive compensation; for the distribution network areas per transformer requiring reactive compensation, a network topological graph of the areas per transformer is constructed according to the line connecting mode, the line length and the load power of the distribution network areas per transformer, the network loss reactive sensitivities of all nodes in the network topological graph are calculated, and reactive compensation points are determined; a reactive optimization model is established with the smallest network loss as a target function, the reactive optimization model is solved through a prime-dual interior point method, a penalty function is introduced to achieve successive warping of a discrete variable in the optimization process, and the best compensation capacity is determined; by comparing the network loss and the voltage quality before and after the optimization configuration, the compensation effect is judged. Through the method, reactive compensation for the distribution network areas per transformer can be optimized, and the accuracy of reactive compensation for the distribution network areas per transformer can be improved.

The invention relates to a meteorological-energy-contained integrated energy system typical scene set generation method, and belongs to the technical field of electric power system operation and analysis. The meteorological-energy-contained integrated energy system typical scene set generation method is characterized in that modeling is performed on the meteorological correlation based on a Copula function theory and a Latin hypercube sampling theory; the correlation between meteorological variables is fully considered; the algorithm can significantly improve accuracy of random production simulation of the meteorological variables; and an optimal quantile theory based on Wasserstein distance is used to perform discretization on randomly simulated continuous variables to obtain a small number of integrated energy system typical scene sets. In the meteorological-energy-contained integrated energy system typical scene set generation method, the optimal quantile theory can generate extreme scenes and can greatly improve the calculating precision. The optimal quantile theory utilizes the discrete variables to enable the meteorological-energy-contained integrated energy system typical scene set generation method to be distinguished from other existing research results. The generated typical scene sets can help an operator to analyze the problem brought to operation and planning of the integrated energy system by nondeterminacy, and the analysis result of typical scenes can provide support for safe and stable running of the system.

The invention discloses a method for solving an uncertain unit commitment problem with a security constraint based on improved constraint ordinal optimization. The method comprises the steps that a discrete variable recognition strategy is blended into a rough model at first, and an ordinal optimization rough module based on a constraint condition is established; and then, a non-valid security constraint reduction strategy is brought into a precise model, and an ordinal optimization precise model aiming at a continuous variable is established. In comparison with a traditional constraint ordinal optimization method aiming at an SCUC model, the improvement strategy disclosed by the invention effectively enhances compactness of an ordinal optimization algorithm, reduces computing redundancy and thus has higher solution efficiency.

The invention relates to an optimization design method for the discretization of magnetic resonance imaging (MRI) superconducting magnets, which belongs to a superconducting magnet design technology and is characterized in that an objective function is set by a weighting method and indexes of magnetic field intensity, uniformity of magnetic field, stray field range, thread consumption, fabrication error, and the like are taken into comprehensive consideration. The use of discrete variables as optimization variables can avoid rounding and discretization errors occurring in common methods. By considering the robustness of errors in design objectives, the optimization design of the robustness is realized. The use of the design method for designing the superconducting magnets can take a plurality of design indexes into consideration in a synthesized manner, thus having the advantages of convenient processing and manufacturing and strong error robustness.

The invention discloses a multifunctional electromechanical system simulation method for simulation of an airplane power supply control management system, belongs to the computer simulation technology, and particularly relates to improvement of a multifunctional electromechanical system simulation method for simulation of an airplane power supply control management system. The multifunctional electromechanical system simulation method is characterized by comprising a hardware system and a software system, wherein the hardware system in a design scheme consists of an industrial control computer, a 64-route discrete variable I/O (input/output) card and a CAN (controller area network) bus communication card, and the software system in the design scheme comprises fuel subsystem simulator software, hydraulic subsystem simulator software, loop control subsystem simulator software and electromechanical management subsystem simulator software. Simulation functions of multiple electromechanical subsystems can be realized by a multifunctional electromechanical system simulator, various electromechanical subsystem simulators share the same hardware resources, and utilization rate of the hardware resources is improved. In addition, the multifunctional electromechanical system simulator is small in volume, light in weight and convenient in carrying. Besides, the multifunctional electromechanical system simulation method is quickly adaptive to changes of a scheme of an electromechanical system by means of updating the software, and is high in flexibility.

The invention discloses an optimization design method of a skeleton structure of an airplane assembly tool, which is used for the optimization design of a skeleton structure of an airplane assembly tool. The method comprises the following steps of: building a profile specification library file; then building a finite element model of a tool skeleton in MSC.PATRAN; generating a BDF file of the finite element model in finite element software; setting an input parameter of a genetic algorithm; transferring MSC.NASTRAN to analyze the BDF file by adopting the genetic algorithm; generating a new variable sequence according to an analysis result; meanwhile, selecting a corresponding profile in the profile library by utilizing other external programs according the new variable sequence; writing the size information of the profile into the BDF file; and then repeating the steps until optimized calculation ends. Due to the adoption of the genetic algorithm, the method solves the optimization problem of discrete variables, lightens the weight of a tool structure under the condition of meeting the rigidity requirement of the tool, shortens the design period of the skeleton structure of the airplane assembly tool and reduces the production cost.

The invention relates to the technical field of infrared spectrum wavelength selection, in particular to an infrared spectrum wavelength selection method based on a discrete multi-universe optimization algorithm. The discrete multi-universe optimization algorithm is applied to the field of the infrared spectrum wavelength selection. According to the method, the problem about the infrared spectrum wavelength selection is mapped into a discrete variable optimization problem, that is, a whole spectrum range is divided into a plurality of sub-ranges firstly, whether a corresponding sub-range is selected or not is represented by a string of binary codes only containing 1 or 0, then a conventional discrete multi-universe optimization algorithm is discretized, and iterative optimization calculation is performed finally, so that the best characteristic wavelength combination is screened out. The infrared spectrum wavelength selection method is mainly applied to the infrared spectrum wavelength selection.

Provided in the invention is a design method for access of a flexible multi-state switch to a distribution network. On the basis of simulation of a multi-scene power flow, distribution network weak positions including distribution network node voltage limiting exceeding and line load heavy loading and the like after access of a distributed power source are calculated and analyzed and a flexible multi-stage switch access location is determined; a system optimization model of access of the flexible multi-stage switch to the distribution network using economic cost minimization as an objective function is established, wherein the objective function includes a yearly conversion cost and a system operation loss cost of initial investment of the flexible multi-stage switch and various constraintconditions are met; with an improved particle swarm optimization method based on a genetic algorithm (GA-PSO), an optimization design model including discrete variables and continuous variables of the flexible multi-state switch is calculated, mixed coding is carried out on the complicated discrete variables and continuous variables, and crossover and mutation operations of the genetic algorithmare introduced into a particle swarm optimization method for improvement; and thus an optimization design plan for access of the flexible multi-state switch to the distribution network is obtained byoptimization calculation.

The invention discloses a relative sensitivity method for optimizing shapes of a discrete truss structure and a continuum structure. In the method, for the problem of variable coupling optimization of the shapes of the discrete truss structure and the continuum structure containing structure shapes and sectional dimensions, the iteration direction of a mixed discrete optimizing problem is determined according to the relative sensitivity of a target and restriction, and the sectional dimension and structure shape variables which can lead the best target performance of the discrete truss structure and the continuum structure are calculated. The regulation for determining the iteration direction is as follows: in a feasible region, the iteration direction is the coordinate direction which enables a target function to decrease most rapidly and the restriction to increase least; and out of the feasible region, the iteration direction is the coordinate direction which enables the target function to increase least and the restriction to decrease most; and if the variable corresponding to the iteration direction is a continuous variable, the discrete variable value is set to be a constant, and all the continuous variables are optimized according to a continuous variable optimization method.

A variable geometry compressor having a housing, a compressor wheel and a plurality of vanes positioned between an exducer of the compressor wheel and a housing volute. The vanes are adjustable through a range of positions, but a vane actuation system is configured to actuate the vanes through three discrete positions from among the range of positions. The actuation system has an actuator that includes an actuator housing containing a diaphragm that divides a housing chamber into two portions. Through the application of a vacuum to either chamber portion, the diaphragm drives a rod to actuate the vanes between two of the discrete vane positions, while a spring returns the vanes to the third position when no vacuum is applied. The actuator is controlled by an open-loop controller.

Generative and inference machine learning models with discrete-variable latent spaces are provided. Discrete variables may be transformed by a smoothing transformation with overlapping conditional distributions or made natively reparametrizable by definition over a GUMBEL distribution. Models may be trained by sampling from different models in the positive and negative phase and/or sample with different frequency in the positive and negative phase. Machine learning models may be defined over high-dimensional quantum statistical systems near a phase transition to take advantage of long-range correlations. Machine learning models may be defined over graph-representable input spaces and use multiple spanning trees to form latent representations. Machine learning models may be relaxed via continuous proxies to support a greater range of training techniques, such as importance weighting. Example architectures for (discrete) variational autoencoders using such techniques are also provided. Techniques for improving training efficacy and sparsity of variational autoencoders are also provided.