91 results about "Penalty coefficient" patented technology

Discrete optimization problem are solved using an analog optimization device such as a quantum processor. Problems are solved using an objective function and at least one constraint corresponding to the discrete optimization problems. The objective function is converted into a first set of inputs and the at least one constraint is converted into a second set of inputs for the analog optimization device. A third set of inputs is generated which are indicative of at least one penalty coefficient. A final state of the analog optimization device corresponds to at least a portion of the solution to the discrete optimization problem.

The invention relates to a rolling bearing fault classifying method based on FOA-MKSVM (fruit fly optimization algorithm-multiple kernel support vector machine), and belongs to the technical field of fault diagnosis of a rolling bearing. The invention aims at providing a rolling bearing fault classifying method which is fewer in initialization parameters, simple for setting parameters, high in global search capability and high in classifying accuracy. The method is characterized by comprising the following steps: extracting characteristics of each vibration signal of the rolling bearing at various states; establishing a multiple-kernel kernel function to achieve the multinucleation of a support vector machine; adopting a training characteristic set as the input of the multiple kernel support vector machine (MKSVM), and carrying out the parameter optimizing for a penalty coefficient C, each kernel function parameter and a kernel function weight gamma m of the MKSVM by utilizing a fruit fly optimization algorithm (FOA); inputting a test characteristic set into an MKSVM model to be tested, and then obtaining the classifying accuracy of the rolling bearing at a normal state, an inner ring fault state, an outer ring fault state and a rolling body fault state. The rolling bearing fault classifying method has the advantages of fewer initialization parameters, simplicity in parameter setting, high global search capability and high classifying accuracy.

Discrete optimization problem are solved using an analog optimization device such as a quantum processor. Problems are solved using an objective function and at least one constraint corresponding to the discrete optimization problems. The objective function is converted into a first set of inputs and the at least one constraint is converted into a second set of inputs for the analog optimization device. A third set of inputs is generated which are indicative of at least one penalty coefficient. A final state of the analog optimization device corresponds to at least a portion of the solution to the discrete optimization problem.

The invention discloses a pipeline leakage detection device based on a particle swarm optimization-variation modal decomposition (PSO-VMD) algorithm and a detection method. Sound emission signals without leakage and with leakage are acquired through a sound emission system; firstly, a preset scale K and a penalty coefficient a for sound emission signal decomposing are obtained through the PSO-VMDalgorithm, variation modal decomposition VMD is performed on the signals, and K intrinsic mode function IMF components are obtained; signal reconstruction is performed by adopting a method based on anenergy value, and an optimal observation signal after denoising is obtained; and a domain characteristic parameter of a time domain of a reconstructed signal is taken, and finally leakage detection is performed through a support vector machine (SVM) algorithm. It is achieved that pipeline leakage after happening can be found in time, and the problem that the mistaken alarming rate of a pipeline is high is solved.

The invention provides a support vector machine method based on chaos and grey wolf optimization, specifically,the grey wolf algorithm chaotization is combined with the support vector machine; two key parameters of penalty coefficient C and kernel width Upsilon of the support vector machine are optimized by the grey wolf chaotization algorithm with outstanding whole searching ability; the best extreme parameter value of the learning machine is obtained so that the application can obtain accurate and intelligent decision effect; the machine can help the decision mechanism effectively to make scientific decision, so the machine has important application value.

The invention provides a method for monitoring the state of a gearbox of a wind power generation set. The method includes the following steps of: collecting historical data of a wind power generation set SCADA system, screening out the active power, wind speed, cabin temperature, principle shaft rotation speed and gearbox oil temperature of the set under the healthy operation condition, and establishing a standard expert database; optimizing the penalty coefficient and the nuclear parameter of a least squares support vector regression machine by using a gravitational search algorithm, and establishing a gearbox oil temperature mapping model under the healthy operation condition by taking the active power, wind speed, cabin temperature, principle shaft rotation speed in the expert database as inputs and the gearbox oil temperature as an output and based on the optimized vector machine model; monitoring the gearbox of the wind power generation set in real time by using the mapping model, inputting the actually measured values of the active power, wind speed, cabin temperature and principle shaft rotation speed to obtain the predicted value of the gearbox oil temperature, defining the quotient of the predicted value of the oil temperature and the actually measured value as a judging index, and judging that a failure occurs in the gearbox of the wind power generation set and giving an alarm, if the statistical property of the judging index is abnormal. The method can be widely applied to early warning of the gearbox of the wind power generation set.

The invention provides a transmission and transformation project construction cost assessment method and device. The transmission and transformation project construction cost assessment method comprises the following steps: input historical sample data of a transmission and transformation project are received; iterations, inertia weight, learning factors, particle velocity of a chaos particle swarm and the population size of the particle swarm are initialized to build a chaos particle swarm model; according to the chaos particle swarm optimization, parameters of the chaos particle swarm model are optimized; according to the historical sample data and the optimized chaos particle swarm model, optimal values of the iterations, inertia weight and learning factors of the chaos particle swarm model are determined; according to the determined optimal values of the iterations, inertia weight and learning factors, penalty coefficients, insensitive coefficients and kernel function parameters of a least square support vector machine model are determined respectively to build the least square support vector machine model; input actual sample data of the transmission and transformation project are received; according to the actual sample data of the transmission and transformation project and the built least square support vector machine model, a construction cost assessment result of the transmission and transformation project is generated.

Provided is a comentropy-based integrated prediction model of traffic flow aiming at factors influencing traffic flow and having characteristics as uncertainty, and the like. A flow prediction model is established by separately utilizing a grey theory having online correction capability and mixture-kernel-based SVM. In order to improve the accuracy of the SVM prediction model, parameters of the SVM prediction model such as penalty coefficient, nuclear parameter, weighting coefficient, and the like are optimized by using an immune clonal optimization algorithm with gradient information. Then an integrated prediction model is established by using the comentropy technology, in order to further improve the accuracy of the prediction model, the weight obtained by the comentropy is corrected online. According to the invention, influences caused by uncertainty factors influencing a traffic flow can be reduced to some extent, the traffic flow can be predicted more accurately, and a preferable guidance is made for the control of the traffic flow.

The embodiment of the invention discloses a data classification prediction method and system based on an improved grey wolf optimizer. The method comprises the steps that historical data is acquired,and the acquired historical data is subjected to normalization processing and classification; the historical data obtained after normalization processing is used as training samples of a support vector machine, and the preset improved grey wolf optimizer is utilized to optimize a penalty coefficient and kernel width of the support vector machine; a prediction model is constructed according to theoptimized penalty coefficient and kernel width of the support vector machine; and to-be-predicted data is acquired, the to-be-predicted data is used as to-be-predicted samples to be imported into theprediction model, and classifications of the to-be-predicted data and a predicted value corresponding to each classification are obtained. Through the data classification prediction method and system,the problems that the grey wolf optimizer is trapped in a local optimal solution and is low in convergence speed can be solved, classification and prediction on problems in specific domains are realized, and decision precision is improved.

The invention discloses a method and a device for configuring the water resource. The method comprises the steps of A, establishing an optimized model by taking land utilization maximization and water consumption minimization as a goal and taking surface water capacity constraint, available water capacity constraint of underground water, surface water irrigation proportion constraint, irrigation water requirement constraint, land utilization constraint, water reservoir capacity constraint and irrigation water capacity non-negativity constraint as constraint conditions; B, disassembling the optimized model into an upper bound sub model and a lower bound sub model according to a region interactive algorithm, introducing penalty coefficients through a type-2 fuzzy type reduction method, converting the upper bound sub model and the lower bound sub model into corresponding linear programming models, and solving the linear programming models; C, selecting different underground water use proportions, bringing the different underground water use proportions into the optimized model to compute, and obtaining a relationship between the underground water use proportions and a crop planting structure. By utilizing the method and the device, disclosed by the invention, the plan on water resource utilization under multiple uncertain conditions can be carried out, and a reasonable agricultural irrigation method can be provided.

The embodiment of the invention discloses an output power prediction method, device and apparatus of a photovoltaic power generation system and a computer-readable storage medium. The method comprisesthe following steps: decomposing the historical output power data of a photovoltaic power generation system in a preset time period by an integrated set empirical mode decomposition, inputting the decomposed sub-sequences and corresponding meteorological data into a pre-constructed kernel limit learning machine prediction model, and determining the output power prediction value of the photovoltaic power generation system according to the prediction results of each sub-sequence output by the kernel limit learning machine prediction model. The historical photovoltaic power data is decomposed byusing a complete set of empirical modes, the nonstationarity of the photovoltaic sequence is suppressed and the prediction accuracy of the output power is improved. Through the good generalization performance and fast learning speed of the kernel limit learning machine, the prediction accuracy and efficiency can be further improved. The improved bat algorithm is used to optimize the kernel parameters and penalty coefficients of the kernel limit learning machine, which greatly improves the accuracy of power prediction.

The invention relates to a stereo matching method based on a self-adaptive weight and local entropy, belonging to the technical field of binocular stereo vision. The method includes redefining the calculation of the window weight according to the gray similarity and the spatial similarity under a self-adaptive weight theory framework; introducing a data smoothing item into the cost function, and determining the penalty coefficient of the data smoothing item by applying the image local entropy; and performing parallax refinement by applying left-right consistency check and weighted median filtering on the initial parallax map obtained by cost aggregation to obtain a high-precision parallax image. The accurate parallax map can be obtained, and the parallax accuracy is improved.

The invention belongs to the technical field of air vehicle guidance and provides a self-adaptive weighting differential game guidance method aiming at a high maneuvering target. The method includes deriving a differential game guidance law under the conditions of symmetrical complete information and asymmetrical incomplete information according to different information modes; self-adaptively adjusting a penalty coefficient in a quadratic index according to estimation results of a state filter and a current maneuvering capacity; and finally, switching the different information modes according to estimation errors and self-adaptively changing a pursuit strategy to further form the self-adaptive weighting differential game guidance method. According to the self-adaptive weighting differential game guidance method, state information measured by a sensor is fully utilized, the information modes are timely adjusted according to changes of outside environments, the problems that reasonable evidences for setting target penalty coefficients are absent and ordinary differential game guidance laws are conservative are effectively solved, guidance accuracy of air vehicles is effectively and scientifically improved, and the method has high application values.

The invention discloses a modulation signal classification method for cuckoo search-improved gray wolf optimizer-least square support vector machine. The method selects a high-order cumulant and a local mean decomposition amount approximate entropy for the characteristic parameter of a modulation signal, and utilizes cuckoo search for the second update of the wolf position to optimize the two keyparameters of a least squares support vector machine model, namely, the penalty coefficient Gamma and the kernel parameter Sigma, so as to obtain the optimal kernel limit learning machine parameter value. The method reduces the influence of noise factor on the signal recognition result, makes up for the defects of under-envelope, over-envelope and boundary effects in the traditional modal empirical decomposition, and effectively improves the defect that the gray wolf optimization global searching ability is poor and is easy to fall into the local optimal solution in processing of high-dimensional data, compared with the original gray wolf optimization result by MATLAB simulation, is it proved that the method can intelligently classify the modulated signal more efficiently and accurately, and has a good application prospect.

The invention relates to an optimization method for constrained problems based on an evolutionary optimization algorithm. For constrained optimization problems, this method adopts the method of penalty function to deal with constrained problems, and uses swarm intelligence optimization algorithm to optimize the function with penalty items. Using this optimization method, the penalty coefficient of each constraint item in the algorithm iteration process is jointly determined by the number of all individuals in the population that meet the constraint conditions, the degree of violation of constraints, and the sum of the objective functions of all individuals in the population. This avoids the problem that the penalty coefficient is too large or too small, prompts the algorithm to achieve a balance between the satisfaction of the constraints and the non-satisfaction of the individuals in the population as much as possible in each iteration, and avoids that the optimization of practical engineering problems cannot give a suitable penalty coefficient The problem.

The present invention is applicable to the field of computers, and provides a method and device for risk prediction based on nuclear extreme learning machines, aiming to solve the problem of inability to determine the optimal value of penalty coefficient and kernel width of nuclear extreme learning machines in the prior art, resulting in risk The problem of low prediction accuracy. The method includes: obtaining the operating data of a predetermined number of enterprises; standardizing the operating data; optimizing the penalty coefficient and kernel width of the kernel extreme learning machine by using the gray wolf algorithm to obtain the optimized penalty coefficient and kernel width; based on the optimized penalty coefficient Construct a prediction model of the kernel extreme learning machine with the kernel width; carry out risk prediction according to the prediction model. Through the technical solution of the present invention, the gray wolf algorithm is integrated into the kernel extreme learning machine to determine the optimal value of the penalty coefficient and the kernel width, build a more accurate prediction model, realize effective prediction of risks, improve prediction accuracy, and assist in It has important application value in the scientific, reasonable and effective prediction of business risk by financial institutions.

The invention relates to a convex-difference-planning-based calculation method of the maximum access capability of a distributed power supply in a distribution network. The method comprises: distribution network structure and parameter information is inputted, and a calculation accuracy value, an initial penalty coefficient and an iteration number of times are set; according to the distribution network structure and parameter information, a distributed power supply maximum access capability calculation model is established by considering the access capability maximization of the distributed power supply in the system; for nonlinear constraints in the maximum access capacity calculation model, conversion into a second-order cone programming model is carried out by linearization and second-order cone relaxation, and the model is solved to obtain an initial solution to the distributed power supply maximum access capability calculation model; a convex difference inequation is introduced into the second-order cone programming model and linearization is carried out, an objective function is updated; a penalty item is added into the objective function, and conversion to a convex difference programming model is carried out; the convex difference programming model is solved and whether a convergence condition is satisfied is determined; an iterative penalty factor is updated and solution calculation is carried out again; and a result is outputted. According to the invention, the calculation speed is increased substantially; and the maximum access plan of the distributed power supplyis obtained quickly.

The invention provides a method and a system for classifying breast tumor images. The method comprises the steps of acquiring lump areas in the to-be-classified breast tumor images on the basis of anFCM algorithm; acquiring mixed characteristics of the lump areas based on a Weka system; and classifying the to-be-classified breast tumor images on the basis of an SVM algorithm according to the mixed characteristics of the lump areas, wherein parameters of the SVM algorithm are optimized by using an improved FOA algorithm in advance. According to the method and the system, the Weka system is used for acquiring the morphology, shape and texture mixed characteristics of the breast tumor lesion areas to perform classification, so that the classification efficiency and accuracy are improved; andin view of the influence of kernel function parameters and penalty coefficients in a support vector machine on the classification performance, parameters of an SVM classifier are optimized by utilizing the improved FOA algorithm, so that the classification performance of the algorithm is improved.

The invention discloses a stereo matching method and a stereo matching device, which can carry out stereo matching on a first image and a second image. The method comprises steps: the first image is used as a reference image, and cost images according to different parallax distances are acquired; cost aggregation is carried out on matching cost values of each pixel point in the first image in the case of different parallax distances; according to the matching cost values after cost aggregation, a first preset penalty coefficient and a second preset penalty coefficient, dynamic programming operation is carried out on the cost images corresponding to the different parallax distances in a preset direction respectively, an accumulated matching cost value of each pixel point in the case of different parallax distances in different preset directions is acquired, and according to the accumulated matching cost value, an actual parallax value of a corresponding pixel point is acquired; and refinement processing is carried out on the acquired actual parallax value of each pixel point, and a parallax estimation map for the first image and the second image is acquired. Thus, the stereo matching accuracy can be ensured, and the complexity during the stereo matching process can be reduced.

The invention discloses a method for detecting a nonlinear oscillation during an industrial process based on improved variational mode decomposition. The method comprises the following steps: (1) collecting a set of loop output signals of a to-be-detected industrial process; (2) calculating a frequency spectrum and a phase correction signal mean frequency spectrum of the loop output signals to determine the mode number and the center frequency initial value; (3) setting the search range and the step size of a penalty coefficient; (4) calculating the sum permutation entropy obtained through a VMD decomposition corresponding to different penalty coefficients, and determining an optimal penalty coefficient; (5) performing the VMD decomposition adopting the determined mode number, the center frequency initial value and the penalty coefficient to select an effective mode; and (6) calculating whether a multiple relation exists between center frequencies of the effective mode, and judging whether a nonlinear oscillation exists or not. According to the method for detecting the nonlinear oscillation during the industrial process based on improved variational mode decomposition, the accuracyand the reliability of a nonlinear detection for the control loop of the industrial process can be improved, a data support is provided for performance evaluations and fault diagnoses, and a foundation is laid for a subsequent positioning work of multi-loop nonlinear oscillation sources.

The invention discloses a fuel cell composite power supply system and a control method, which belong to the field of fuel cells and can realize the lowest hydrogen consumption of the whole system andimprove the efficiency and the service life of the whole system. The system comprises a fuel cell module, a super capacitor module, a battery module, a data collecting module and a composite power supply controller, wherein the fuel cell module is used for providing a main power source for a vehicle; the super capacitor module is used for providing transient peak power for the vehicle; the batterymodule is used for supplementing remaining vehicle requirement power; the data collecting module is connected with the fuel cell module, the battery module, the super capacitor module and a load module respectively; and the composite power supply controller realizes optimal distribution of current to the fuel cell module, the battery module, and the super capacitor module according to the collected state information of the battery module, the super capacitor module, the fuel cell module and the load module and based on the minimum hydrogen consumption strategy considering a penalty coefficient.

The invention discloses a machine vision based precision compensation method. The method comprises the following steps: arranging a camera module and a precision compensation target; through identifying a specific element in a calibration member, determining a distortion coefficient and an intrinsic rotation angle of the camera module; carrying out approximation and fitting on identification values of multiple known points by use of a regularization network interpolation algorithm and a preset punishment coefficient; identifying an actual position point set corresponding to multiple characteristic points through the camera module; and according to an original position point set and an actual position point set, calculating rotation conversion data and offset conversion data by use of an iteration algorithm, and then distributing the rotation conversion data and the offset conversion data to the precision compensation target. The method provided by the invention has the following advantages: errors caused by manual positioning by use of a conventional vision positioning device or an automatic device are substantially reduced, and a series of complete precision compensation schemes ranging from parameter compensation of the camera module itself to coordinate and path compensation of a target of the automatic device are realized.

In a speech synthesis, a selecting unit selects one string from first speech unit strings corresponding to a first segment sequence obtained by dividing a phoneme string corresponding to target speech into segments. The selecting unit performs repeatedly generating, based on maximum W second speech unit strings corresponding to a second segment sequence as a partial sequence of the first sequence, third speech unit strings corresponding to a third segment sequence obtained by adding a segment to the second sequence, and selecting maximum W strings from the third strings based on a evaluation value of each of the third strings. The value is obtained by correcting a total cost of each of the third string candidate with a penalty coefficient for each of the third strings. The coefficient is based on a restriction concerning quickness of speech unit data acquisition, and depends on extent in which the restriction is approached.

The invention discloses a method for removing an abnormal spectrum in an actual measurement spectrum curve based on a support vector machine model. A threshold value is set by using the support vector machine classification method thought in the machine learning theory to remove the abnormal spectrum, automatic parameter optimization is carried out through a cross validation method to find out the optimal model parameter, then spectrum data are classified, the problems that due to the fact that the threshold value is set manually or is constantly adjusted, subjectivity is low, and efficiency is low are solved, the method can be applied to processing mass spectrum data in a large-scale mode, and precision and accuracy are effectively improved. A selected RBF kernel function has the advantages of being high in generalization ability and high in convergence rate. The parameter selection step for carrying out optimization on a penalty coefficient C and the interval parameter gamma in the RBF kernel function is further added, an SVM dichotomy algorithm model, namely, the support vector machine model, is built in combination with a training spectrum, and the final result of removing the abnormal spectrum is further optimized.

The invention discloses a soft measuring method and system for key variables of a lysine fermentation process based on PSO-FSVM. The method relies on a hardware platform, a measuring instrument and a computer system software used for intelligent computing. The method includes the following steps: analyzing the technological mechanism of the lysine fermentation process, selecting appropriate auxiliary variables, and establishing a training sample database according to historical pot batch data; mapping a training sample into a high dimensional kernel space, and calculating the fuzzy membership degree corresponding to each sample point in the kernel space; then conducting on-line optimization on a kernel function parameter and a penalty coefficient by a particle swarm algorithm, training the fuzzy training sample by a fuzzy support vector machine, and establishing a soft measurement model; finally, achieving prediction of the key state variables according to the latest to-be-predicted pot batch input vector. The soft measuring method and system achieve the on-line real-time prediction of the key state variables in the lysine fermentation process, and are of great significance for the parameter prediction and the optimal control of the lysine fermentation process.

The invention relates to a forecasting method of the treating effect of interferon on chronic hepatitis B. The forecasting method comprises the following steps of: collecting information related to chronic hepatitis B patients; inputting treatment outcome influencing factors of the patients; dividing the treating effect evaluation into three classes; determining specific scores of all levels of the influencing factors by taking the influencing factors as independent variable and the treating effect as dependent variable and building a rating scale and a model; randomly dividing collected illness cases into a training set and a test set, building a rating scale and a model by taking the training set as a sample, and detecting the accuracy of the rating scale and the model by taking the test set as a sample; taking the Youden index which is shown as a formula: sensitivity+specificity-1 as the evaluation index of the rating scale and the accurate model performance, and when the Youden index takes a maximum value, prompting that the forecasting performance is optimal; and enabling the forecasting performance to be optimal by adjusting parameters of mutation rate, mating rate, sub-algebra generation, and the like of a genetic algorithm as well as parameters of kernel training parameters, outlier penalty coefficients, and the like of a support vector machine.

The invention discloses a boiler combustion efficiency predicting method based on a support vector machine incremental algorithm. The boiler combustion efficiency predicting method based on the support vector machine incremental algorithm is characterized by including the following steps: (1) a kernel function is selected; (2) an initial data set is formed; (3) the initial data is pre-treated; (4) a training sample is taken out and tested; (5) a sensitivity coefficient Epsilon is 0.0001, a training precision is 0.00001 and the default values of a penalty coefficient C and a width coefficient sigmate Sigma are respectively 10 and 0.0001; (6) generalization is determined; (7) the optimum coefficient pair is selected; (8) an initial classifier Omega 0, a support vector set and a non-support vector set are obtained through training; (9) sample points which are not in line with a generalized karush-kuhn-tucker (KKT) condition, namely yif (xi)>1 are found out in a newly added sample set X1; (10) a new set is formed; (11) in terms of X, a classifier Omega and a support vector SV are determined; (12) a support vector machine predicting model on boiler combustion efficiency is established. Less input coefficients are input so as to facilitate measuring, a complicated calculation process is removed, training time of working conditions of boiler combustion is shortened, a requirement for online calculation of a distributed control system (DCS) is met and prediction precision is high.

The invention relates to a rolling bearing fault diagnosis method for optimizing an FSVM based on an improved PSO algorithm, and belongs to the technical field of mechanical engineering automation. The rolling bearing fault diagnosis method comprises the steps that firstly, a wavelet decomposition method is utilized to extract feature vectors of fault data of a rolling bearing; then penalty coefficients and kernel function parameters of the FSVM are optimized by using the PSO algorithm, and an optimized FSVM model is constructed; the extracted feature vectors are input into the optimized FSVMfor training to obtain a rolling bearing fault diagnosis model, and lastly, fault diagnosis is carried out by using the rolling bearing fault diagnosis model. The rolling bearing fault diagnosis method for optimizing the FSVM based on the improved PSO algorithm can obtain the fault diagnosis precision rate as high as possible in a short time.