33results about How to "Small data requirements" patented technology

The invention relates to an automatic assembly method for a three-dimensional ship modeling part, which judges the accurate component attribution of the part thereof and two components with the same component name and different installation positions by taking a part code and gravity center data as a basis and taking a principle that the average gravity center distance sum of parts in the component is smallest as a reference. The method has the advantages of simple principle, little data demand and fast analysis process, is proved by using of having the judgment accuracy rate of 100 percent, is an optimal analysis method for developing a ship body automatic assembly program, is suitable for assembly analysis of the ship body part, also has certain reference value to assembly analysis of other industries and specialties and is a new assembly analysis method.

The invention provides a method for forecasting mid-long-term electricity consumption in combination with a population index. The method is characterized by comprising the steps of firstly, obtaining electricity consumption of each user and a resident population number in each of past k years in a to-be-forecast region; secondly, calculating a per-capita electricity consumption forecasting value by considering a per-capita electricity consumption increment coefficient; thirdly, calculating a resident population number forecasting value; fourthly, multiplying the per-capita electricity consumption forecasting value by the resident population number forecasting value to obtain a forecasting value of the sum of electricity consumption of residential users and commercial users; fifthly, calculating electricity consumption forecasting values of industrial users, electricity consumption forecasting values of non industrial users and electricity consumption forecasting values of other users; and finally, calculating a total electricity forecasting value. The forecasting method is simple in algorithm, can reduce data demand amount, is high in operability, effectively grasps key points of electricity consumption forecasting for different electricity users, is high in forecasting result precision, and can provide marketing decision support for power supply enterprises.

The invention proposes a method for grid arrangement of air pollutant monitoring stations. The method includes three steps: the first step is coarse grid layout, according to the investigation of the distribution of local population and industry, the frequency of pollution, and the scope of pollution influence, the hotspot area is encrypted on the basis of a uniform grid ; The second step is to improve the grid fitness of the hotspot area. According to the local meteorological conditions, the actual effective monitoring range of the monitoring station is reversely obtained based on the adjoint probability method, which provides data reference for the size of the fine grid; the third step is the network In the grid optimization step, according to the effective monitoring range of the monitoring station, the optimization analysis technology is used to solve the grid distribution density and grid aspect ratio, so as to realize the full coverage of the monitoring area. The present invention can efficiently and accurately propose the arrangement plan of the monitoring grid according to the actual situation, has guiding significance for the position selection of the monitoring site and the monitoring effect evaluation of the existing site, and is helpful for the management and improvement of urban air.

The invention relates to a geometric method for evaluating capacity recession of a lithium ion battery. The geometric method for evaluating capacity recession of the lithium ion battery comprises the steps that (1) geometrical characteristic values under different working conditions are extracted, wherein the number of the geometrical characteristic values is four, and the four geometrical characteristic values are the duration of continuous charging during constant-voltage charging, the maximum curvature radius of a current characteristic curve at the constant-voltage charging stage, the area under the condition of the constant-voltage charging current characteristic curve, and the maximum gradient of a discharge voltage characteristic curve at the early stage; (2) an intrinsic manifold based on the Laplace characteristic mapping method is established, namely the four geometrical characteristic values obtained in the first step are used as input of the Laplace characteristic mapping method to construct a low-dimensional manifold, namely the intrinsic manifold, embedded into a high-dimensional space; (3) a geodesic line on the manifold is used for geometric evaluation of the capacity of the battery. The geometric method for evaluating capacity recession of the lithium ion battery has the advantages that essential recession or the health state of the battery can be determined effectively, studying of a complicated electrochemistry mechanism is not needed, a complicated electrochemistry model does not need to be established, and the method can be realized easily.

The invention discloses a least squares support vector machine power prediction method based on the maximum correlation entropy criterion. The steps include: Step 1. Construct an input data set based on the power consumption in the same period of three years in history and the corresponding monthly average temperature, and use Least squares support vector machine method to build a forecasting model of electricity sales in the month to be tested; step 2, data preprocessing; step 3, normalize the input data set; step 4, for historical electricity consumption, small temperature samples Data, select the least squares support vector machine model to predict electricity sales; step 5, the key parameter σ of the model 0 and σ are optimized to determine the optimal parameters; step 6, the key parameter σ 0 Optimize with σ; step 7, select monthly relative error and annual average relative error as the evaluation index of the prediction result, and calculate the prediction accuracy at this time; step 8, make at least 3-5 predictions, and take the average value. The method of the invention can effectively predict electricity sales for users.

The invention provides a method for rapidly analyzing ecological water need season difference. Aiming at the deficiency that different types of ecological water need computation methods can not rapidly analyze the ecological water need time difference under the condition of the lack of ecological and hydrological related information of an ecological system at present, an ecological target the ecological system of which changes following the time is divided into a critical period ecological target and a water need quantity and time change target. The requirement of the biological reproduction period with obvious contradiction of ecological need water and production need water on habitat environment factors is used as a critical period ecological target, and the critical period ecological water need quantity is determined; and an average natural runoff annual time change rate for years is used as a critical period ecological target, finally, the ecological water need season difference is determined, thereby meeting the requirement of rapidly analyzing the ecological water need season difference. The invention avoids complex computational effort and information quantity demand and effectively achieves the purpose of rapidly analyzing the ecological water need season difference.

The invention relates to a method and a device for monitoring short-circuit capacity on line, in particular to the method and the device for monitoring the short-circuit capacity on line based on dynamic transfer impedance. The method comprises the following steps of: (A), replacing the conventional form of a single power supply of an equivalent target of a power system with a form of a multi-power supply to a single load node; (B), computing a pure network admittance matrix of a system according to the power grid topology information, and converting the pure network admittance matrix into an order increasing form in an X-Y coordinate; (C), aiming at power supply nodes and a load node, amending the network admittance matrix; (D), computing the dynamic transfer impedance from each power supply node to the load node; and (E), computing the internal electric potential of each power supply node according to the measured values of a power angle, terminal voltage, terminal current and the like of a motor in the running process of the system, thus computing the short-circuit capacity value of the nodes. As the method and the device are based on the dynamic transfer impedance, the change of the short-circuit capacity of the nodes can be traced quickly and precisely without any hypothesis or approximation, and the method and the device have promotional and applicable value.

The invention discloses a library construction method and reagent for pooling of large-sample-size PCR products based on high-throughput sequencing. The library construction method comprises the following steps: separately subjecting each of a plurality of PCR products derived from DNAs of different samples to annealing with a random primer having a tag sequence and carrying out an isothermal amplification reaction under the action of an isothermal amplification enzyme; mixing isothermal amplification products derived from different samples and carrying out piece selection on the mixed products; subjecting the products of the piece selection to repairing of 5' phosphorylated blunt ends and A addition reaction of 3' ends; connecting the products with linker sequences so as to obtain ligation products capable of distinguishing library information; and subjecting the ligation products to PCR amplification to obtain a library applicable to high-throughput sequencing. The method provided bythe invention realizes pooling without dependence on an ultrasonic breaking instrument, is lowered in library construction cost and complexity, reduces data waste, and improves the base randomness and coverage depth uniformity of sequencing data, thereby reducing the sequencing cost of a single sample.

The invention discloses a quick detecting method for a harmonic wave of an electric power system, and particularly relates to a method for detecting phase and amplitude of signals with different frequencies in the electric power system. The method comprises the following steps of (1), acquiring sampling points of which the number is at least two times of that of the harmonic waves through using two times of highest multiplied frequency as a sampling rate; (2), constructing a sine signal and a cosine signal which correspond with each harmonic wave frequency; (3), calculating a harmonic wave detecting parameter through fitting the coefficient before each sine item and cosine item; and (4), combining the coefficients in front of the sine item and the cosine item with the same frequency for obtaining the amplitude and the phase of the harmonic wave with each frequency. Compared with the prior art, the quick detecting method is advantageous in that a small volume of data are required; a result can be calculated just through the data with length of one half of that of a fundamental frequency signal; relative to an FFT (fast Fourier transform) method, a problem of frequency leakage can be prevented; and the method is compatible with clutter interference to a certain extent.

The invention discloses a differential equation-based rapid detection method for harmonious waves in a power system and particularly provides a method for rapidly detecting the phases and the amplitudes of signals different in frequency in the power system. The method comprises the following steps of (1) collecting sampling points at a sampling rate higher than two times a highest frequency multiplication, wherein the number of sampling points is one less than at least four times the harmonic quantity; (2) constructing a homogeneous linear differential equation of a harmonic and fitting an equation coefficient with two times the highest frequency multiplication as an order; (3) calculating the complex root of the differential equation by adopting the above equation coefficient, and calculating the frequency, the amplitude and the phase of the harmonic according to the complex root. Based on the above manner, the required amount of data is small and a result can be calculated only based on data much less than that of a cycle length of a fundamental frequency compared with the prior art. Compared with the fast Fourier transform method, the problem of frequency leakage is avoided. Meanwhile, the method can be used for calculating the frequency of each actual waveform in harmonic signals in the self-adaptive manner, and is strong in anti-jamming capability and anti-noise capacity.

The invention provides an air conditioner cooling load real-time prediction algorithm applied to an embedded type control system and relates to the field of central air conditioner system control. The air conditioner cooling load real-time prediction algorithm comprises the following steps: 1, acquiring meteorological parameters for the outdoor environment prediction of a predicted day; 2, selecting the stable running representative day of an air conditioner system as a reference day, and calculating so as to obtain the hour-by-hour load distribution coefficient of the reference day; 3, calculating the maximum hourly cooling load on the predicted day; 4, calculating the load distribution condition on the predicted day; 5, carrying out online correction on predicted load on the current running day; and 6, comparing each actual hourly cooling load data with each predicted hourly cooling load data on the predicted day, and correcting the load distribution coefficient of the reference day according to deviation conditions. The algorithm carries out prediction according to the load distribution coefficient of the basic day and the outdoor prediction meteorological parameters so as to obtain the maximum hourly load on the predicted day, and carries out calculation so as to obtain load distribution on the whole predicted day. The algorithm has the characteristics of being small in data quantity demand, simple in calculation, accurate in prediction results and great in universality.

The present application discloses a method, device, equipment and medium for obtaining the contribution degree of air pollutant sources. The method includes: obtaining the atmospheric monitoring parameter data of the target point in the current period; obtaining the same historical period of the target point in the previous year Atmospheric monitoring parameter data; To the atmospheric monitoring parameter data of the current period and the atmospheric monitoring parameter data of the same historical period of the previous year, regression analysis is carried out to obtain the influence coefficient of each atmospheric monitoring parameter; using the current Atmospheric monitoring parameter data and the influence coefficient of the period are used to calculate the contribution degree of air pollutant sources. The method of the present application correlates the data of each atmospheric monitoring parameter of the target point with each influence coefficient to calculate the contribution degree of the source of atmospheric pollutants. The data demand is small and easy to obtain, and the data processing is simple, which can effectively judge the influence of each atmospheric monitoring parameter. The impact weight can fully meet the requirements of analyzing local contribution, meteorological change contribution and regional transmission contribution.