77results about How to "Guaranteed Predictability" patented technology

The invention provides a photovoltaic power prediction method based on a ground-based cloud chart. The method comprises the following steps: 1) pre-estimating the movement of a cloud cluster; 2) predicting irradiance on the ground; 3) establishing a comprehensive photoelectric conversion model; 4) and carrying out photovoltaic power prediction. A plurality of digital picture processing techniques are comprehensively applied, so that requirements of problems are highlighted, pertinency and practicality are high, and combined with the traditional statistical method, the photovoltaic power prediction method provides effective technical support and method reference for a photovoltaic power refinement prediction; models are established with multi-steps based on synthesizing of multi-factors; key links in the photovoltaic prediction field are all included in the description of the method, so that universality and robustness are high; in order to combine concrete scenes and application objects used during usage, adjustment parameters are reserved in some links, so that operation is convenient and prediction effect is guaranteed.

The invention discloses a method for technically controlling secure access to an embedded system memory. The method includes the following steps of initially designing a system of a memory secure access mechanism, wherein an MPU memory protection unit is introduced, domain attributes of different grades are distributed, and an access attribute and a system initiating progress are set; secondly, designing the secure access to the memory for switching of embedded operation system process context, wherein a domain control module is set, the system process context is set, and switching of the system process content is achieved in the domain control module. The illegal access of all address space resources are detected and limited through a processor MPU hardware mechanism, the boundary control and context switching of any task of the embedded real-time multi-task operation system for accessing resources of other tasks is achieved, distributing and designing of cache, writing buffer and MPU attributes of different storage spaces in the system are achieved, and the control over secure access to all background domains in the memory is ensured.

The present invention is madder dye extracting process and its usage in dyeing mercerized wool and mercerized wool fabric. The madder dye extracting process includes the following steps: washing madder root as material, fermentation at 20-50 deg.c for 20í½30 hr, and extracting madder dye through heating at 70-90 deg.c and condensing evaporated gas for 0.5-1.5 hr. The dyeing process includes two steps: the first step of ecological pre-treatment of mercerized wool or mercerized wool fabric in treating liquid comprising sodium bicarbonate, coconut oil biethyl alcohol acrylamine as detergent and neutral organic phosphate in certain proportion at 20-60 deg.c and bath ratio of 1:20-50 for 10-60 min; and the second step of harmless mordant dyeing of the pre-treated mercerized wool or mercerized wool fabric. The present invention has rich dye resource, good dyeing effect and less damage of mercerized wool.

The invention belongs to the technical field of evaluation of components in mud shale, and discloses a method of evaluating volumes of components in mud shale by means of a conventional logging curve. The method includes the steps of: 1) on the basis of organic carbon analysis, porosity test and total rock inspection test of the mud shale after extraction, with combination of densities of the components in the mud shale, calibrating the volumes of the components in the mud shale and establishing a component volume model of the mud shale; 2) on the basis of evaluation of total organic carbon content with a [delta]logR method, with combination of a relationship of organic carbon before and after the extraction, calculating volume of kerogen, and performing optimizing calculation to obtain a BP neural network model of mineral components and pore volumes in a cross validation manner. The method, on the basis of ensuring that the sum of the volumes of the components in mud shale is 1, achieves the advantages of multiple inputs and multiple outputs of the BP neural network and solves a complex nonlinear problem between the components in mud shale and logging response.

A multi-model combined prediction method for the short-term power of a wind farm is used for improving the prediction precision of the short-term power of the wind farm.According to the technical scheme, the method comprises the steps that single-step and multi-step prediction of variable-weight dynamic combination are conducted on the wind speed by adopting multiple prediction models according to actual data collected in the wind farm, wherein multi-step prediction is implemented by adopting a rolling prediction method and equivalent dimensions addition of data, and the weight proportion among the models is regulated by analyzing the prediction precision of the models in real time; a prediction result of the output power of the wind farm is obtained according to a wind speed power curve fitted by measured data.According to the prediction method, the weight proportion among the models is changed by analyzing the prediction precision of the models on the basis of a data-driven idea, not only is comprehensive utilization of data information of different prediction methods achieved, but also the condition that the real-time prediction precision is influenced by the fixed weight coefficient can be prevented, therefore, the wind speed prediction precision is improved, and the prediction effect on the short-term power of the wind farm is guaranteed.

The present invention relates to a method of dyeing mercerized wool or mercerized wool fabrics with yellow chrysanthemum paint, which is taken as the paint, and the dyeing method is divided into two steps including ecological pretreatment and subsequent innocuous mordant dyeing: the first step, treatment conditions: baking soda of 1-3 percent of cloth weight, 1-5 percent coconut oil diethanol amide as detergent of 1-5 percent of cloth weight, 0.2-1.0g/L neutral phosphate, temperature 20-60 DEG C, duration 10-60min, bathing ratio 1 to 20-50; the second step, mordant dyeing: the mercerized wool or mercerized wool fabrics with ecological pretreatment is treated with mordant front dyeing, single bathing or post dyeing. The yellow chrysanthemum resources are rich, easy for expanding plantation, with sound dyeing and less damage to mercerized wool during dyeing.

A pedestrian trajectory prediction method based on the space-time attention mechanism comprises the following steps: acquiring image information in a scene, and extracting position information of a pedestrian in an image coordinate system; performing data preprocessing to obtain historical track coordinates of each pedestrian in the scene; by adopting an encoder Encoder, encoding the historical trajectory of the pedestrian, and outputting a feature tensor; and by adopting a decoder, iteratively predicting future trajectory coordinates of the pedestrian, wherein the encoder integrates the historical track information of each pedestrian and the interaction information among different pedestrians in the same scene through an attention mechanism. Real-time effective prediction of the pedestrian trajectory in an automatic driving practical application scene is realized, the method adapts to the processing capability of a vehicle-mounted low-power-consumption processor, the accuracy of pedestrian trajectory prediction is improved, a reliable basis is provided for practical automatic driving decision making, and the safety of automatic driving is greatly improved.

The invention provides an electricity demand predication method based on temperature and economic growth, which comprises steps of selecting the best economic indicator, acquiring historical data, acquiring average temperature for each mouth/quarter in a historical annual sample interval through calculation, acquiring an economic growth index for each mouth/quarter in the historical annual sample interval through calculation, building a year-after-year monthly/quarterly econometric model with whole society electricity as an explanatory variable, acquiring an average temperature predication value for a target mouth/quarter through calculation, acquiring an economic growth index predication value for the target mouth/quarter through calculation, and acquiring a whole society electricity predication value for the target mouth/quarter through calculation. The method of the invention has the advantages of simple prediction process, good prediction effects and the like, an effective prediction method is provided for electricity demand in a mouth/quarter level, and important reference basis is provided for an energy director and a power market analysis and predication person.

The invention discloses a risk model training method, a risk model training device, a risk identification method, a risk identification device, equipment and a medium. The risk model training method comprises the steps of: performing risk value labeling on historical travel data, and acquiring original training data; conducting companion analysis and port drift analysis on the original training data to obtain target training data; splitting the target training data according to preset time to obtain a training set and a test set; adopting a decision tree algorithm to train the target trainingdata in the training set, so as to obtain an original risk model; and adopting the test set to test the original risk model, so as to obtain a target risk model. The risk model training method effectively solves the problem that a current risk model has low identification efficiency and the accuracy rate of the current risk model is not high.

The invention provides a performance state detection method and device for an optical module in engineering application and electronic equipment. The method comprises the steps: obtaining a theoretical value and an actual value of a key parameter of the optical module under a set temperature value; calculating the difference value between the theoretical value and the actual value of the key parameter, and automatically determining the performance health state of the optical module through the corresponding relation between the pre-constructed difference value of the key parameter and the performance health state of the optical module, thereby ensuring the performance monitoring performance of the optical module and the real-time updating and predictability of the performance health stateof the optical module.

The invention discloses a multi-exit evacuation method based on prediction time. The multi-exit evacuation method comprises the following steps: 1, proposing prediction evacuation time to acquire a formula; 2, carrying out optimal planning based on a global path of A* algorithm; and 3, calculating a formula of exit evacuation capability. According to the multi-exit evacuation method based on the prediction time disclosed by the invention, by fully taking consideration of key points including a distance of every pedestrian to the exits, pedestrian expected speed, the total number of the pedestrians corresponding to each exit and different evacuation capability of different width exits, of crowd evacuation from multiple exits, a calculation method of evacuation time required by selection ofdifferent exits by the pedestrian is given, and the evacuation is simulated according to actual scenes. An experimental result proves that compared with a conventional mode, the time prediction modeldisclosed by the invention has the advantages that the evacuation to the pedestrian is simulated and predicted preferably, and effective help is provided for drawing an emergent evacuation plan.

The invention discloses a junk mail identification method and device, and relates to the technical field of computers. One specific embodiment of the method comprises the steps of analyzing a target mail, obtaining mail information of the target mail and key information with a set label, and splicing the mail information and the key information to obtain text data; carrying out word segmentation on the text data to obtain a word segmentation result, and carrying out vectorization representation on the word segmentation result by utilizing a pre-trained word vector model to obtain a corresponding feature matrix; inputting the feature matrix into a pre-trained mail recognition model for recognition, and outputting a recognition result of the target mail; wherein the mail identification modelis used for identifying whether the target mail is a junk mail or not. According to the embodiment, the junk mail identification accuracy is improved, the maintenance cost is reduced, and the junk mail identification method is not influenced by human factors.

The invention belongs to a model transfer method, and discloses a soil nutrient model transfer method based on canonical correlation analysis and linear interpolation. The method comprises steps: 1) obtaining soil spectroscopic data in different regions, setting a main sample and slave samples; 2) dividing a main sample calibration set and a testing set, using a partial least squares method to establish a main sample calibration model, and evaluating a model effect of the main sample calibration model; 3) dividing a slave sample standard set and an unknown set; 4) performing spectrum pretreatment on the slave samples; 5) using canonical correlation analysis combined with linear interpolation (CCA-LI) algorithm to transfer the model, to obtain a prediction result of the unknown set of the slave samples. The method realizes to use a soil nutrient content model to solve problems of soil nutrient content prediction in different regions. The method ensures the prediction effect of the model, and reduces time for soil nutrient chemical method measurement, reduces cost, saves manpower and material resources, and rapidly and simply realizes prediction of soil nutrients.

The invention relates to a scenic region congestion state issuing system based on image acquisition. The system comprises a computer control sub-system and a plurality of scenic spot image acquisition sub-systems. The scenic spot image acquisition sub-systems are respectively disposed in multiple scenic spots in a scenic region. The computer control sub-system is disposed in a control machine room in the scenic region. Each scenic spot image acquisition sub-system is used for acquiring the scenic spot images of the corresponding scenic spot and processing the images. The computer control sub-system is wirelessly connected with the scenic spot image acquisition sub-systems and issues scenic region congestion state according the image processing results of the scenic spot image acquisition sub-systems. By the system, people number of the scenic spots in the scenic region can be estimated through image identification, the scenic region congestion state of the scenic region can be determined by the estimation results, and important reference data is provided for tourists during scenic region and scenic spot selection.

The invention discloses an electronic equipment fault prediction method based on an LSTM enhancement model, and relates to a method for performing memory enhancement on a long-short-term neural network. In order to overcome the problem that the precision is gradually reduced along with the time sequence length when an existing LSTM model is adopted to predict an equipment fault, the method specifically comprises the following steps: step 1, collecting monitoring data in the operation process of electronic equipment, and preprocessing the monitoring data; 2, building an LSTM enhancement model,and inputting the preprocessed monitoring data into the LSTM enhancement model for training to obtain a trained LSTM enhancement model; and step 3, according to the trained LSTM enhancement model andthe monitoring data in a period of time, predicting the fault of the electronic equipment.

The invention discloses a method for calculating and identifying protein kinase phosphorylation specific sites, which is characterized in that, the method comprises the following steps: a) establishing a new protein sequence structure characterization method-amino acid three-dimensional character scoring based on the active ingredient analysis method; b) using the amino acid three-dimensional character scoring for characterizing the structure features of the protein kinase phosphorylation specific sites; c) using a Fisher criterion scoring method for selecting parameters which are closely related to the features of the protein kinase phosphorylation specific sites; and d) and establishing a protein kinase phosphorylation specific site identification model by a radial basis kernel support vector machine, carrying out the self-replacement verification respectively, and using the retaining 1/10 cross-verification and the external verification for verifying the predictive capability of the method. The method can be used for the identification of the protein kinase phosphorylation specific sites, explore the protein phosphorylation rules under physiological and pathological states, further elaborate the nature of life and the disease pathogenesis and provide the important support for developing new drugs.

The invention relates to the technical field of artificial intelligence, in particular to a label-free six-dimensional object attitude prediction method and device based on reinforcement learning. According to the technical scheme, the method comprises: obtaining a to-be-predicted target image, wherein the target image is a two-dimensional image comprising a target object; according to the targetimage, carrying out attitude prediction by adopting a pre-trained attitude prediction model to obtain a prediction result, wherein the attitude prediction model is a model obtained by carrying out reinforcement learning according to a sample image; and determining a three-dimensional position and a three-dimensional direction of the target object according to the prediction result. According to the embodiment of the invention, the attitude prediction model is trained by introducing reinforcement learning, and according to the target image, the attitude prediction model obtained by pre-trainingis adopted to perform attitude prediction, so that the problem of six-dimensional object attitude estimation based on the two-dimensional image can be solved under the condition of no real attitude annotation, and the prediction effect of label-free six-dimensional object attitude prediction is ensured.

The invention discloses a lightweight image classification method, system and device based on model pruning. The method comprises the following steps: calculating the importance of a mutual information evaluation channel; performing sorting and grouping according to the importance of the channels; setting an attention regularization loss function according to grouping and optimizing the model; pruning unimportant channels; and optimizing the model again, and recovering the precision of the model. According to the method, system and device, the compression effect on the deep neural network is obvious, the precision loss caused by pruning is reduced while the storage and calculation consumption of a large-scale deep network model is reduced, and the method, system and device can be better applied to mobile terminal equipment for image classification.

The invention belongs to a model transfer method, and discloses a soil nutrient model transfer method of different regions based on multi-algorithm recommendation. The method comprises the following steps that (1) soil spectroscopic data of the different regions is acquired, and a main sample and an auxiliary sample are set; (2) a calibration set and a testing set of the main sample are divided, while a calibration model of the main sample is built based on partial least squares, and the model effect is evaluated; (3) a standard set and an unknown set of the auxiliary sample are divided; (4) the auxiliary sample is subjected to the spectroscopy pretreatment; (5) through multi-algorithm model transfer, a prediction result of the unknown set of the auxiliary sample is obtained; (6) through evaluation and analysis, a model transfer algorithm with the best effect is recommended. According to the method, solving the problem of soil nutrient content prediction in the different regions by application of a soil nutrient content model is achieved, the model prediction effect is guaranteed, while the chemical method measurement time of soil nutrients is reduced simultaneously, so that the cost is lowered, the manpower and material resources are saved, and the soil nutrient prediction is achieved in a fast and simple way.

The invention belongs to a model transfer method, and discloses a segmented direct correction and slope/intercept correction combined soil nutrient model transfer method. The method comprises the following steps of: 1) obtaining soil spectrum data among different regions, and setting primary and secondary samples; 2) dividing a primary sample correction set and a primary sample verification set, establishing a primary sample correction model via a partial least squares method, and evaluating the model effect of the model; 3) dividing a secondary sample standard set and a secondary sample unknown set; 4) carrying out spectrum preprocessing on the secondary samples; and 5) carrying out model transfer by adoption of a segmented direct correction and slope/intercept correction (PDS-S/B) combined algorithm so as to obtain a prediction result of the secondary sample unknown set. According to the method provided by the invention, the difficulty of predicting soil nutrient contents among different regions is solved by using a soil nutrient content model, so that the time of measuring the soil nutrients via a chemical method is shortened while the prediction effect of the soil is ensured, the cost is reduced, the manpower and material resources are saved, and the soil nutrient prediction is rapidly and simply realized.

The invention relates to the field of an embedded system, and discloses a method for realizing an affair mechanism of an embedded system and a device thereof, and aiming at the problem that the data can not be automatically reset when the data is abnormally configured in the embedded system. The method comprises the following steps of: a master control plate receives and analyzes webmaster data, configures the analyzed data onto a service plate, and completes the storage of the web master data and the reset of the service plate by judging service plate configuring information. The device comprises a receiving and analyzing module, a configuring module, a storing module and a data resetting module. The invention can be widely applied to the communication industry, can automatically complete the resetting function when the configuration of the business data is fault to finally guarantee the stable running of the equipment, improves the equipment efficiency of the equipment, greatly reduces the cost of the equipment, and has good development potential and application market.

The invention discloses a link prediction method in a social network, which belongs to the field of data mining, and comprises the following steps of: preprocessing, constructing context sensing embedding, constructing a link prediction candidate set, and combining the following text sensing embedding with the link prediction candidate set for similarity calculation. Context scenes with similar behavior modes of users are found by constructing context sensing embedding, so that information missed by a previous traditional method is captured, potential social influence is instantiated, link prediction accuracy in a social network is improved, accurate representation of the users in the social network is obtained by the link prediction accuracy and node structure information, the accuracy and performance of social network link prediction are improved, and the problems that only influence interaction between nodes and adjacent nodes is considered in a traditional method, and hidden socialinfluence information is not mined from the level of the whole social network are solved.

The invention discloses an inspection robot dynamic path planning method, device and equipment based on a particle filter algorithm, and a computer readable storage medium. The particle filter algorithm is adopted to carry out environment situation analysis while the inspection robot dynamic path planning operation is carried out, and therefore, the planning process has certain predictability for dynamic sudden obstacles; every time the inspection robot executes a planned step length, the states of the obstacle and the inspection robot are predicted and updated by adopting a particle filtering algorithm on the basis of current environment information obtained by a sensor, and according to the updated environment information, the particle filtering algorithm is iteratively executed for preset times in each path rolling window to optimize a local path, and the convergence process of the optimal solution is accelerated while the estimation accuracy of the algorithm is improved; and a particle filter algorithm and a rolling optimization strategy are combined, and a rolling window is adopted to locally plan paths in a segmented manner, so that the computational complexity of global path planning is reduced while the predictability of a planning process is ensured.

The invention provides a method and device for training a prediction model used for predicting customer transaction behaviors, relates to the technical field of artificial intelligence, and can be applied to the technical field of finance. The method comprises the following steps: determining a plurality of pre-selected feature types according to a customer sample data set, wherein the customer sample data set comprises customer basic information data and telephone consultation information data of a customer for a target product; determining a plurality of target feature types from the plurality of pre-selected feature types, wherein the target feature types are data types of which the contribution degree to the customer purchase behavior result is greater than a preset contribution degree threshold value and the correlation between every two target feature types is smaller than a preset correlation threshold value; screening out data related to the target feature type from the customer sample data set to generate a target training set; and training a preset algorithm by using the target training set to generate a prediction model.