36results about How to "Avoid forecast errors" patented technology

The invention discloses a user behavior prediction method and device, relates to the technical field of information and is mainly intended to predict user behaviors more accurately. The method comprises: analyzing a user behavior log of a target user to obtain a user attribute vector and a user behavior index vector; inputting the user attribute vector and the user behavior index vector to a pre-trained behavior prediction decision tree model to obtain predicted behavior probability of the target user, wherein the behavior prediction decision tree model is acquired by using a preset decision tree algorithm to train a sample user attribute vector, a sample user behavior index vector and a sample user; determining predicted behaviors of the target user according to the predicted behavior probability. The user behavior prediction method and device are applicable to prediction of user behaviors.

The invention relates to a waiting time predicting method for a queuing machine. The waiting time predicting method comprises following steps of collecting historical arriving moments of customers, the line lengths at the arriving moments of the customers and waiting time in a plurality of historical service periods, generating a customer waiting time table, recording the arriving moments of the customers and the line lengths at the arriving moments of the customers when the customers arrive, acquiring information about the customers required to wait in line, counting the waiting time of the customers by virtue of the customer waiting time table, then acquiring information about one queued customer accepting the service, recording actual waiting time of the queued customer, recounting the waiting time of the rest queued customers, and carrying out cyclic operation until the last service period ends. The waiting time predicting method has the advantages that the accuracy is high, a prediction error is avoided, the customers can conveniently and reasonably arrange the time, and the time availability is improved; meanwhile, the waiting time can be corrected according to queuing information of the customers at any time, so that the customers cannot miss service chances, and the vacant number rate is decreased.

The invention discloses a shopping mall building air conditioner cooling load prediction method based on GBDT, a storage medium and equipment, and the method comprises: collecting cooling load data, and carrying out the normalization processing to serve as the cooling load energy consumption prediction; establishing a load prediction model based on a gradient lifting decision tree algorithm; inputting the preprocessed data into a prediction model for training, selecting a grid search-cross validation mode, and optimizing the three hyper-parameters with the maximum influence on the performanceof the GBDT model; establishing a final cold load prediction model by completing parameter optimization of the prediction model, and obtaining a predicted cold load curve according to the parameters and the structure of the prediction model; and evaluating the prediction performance of the prediction model, adopting the prediction error for evaluation, enabling the deviation between the true valueand the prediction value to form the prediction error, and completing mall building air conditioner cooling load prediction. The method has good prediction precision, universality and applicability,and is especially suitable for large public buildings with periodically changing cold loads.

The invention discloses a preliminary assessment and prediction method for earthquake casualties, which specifically includes the following steps: 1) analyzing the main factors affecting the earthquake casualties; 2) determining the relationship between the influencing factors and the casualties; 3) according to the determined function Relationships are modeled. Compared with other models, the error rate of the model of the present invention is smaller, indicating that the model of the present invention has higher accuracy and practical reference.

The invention relates to a soil type merging and multiple regression-based soil manganese content prediction method. The divide-and-conquer control of different spatial variation characteristics showed by soil trace elements in different soil types is involved; the spatial heterogeneity of the soil trace elements can be detected by analyzing a discrete degree of spatial distribution of effective manganese content of soil; the multicollinearity problem in local regression analysis can be diagnosed through the local regression analysis; and especially in a spatial prediction process, a comprehensive prediction model is built by measuring spatial distribution characteristics of the effective manganese content of the soil under the sampling densities of different soil samples.

The invention provides a photovoltaic output power prediction method and system and a storage medium. The method comprises the steps of obtaining photovoltaic output data and meteorological data in apreset historical time period; decomposing the photovoltaic output power sequence by adopting a wavelet packet decomposition algorithm to obtain 2n decomposition sequences, and respectively carryingout single-branch reconstruction on the 2n decomposition sequences to obtain 2n reconstruction sequences; combining the 2n reconstruction sequences with the corresponding meteorological data to obtain2n data sets; respectively inputting the 2n data sets into corresponding pre-trained long-short-term memory neural network units to obtain prediction results of the 2n long-short-term memory neural network units on photovoltaic output power in a preset future time period; and performing weighted summation on the prediction results output by the 2n long-short-term memory neural network units to obtain a comprehensive prediction result of the photovoltaic output power in the preset future time period. The prediction precision can be improved.

The application discloses a tumor volume change tendency chart generating device after ablation. The generating device comprises an image acquiring module, a characteristic extracting module, a clinical parametric quantity acquiring module and a tendency chart generating module, wherein the image acquiring module is used for acquiring ultrasonic images of a target area, and acquiring images of a tumor-like lesion interest area; the characteristic extracting module is used for extracting image secretomics characteristics from the images of the tumor-like lesion interest area; the clinical parametric quantity acquiring module is used acquiring clinical parametric quantity information; and the tendency chart generating module is used for inputting the image secretomics characteristics and theclinical parametric quantity information into a forecasting model so as to obtain a tumor volume change tendency chart within a target time range. The generating device can accurately predict the tumor volume change trend after ablation. The application further discloses tumor volume change tendency chart generating equipment after ablation, and a storage medium, and the tumor volume change tendency chart generating equipment and the storage medium have the beneficial effects.

The invention discloses a lithium battery remaining charging time prediction method for guaranteeing electrical safety. The lithium battery remaining charging time prediction method is characterized in that: detection data are obtained through a charging data detection step; a battery charge state model prediction step and a remaining charging time model prediction step are used for acquiring battery charge states and outputting the remaining charging time through model prediction; in a battery charge increment comparison step, the battery charge states in the battery charge state model prediction step are compared to obtain a result; and in a model retraining step, models in the battery charge state model prediction step and the remaining charging time model prediction step are iteratively updated according to the result. The lithium battery remaining charging time prediction method has the advantages that the prediction model adaptively matches the aging state of a lithium battery, calculation resources are reasonably distributed at the local end and the cloud end, the hardware dependence is low and the like, the electrical safety of lithium battery charging is ensured through accurate prediction of the remaining charging time, and the application is wide.

The invention relates to a commodity brand feature acquisition method and device, a sales volume prediction method and device and electronic equipment, and belongs to the technical field of artificialintelligence. The method comprises the steps of: obtaining historical daily average sales volume sequences corresponding to multiple commodity brands including a target commodity brand; converting the historical daily average sales sequence corresponding to each commodity brand into a sentence character string; forming a matrix array based on all the sentence character strings, wherein each row in the matrix array corresponds to one sentence character string; converting each sentence character string in the matrix array into a corresponding sentence vector based on a word vector model to obtain a semantic vector matrix; and clustering each sentence vector in the semantic vector matrix to obtain a label corresponding to the class where the clustered target commodity brand is located, the label being a brand feature corresponding to the target commodity brand. A high-dimensional sparse feature matrix brought by adopting one-hot coding is avoided, so that the time and space required fortraining the sales prediction model are reduced, and the prediction precision of the sales prediction model is improved.

The invention discloses a single-soldier guided rocket dynamic target prediction and guidance method, which can effectively reduce the normal overload of a rocket in the flight process and ensure therelatively high hit accuracy. The single-soldier guided rocket dynamic target prediction and guidance method comprises the following steps of: measuring the motion state of the target in a ground inertial coordinate system by ground fire control equipment, transforming into the inertial coordinate system of the rocket and transmitting to a computer on the rocket; using the computer on the rocket to estimate the flight time required for the rocket to hit the target and calculating the point of intersection of a guided missile and the target according to the motion state of the target, the initial position of the rocket in the inertial coordinate system of the rocket and the set speed of the single-soldier guided rocket; in the current guidance period, controlling the rocket to fly to the target point according to the set guidance law by taking the point of intersection of the guided missile and the target as the target point; when the current guidance period ends, obtaining the positioninformation of the target in the inertial coordinate system of the rocket by solving with the computer on the rocket, solving the remaining flight time of the rocket hitting the target, and updatingthe point of intersection of the guided missile and the target; and entering the next guidance period, and repeatedly executing till the rocket hits the target.

The invention discloses a content caching method based on deep learning. The method comprises the following steps: (1) collecting user request information of edge nodes to construct a time request sequence; (2) calculating content popularity according to the request sequence data; (3) carrying out maximum and minimum normalization processing; (4) converting a time sequence prediction problem into a supervised learning problem; (5) offline training a popularity prediction model based on the time convolutional network; (6) calling a popularity prediction model to predict popularity, performing weighted summation on predicted popularity data and historical popularity data based on index average, and calculating a content value; and (7) performing cache decision by using the LRU. Popularity distribution of the corresponding content can be predicted only according to the characteristic of the content request sequence, meanwhile, balance can be achieved between long-term memory and short-term sudden memory in combination with historical content popularity information, and good effects can be achieved on the aspects of prediction accuracy and cache hit rate improvement.

The invention discloses a thin sand body plane distribution prediction method based on fractional order Hilbert transform. The thin sand body plane distribution prediction method comprises the following steps: firstly, calculating fractional order Hilbert transform of three-dimensional seismic data of 10 orders; and dividing the sand body development section into a plurality of sand body development small layer sections in the longitudinal direction, extracting an attribute plane graph of each sand body development small layer section, and optimizing an amplitude attribute plane graph corresponding to the optimal order based on the development condition of the actual drilling well in each sand body development small layer section to perform thin sand body plane distribution prediction. Thethin sand body plane distribution prediction method makes full use of the advantages that fractional order Hilbert transform can improve the longitudinal resolution, can realize perfect correspondence of the event and the acoustic logging curve on the seismic profile, effectively predicts the plane distribution characteristics of the thin sand body, and improves the accuracy of the plane distribution prediction of the thin sand body.

The embodiment of the invention discloses an image transformation system and method, a storage medium and electronic equipment. The image transformation system mainly comprises an image acquisition assembly used for acquiring a first image containing a target object and a second image containing a reference object; the image transformation component is used for inputting the first image and the second image into an image transformation model and obtaining a target image obtained by the image transformation model; wherein the image transformation model adopts a deformable attention mechanism to predict features of a target object and a reference object in a target image, the predicted features are utilized to further obtain the target image, and a partial region of the reference object in the target object is transformed into the target object. According to the invention, the transformation effect of image transformation can be effectively improved.

The invention provides an image defogging system based on an adversarial neural network, and the invention comprises the following steps: S1, selecting an image data set of RGBD, and making a defogging data set through employing an atmospheric scattering model; s2, normalizing the size of the picture in the data set to a * a; s3, building an adversarial neural network defogging model, wherein the model is divided into two parts: a generative network and a discrimination network; s4, training the adversarial neural network model by using the data set; and S5, storing the trained model, and inputting a foggy image to obtain a clear image. The invention does not need to manually extract features, effectively avoids intermediate variable prediction errors, realizes end-to-end defogging, and is simple and wide in applicability.