199 results about "Moving-average model" patented technology

Multi-step statistical modeling in one embodiment of the present disclosure enables anomaly detection, forecasting and / or root cause analysis of the energy consumption for a portfolio of buildings using multi-step statistical modeling. In one aspect, energy consumption data associated with a building, building characteristic data associated with the building, building operation and activities data associated with the building, and weather data are used to generate a variable based degree model. A base load factor, a heating coefficient and a cooling coefficient associated with the building and an error term are determined from the variable based degree model and used to generate a plurality of multivariate regression models. A time series model is generated for the error term to model seasonal factors which reflect monthly dependence on energy use and an auto-regressive integrated moving average model (ARIMA) which reflects temporal dependent patterns of the energy use.

A method of predicating ultra-short-term wind power based on self-learning composite data source includes following steps. Model parameters of an autoregression moving average model are obtained by inputting data. A predication result is obtained by inputting data required by wind power predication into the autoregression moving average model. A post-evaluation is performed to the predication result by analyzing error between the predication result and measured values, and performing model order determination and model parameters estimation again while the error is greater than an allowable maximum error.

Multi-step statistical modeling in one embodiment of the present disclosure enables anomaly detection, forecasting and / or root cause analysis of the energy consumption for a portfolio of buildings using multi-step statistical modeling. In one aspect, energy consumption data associated with a building, building characteristic data associated with the building, building operation and activities data associated with the building, and weather data are used to generate a variable based degree model. A base load factor, a heating coefficient and a cooling coefficient associated with the building and an error term are determined from the variable based degree model and used to generate a plurality of multivariate regression models. A time series model is generated for the error term to model seasonal factors which reflect monthly dependence on energy use and an auto-regressive integrated moving average model (ARIMA) which reflects temporal dependent patterns of the energy use.

The invention discloses a power system abnormal data identifying and correcting method based on time series analysis. The power system abnormal data identifying and correcting method includes data preprocessing, time series modeling, abnormal data identifying and abnormal data correcting. Data preprocessing includes the step of identifying and correcting missing data in data to be detected and data suddenly changing to be zero. Time series modeling comprises the steps of conducting time series analyzing on the preprocessed data to be detected and establishing a model according to the time series, and a difference autoregression moving average model is used for modeling the data to be detected. According to abnormal data identifying, the fitting residual series of the established difference autoregression moving average model is analyzed, an error confidence interval is set, and abnormal data are identified. According to abnormal data correcting, a neural network method is used for establishing a prediction model for correcting the abnormal data, the data value of the moment when the abnormal data exist is predicted, and the abnormal data are corrected. The power system abnormal data identifying and correcting method is easy to implement and high in accuracy.

Pursuant to one aspect of the invention, a prefilter module that incorporates an inverse filter is used in conjunction with an encoder. The inverse filter has an inverse frequency response of a frequency response of a filter that simulates speech having transmission path characteristics, such as telephone-channel bandwidth speech, and / or noisy speech. The inverse filter is used to compensate transmission path characteristics of an input signal. The inverse filter can be designed using several methods, such as, for example, an autoregressive model or a moving average model. Pursuant to a second aspect of the invention, a parameter preprocessor is used in conjunction with a decoder. The parameter preprocessor performs pitch rectification through use of a medium and linear filter, and updates spectral amplitudes and voicing parameter depending on the pitch rectification. The inverse filter and parameter preprocessor, used in conjunction with an encoder and decoder, respectively, improve signal processing and parameter estimation.

The invention discloses a photovoltaic generation power ultra-short term prediction method based on a time series model. The method is characterized by comprising the following steps of collecting and normalizing historic power data of a photovoltaic power station; establishing a fitting equation according to the normalized historic power data, and determining the order of a model according to the established fitting equation and residual variance, namely, the values of p and q; determining the value of A; establishing an auto-regressive moving average model. According to the photovoltaic generation power ultra-short term prediction method based on the time series model, by establishing the prediction model for the ultra-short term prediction of photovoltaic generation power through the historic power data, the fitting equation and the auto-regressive moving average model, so that the aim of short-term accurate prediction of the photovoltaic generation power can be achieved according to the exiting model and the existing data.

The invention discloses a hybrid prediction method for network traffic. The method is characterized in that a network traffic prediction method with an extreme learning machine (ELM) being compensated by autoregressive integrated moving average (ARIMA), namely, a network traffic prediction method with the ELM being compensated by a Farctal autoregressive integrated moving average model is provided through a self-similarity analysis of a network traffic sequence. The method comprises the following steps: firstly, predicting the network traffic sequence with the ELM; secondly, correcting an error sequence of network traffic prediction through an ARIMA model; and lastly, overlaying an ELM predicted value and an ARIMA model correction value to obtain a final predicted value. According to the method, prediction error data is fitted by the ARIMA model, and the predicted value of the ELM is overlaid with a residual error of ARIMA prediction to obtain the final predicted value. Residual error compensation is performed through the ARIMA model, thereby effectively increasing the prediction accuracy.

The invention discloses an expressway traffic flow forecasting method based on time series. The expressway traffic flow forecasting method includes the steps of selecting one time scale, and carrying out statistics to build the traffic flow time series Q=(x); setting the value range of the number p of autoregression items and the number q of moving average items according to the selected time scale; solving the number p of the autoregression items and the number q of the moving average items; fitting the optimal number p of the autoregression items and the optimal number q of the moving average items through the maximum likelihood estimation in cooperation with the traffic flow time series Q to obtain an optimal ARMA model, and obtaining weight parameters of historical measured values and weight parameters of historical error values; solving the traffic flow forecasting series (please see the specifications) under the different time scales. By means of the expressway traffic flow forecasting method, an obtained time series model can better meet the requirement for forecasting various kinds of flow of an expressway, and the forecasting universality is improved; operation is simple, the forecasting efficiency is improved, the forecasting speed is increased, and the engineering requirement of traffic forecasting of the expressway is met.

The invention provides a transformer fault prediction method based on monitoring data of a dissolved gas in the oil of a transformer. The method includes the following steps of: optimization of historical online data of the dissolved gas in the oil of the transformer, model identification of the optimized data, estimation of auto-regression moving average model parameters, model checking and establishment. With the transformer fault prediction method adopted, the content of the characteristic gas in the oil of the transformer at any time point in the future can be predicted, and therefore, the faults of the transformer can be judged, and maintenance measures can be put forward. Compared with the prior art, the method can improve sample quality, embody individual characteristics of the transformer and reflect a characteristic that the dissolved gas in the oil changes with time. Since data obtained through adopting the method do not change abruptly, the method can make more stable and concise physical interpretation compared with a prediction model which is established through adopting traditional machine learning. With the transformer fault prediction method adopted, the accuracy of online data prediction of the dissolved gas in the oil of the transformer can be improved, and fault prediction and maintenance measures can be more accurate and reliable; a reliable guarantee can be provided for the maintenance and use of the transformer; and the service life of the transformer can be prolonged.

The invention discloses a wind power predication value pre-evaluation method based on wind power longitudinal time probability distribution. The method specifically comprises the following steps of: (1) fitting to obtain a longitudinal time probability distribution piecewise function according to a force output value of a wind power plant and probability distribution result of the force output value at the same time each day; (2) predicating the wind power through an autoregressive integrated-moving average model (ARMA) in a time sequence model, thus acquiring a predication value; and (3) pre-evaluating the wind power predication value: determining the appearing probability of the predication value according to the longitudinal time probability distribution piecewise function, or determining the reliability of the predication value by setting different confidence levels, in order to pre-evaluate the predication value before the real value appears; and a dispatching department can reasonably judge and accept or reject the predication value according to the appearing probability of the predication value and the reliability of the predication value.

The invention relates to a coal-fired boiler system hybrid modeling method. In the method, a real-time data driving method is firstly used for establishing a local prediction model; concretely, the collected real-time process running data is taken as a sample set of data driving; based on the set, a local controlled autoregressive moving average model in the form of discrete difference equation based on a least square method is established; secondly, a local error intelligent prediction model is established by an error data driving method; concretely, based on the sample set of the error data driving, a supporting vector optimization method is adopted to establish the local error intelligent prediction model. The local error intelligent prediction model is established by that: the error performance index is given according to process requirements and the judging is then carried out. The modeling method provided by the invention can effectively reduce the error between the model and practical process parameters, compensates the shortages of traditional controllers, ensures that the control device runs under the best state, leads the process parameters of production to be strictly controlled, and effectively improves the precision of the models.

Aspects of the present invention are related to systems, methods and apparatus for automatic quality assessment of a video sequence. According to a first aspect of the present invention, a quality index may be generated by combining a spatial quality index and a temporal quality index. According to a second aspect of the present invention, a spatial quality index may be calculated using a modified exponential moving average model to pool multi-scale structural similarity indices computed from test frame—reference frame pairs. According to a third aspect of the present invention, a temporal quality index may be generated by averaging multi-scale structural similarity indices computed from difference image pairs, wherein one difference image is formed between reference frames and another difference image is formed between a reference frame and a test frame.

The invention discloses a method and a system for predicting a flow of a self-adaptive differential auto-regression moving average model. The method and the system are used for causing a model to be more fit with a data trend of a present flow. The technical scheme comprises the following steps: utilizing an ARIMA (Autoregressive Integrated Moving Average) model to forecast the flow, and alarming when a practical value is deviated from a predicted confidence interval; while alarming, starting an alternative plan to monitor a flow data, for preventing an abnormal data from entering into ARIMA model prediction; and when the ARIMA model normally runs, judging if a parameter of the ARIMA model is still suitable in real time, and if not, automatically relearning and acquiring a new model parameter by relearning, thereby promoting the accuracy for model prediction.

A multi-model dynamic soft measuring modeling method comprises the steps of establishing multiple sub models by utilizing a self-adaptive fuzzy core clustering method and a least square support vector machine; then taking a probability distribution function constructed by a proof synthesis rule as a weight factor to perform fusing on sub model output to obtain the output of multiple models; finally performing dynamic estimation on predicted errors of the multiple models by combining an autoregression moving average model.

The invention belongs to the technical field of deposit hydrogeological exploration and relates to a water inflow forecasting method based on wavelet transform and ARMA-SVM (auto-regressive moving-average model-support vector machine). The water inflow forecasting method comprises the following steps: collecting and analyzing the water inflow account data of a mine, then selecting a modeling sample and an inspection sample, performing dyadic wavelet decomposition and reconstruction on the modeling sample, extracting a high-frequency signal and a low-frequency signal in an original time sequence, utilizing the ARMA model to build a high-frequency signal model, meanwhile, utilizing the SVM to build a low-frequency signal model, synthesizing the high-frequency signal model and the low-frequency signal model to obtain a final water inflow prediction model, and finally, utilizing the inspection sample to inspect the final prediction model to realize water inflow prediction. While the low-frequency signal is fully fit, over-fitting of the high-frequency signal is avoided, the working principle is reliable, the prediction method is simple, the prediction precision is high and the prediction environment is friendly.

The invention provides a structural damage detection method based on a nonlinear output frequency response function. Damage detection is performed on an engineering structure by adopting a NARMAX:nonlinear auto- regressive moving average with exogenous input) and the nonlinear output frequency response function (NOFRF) analysis method. The structural damage detection method based on the NOFRF mainly comprises the following three steps: firstly, identifying the NARMAX model of a system by utilizing experimental data, and obtaining a nonlinear auto-regressive exogenous model (NARX model) of the system according to the obtained NARMAX model; secondly, according to the obtained NARX model, calculating the NOFRF of the system and indexes related with the NOFRF; and finally, judging whether the system is damaged or not through comparing the indexes related with the NOFRF of the system under different states. The damage detection method provided by the invention is simple in operation and convenient in calculation. Another effective path is provided for structural damage detection.

The invention discloses a method for predicting the remaining life of a lithium battery. The method comprises two stages; at the first stage being a decomposition stage, a complicated state-of-health(SOH) sequence is decomposed into a limited number of intrinsic mode functions (IMF) and a residual function by using an EMD; and at the second stage being a prediction stage, all functions after decomposition are predicted by using an auto-regressive integral moving average model (ARIMA) and then all prediction values are added to obtain an overall SOH prediction value, so that the remaining lifeof the battery is obtained. With full consideration of the local fluctuation portion of the battery SOH sequence, prediction becomes real and effective.

The invention provides a space-time estimation and prediction method for PM2.5 concentration distribution. The space-time estimation and prediction method for PM2.5 concentration distribution comprises: collecting and correcting fine-grained aerosol optical thickness (AOD), calculating a regression model of fine-grained PM2.5, and predicting fine-grained PM2.5 concentration distribution. By comparing several regression models with a machine learning model, an XGBoost model is determined as an estimation model under the framework, the minimum root mean square error (RMSE) is 32.86 [mu]g / m<3>, and the maximum R2 is 0.71. 10 times of verification and space-time comparison with a traditional time series prediction model, namely a seasonal autoregressive differential moving average (SARIMA) model, are carried out; the prediction precision of ConvLSTM is higher, the total average prediction RMSE is 14.94 [mu]g / m<3>, and the prediction precision of SARIMA is 17.41 [mu]g / m<3>. Moreover, the ConvLSTM is relatively small in fluctuation in time and relatively good in stability, and the spatial difference of prediction precision can be relatively well eliminated in space.

The invention discloses a passenger flow volume prediction method and device, and the method and device are used in the technical field of passenger flow volume prediction. In some feasible embodiment of the invention, the method comprises the steps: obtaining characteristic attributes, affecting the passenger flow volume, from multi-source data, wherein the multi-source data comprises an intelligent card swiping data, meteorological data, and motor vehicle GPS data; and predicting the future passenger flow volume based on the characteristic attributes through employing an autoregression integration moving average model and an artificial neural network. According to the technical scheme of the invention, the method and device can predict the future passenger flow volume based on the characteristic attributes through employing the autoregression integration moving average model and the artificial neural network, and can improve the prediction precision of the passenger flow volume.

The invention discloses a hydrological time series prediction method based on a wavelet neural network and a difference autoregression moving average model. The method comprises: obtaining hydrological time series data and performing normalization processing; performing discrete wavelet decomposition on the normalized hydrological time series, to obtain a scale changing series and a plurality of wavelet transforming serieses; using an ARIMA model to perform fitting prediction on the scale changing series, to obtain a prediction value series, and performing wavelet reconstruction to obtain a normalized water level prediction series; using a WNN model to perform training fitting on the wavelet transforming serieses, to obtain prediction value serieses; performing reverse normalization on a normalized water level time series, to obtain a prediction value of an original series. The invention provides a new combination prediction model for water level and flow prediction of rivers and lakesfor water conservancy and hydropower industries. Prediction precision of the model is better than that of a conventional single neural network model and existing combination prediction methods. The method has high application value for flood control and drought relief, and irrigation and power generation.

A system, method and computer program for forecasting energy price is provided that includes an adaptive hybrid forecasting engine. The adaptive hybrid forecasting engine is operable to generate an energy price forecast based on both a prediction utility and a correction utility. The prediction utility may implement a linear modeling algorithm for predicting energy price based on historical data. The linear modeling algorithm may be a multiplicative seasonal ARIMA (Autoregressive Integrated Moving Average) model, for example, which includes both a regular ARIMA and seasonal ARIMA model. The correction utility may implement an adaptive dynamic correction algorithm that is operable to adapt the energy price forecast based on current or near-current conditions. The adaptive dynamic correction algorithm may be a LL (lazy learning) algorithm.

The invention discloses a method for predicting and correcting an integrated circuit (IC) manufacturing process based on time series analysis, which belongs to the field of IC manufacturing techniques. The data related to the IC manufacturing results are separated by establishing a process model to extract the easy-to-control process parameters related to the manufacturing results; the influence values of the other parameters are arranged in time order to form a time sequence; the overall change trend of the time sequence (potential change trend of the manufacturing results) is predicted by using the time series analysis algorithm such as autoregressive integrated moving average (ARMA) model; the predicted process fluctuation is compensated by regulating the easy-to-control process parameters to stabilize the manufacturing result and thus to realize the dynamic treatment of process parameters. The invention optimizes the process control without hardware investment and increase in manufacturing cost, and greatly increases the stability of manufacturing results.

A video abstraction method based on attention expansion coding and decoding network comprises: regarding a video abstract as a sequence-to-sequence learning process; using time domain correlation between video frames is utilized to obtain a video frame feature sequence from an original video in the SumMe or the TVSum through a pre-training network; taking the video frame feature sequence as inputof an encoder network in an attention expansion coding and decoding network to obtain a semantic information sequence of the video frames, and then obtaining a score corresponding to each video framethrough a multiplicative attention decoding network; the scores of all the video frames forming an abstract sequence; obtaining a semantic information sequence of an abstract sequence through a retrospective encoder, constructing global semantic discrimination loss, introducing a moving average model, learning semantic correlation between the abstract sequence and a video frame feature sequence, obtaining a new abstract sequence retaining important information of an original video, and finally selecting a set final abstract through the new abstract sequence. The robustness of the model is enhanced.

The invention relates to a coal-fired boiler system hybrid controlling method. In the method, a real-time data driving method is firstly used for establishing a process model; concretely, the collected real-time process running data is taken as a sample set of data driving; based on the set, a local controlled autoregressive moving average model in the form of discrete difference equation based on a least square method is established; secondly, a typical response curve method is used to design the proportional integral derivative controllers of the process model; the designed proportional integral derivative controllers of the process model are then used to design the prediction proportional integral derivative controllers. The control method of the invention compensates the shortages of traditional controlling, effectively facilitates the design of the controllers, ensures the improvement of the controlling performance and complies with the given production performance indexes at the same time. The coal-fired boiler system hybrid controlling method effectively reduces the error between the ideal process parameters and practical process parameters, ensures that the control device runs under the best state and leads the process parameters to be strictly controlled.

The invention provides an error correction-based method for ultra-short term prediction of wind speeds of extreme learning machines. The method comprises the following steps of: S1, normalizing wind speed history data to obtain a normalized data set; S2, establishing an extreme learning machine model, and carrying out wind speed prediction by utilizing the extreme learning machine model and the normalized data set so as to obtain a preliminary predicted value set and an error set; S3, judging whether a sequence of the error set is stable or not, if the judging result is positive, inputting theerror set into an auto-regression moving average model to obtain a first error prediction sequence, and if the judging result is negative, inputting the error set into an auto-regression integral moving average model to obtain a second error prediction sequence; and S4, superposing the preliminary predicted value set with the first error prediction sequence or the second error prediction sequenceso as to obtain a final wind speed predicted value set. According to the error correction-based method for ultra-short term prediction of wind speeds of extreme learning machines, wind speeds are predicted through correcting errors, so that the advantage of improving the wind speed prediction precision is provided.

The invention relates to an electrical tomography image reconstruction method based on a convolutional neural network. The method comprises the following steps: solving a positive problem of electrical tomography by adopting a finite element method; Designing a convolutional neural network structure to enable the convolutional neural network structure to be suitable for an electrical tomography image reconstruction process; Determining a loss function; Updating the network parameters by adopting a small-batch gradient descent strategy, and synthesizing the parameters obtained by each round ofiteration by using a moving average model to determine a final parameter updating value; After the iteration is finished, obtaining a convolutional neural network of which the connection weight and the threshold are determined; When the image is reconstructed, taking the actually measured boundary measurement value as a trained convolutional neural network input layer neuron, wherein the output ofthe output layer neuron is the value of each pixel point in the image.

The invention discloses a channel quality indication predicting and compensating method and a channel quality indication predicting and compensating system. The method comprises the following steps of: accumulating CQI values of the same subband or full bandwidth at same intervals to form a CQI time sequence; identifying an autoregressive summarization moving average model, namely according to the CQI time sequence, identifying the process of the autoregressive summarization moving average model and determining parameters in the autoregressive summarization moving average model; estimating the parameters of the autoregressive summarization moving average model to determine the autoregressive summarization moving average model; and predicting the CQI values based on the determined autoregressive summarization moving average model. The method can effectively reduce the difference of the original CQI values and a true value at the current moment so as to save communication resources and energy and improve the system communication capacity; and the method also has the advantages of low complexity of algorithms and convenient implementation.

The invention discloses a bus-network-based system and method for predicting passenger requirements. According to the method, the unevenness, the abruptness, the periodicity and other factors are compressively taken into consideration, and predicting of the passenger requirements in a bus network is finally obtained through predicting models such as a Poisson model changing along with time, a Poisson model weighing time changing and a moving average model synthesizing autoregression and an integrating frame based on a sliding window. The demand of passengers obtained through prediction of the system and method can provide a more convenient, faster and more comfortable bus trip environment for the passengers, and therefore the bus waiting time of the passengers is shortened, and overcrowding or over loosening of buses is avoided.

The invention discloses a monthly power consumption prediction method that comprehensively considers various economic factors, adopts the X-12-ARIMA model to decompose the monthly power and various economic factors into seasons; uses stepwise regression analysis to study the economic quantities and power consumption The correlation degree and regression model of electricity consumption were used to obtain preliminary prediction results; the annual power consumption forecast was carried out by using polynomial fitting, and the existing monthly power consumption forecast results were adjusted; the autoregressive integral sliding average Seasonal forecast corrections are carried out in each month to obtain a monthly electricity forecast model with good accuracy. Using the above technical solution, after seasonal decomposition of monthly electricity and economic volume, not only can use periodicity for forecasting, but also can effectively reduce the impact of volatility on regression analysis fitting accuracy and forecasting accuracy, and obtain good forecasting results.