391 results about "Wavelet neural network" patented technology

The invention provides a fault predicting and diagnosing method suitable for a dynamic complex system. The method can be applied in the field of fault prediction and diagnosis of dynamic complex systems of spacecrafts and the like. The method comprises the following steps of: performing failure mode and effect analysis (FMEA) on the dynamic complex system to obtain a main fault mode and corresponding performance detection parameters, dividing the performance detection parameters into slowly variable data and fast variable data, pre-processing the performance detection parameters, establishingan autoregressive moving average model (ARMA) aiming at the slowly variable data to perform time sequence prediction, establishing a multi-resolution wavelet neural network aiming at the fast variable data to perform time sequence prediction, performing fault early warning on the time sequence prediction results by establishing a prediction interval model, and performing fault diagnosis by establishing a D-S (Dempster-Shafer) evidence theory-based multi-signal fusion model. The method can be used for predicting and diagnosing the faults of the dynamic complex system with high precision, and has strong universality.

The invention discloses a depth convolution wavelet neural network expression identification method based on auxiliary tasks, and solves problems that an existing feature selection operator cannot efficiently learn expression features and cannot extract more image expression information classification features. The method comprises: establishing a depth convolution wavelet neural network; establishing a face expression set and a corresponding expression sensitive area image set; inputting a face expression image to the network; training the depth convolution wavelet neural network; propagating network errors in a back direction; updating each convolution kernel and bias vector of the network; inputting an expression sensitive area image to the trained network; learning weighting proportion of an auxiliary task; obtaining network global classification labels; and according to the global labels, counting identification accuracy rate. The method gives both considerations on abstractness and detail information of expression images, enhances influence of the expression sensitive area in expression feature learning, obviously improves accuracy rate of expression identification, and can be applied in expression identification of face expression images.

The invention discloses a complex neural network channel prediction method, and mainly aims to solve the problem of channel fading caused by channel time variation in an MIMO (Multiple Input Multiple Output) system. According to the technical scheme, the complex neural network communication prediction method comprises the following steps: 1, measuring a channel by a base station to obtain a channel coefficient training sequence containing an estimation error; 2, acquiring a corresponding training sample and desired output according to the obtained channel coefficient sequence; 3, inputting a training sample to perform complex wavelet neural network training in order to obtain a final network weight; and 4, performing channel coefficient prediction through a trained complex wavelet neural network by the base station. The method is simple, convenient and feasible, has a good effect, and is suitable for lowering the influence of the channel time variation on an MIMO system channel.

The invention provides an MEMS (Micro-electromechanical Systems) gyroscope random error predication method based on a grey wavelet neural network. The predication method comprises the following steps: carrying out pretreatment on output data of an MEMS gyroscope, collecting the output data of the MEMS gyroscope, carrying out wavelet analysis on the output data, and selecting a Db4 wavelet function so as to carry out de-noising processing on the output data of the gyroscope; grouping the output data of the MEMS gyroscope after de-noising processing, and determining an input vector and a target vector; building a grey wavelet network predication model, determining the input node number of the grey wavelet network, the outputting node number and the hidden layer node number, and initializing the network; training the built network, and storing the network for predicating the trend of gyroscope random error. Compared with a traditional gyroscope random error modeling method, according to the method, a grey theory is combined with the wavelet neural network, and thus the predication accuracy of the MEMS gyroscope random error is improved, and the predication accuracy is obviously improved compared with the traditional method.

The present invention belongs to the field of tunnel and underground engineering technology. In the concrete, it relates to a shield tunnel settlement control method based on ground-detecting radar. Said control method includes the following steps: adopting ground-detecting radar equipment to make sectional and longitudinal detection of slurry injection after shield tunnel wall, determining and estimating dielectric constant of slurry injection material, making model testing detection to obtain data, extracting characteristic image and characteristic wave, utilizing wavelet neural network automatic identification method to obtain slurry injection layer distribution form, combining said slurry injection layer distribution form with ground settlement monitoring data and making analysis, then defining further construction measure for implementing settlement control according to the detected and analyzed result.

The invention discloses a multi-crack damage identification apparatus and method for a cantilever flexible beam. The apparatus comprises a flexible beam, a mechanical clamping device, a detection device, a vibration exciter, a signal generator, a power amplifier, a fiber Bragg grating demodulator and a computer. One end of the flexible beam with multiple cracks is fixed through the mechanical clamping device; a multi-channel FBG optical fiber sensor group is adhered on the flexible beam and is evenly distributed along the length direction of the flexible beam; a signal generated by the signal generator is amplified by the power amplifier and then enters the vibration exciter; the vibration exciter excites low-order mode vibration of the flexible beam; the vibration results in curvature changes at measurement points of each sensor of the beam; the sensors input the curvature changes into the fiber Bragg grating demodulator in the form of wavelength changes; a demodulated signal enters the computer for modal analysis of the flexible beam, drafting of a modal curvature curve, calculation of damage indexes and establishment of a wavelet neural network identification model; and finally, the purpose of accurate identification of the depth and the location of the cracks of the flexible beam is achieved.

The invention discloses an aviation direct-current converter online fault combined prediction method based on fractional order wavelet transformation. The method includes: (1) monitoring and collecting output voltage signals of the aviation direct-current converter in real time, calculating output voltage change rate of different moments, and using the output voltage change rate as converter performance degradation parameters; (2) conducting abnormal value rejection and missing data filling on performance degradation data by using a 3 sigma method and an interpolation method; (3) conducting fractional order wavelet transformation on the performance degradation data, the performance degradation data is decomposed into subcomponents with different scales, and determining noise components and removing the noise components by calculating a combination entropy between high-frequency components and environment data; (4) building a predication model of the high-frequency components in decomposition data by using a wavelet neural network, building a prediction model of low-frequency components by using a gray neural network, and conducting time sequence prediction; and (5) stacking predication values of the high-frequency components and the lower-frequency components to obtain a final predication value, conducting performance evaluation and fault predication on the aviation direct-current converter by combining fault threshold values. The aviation direct-current converter online fault combined prediction method removes disturbances caused by environment factor fluctuation in performance degradation data, restores real performance degradation data, simultaneously decomposes the performance degradation data into different frequency subcomponents with strong regularity, predicts the subcomponents by using a combined prediction model, enables prediction risks to be dispersed, and improves online fault prediction correctness.

The invention discloses a short-term prediction method for occupancy of an effective parking space of a parking lot. The short-term prediction method comprises the steps as follows: 1) determining a time sequence of the occupancy of the effective parking space of the parking lot; 2) carrying out wavelet decomposition to the time sequence of the occupancy of the effective parking space through a wavelet function, thus obtaining a low-frequency coefficient vector and a high-frequency coefficient vector; implementing the wavelet reconstruction to the low-frequency coefficient vector and high-frequency coefficient vector, so as to obtain the time sequence of N+1 reconstructions; 3) establishing a wavelet neural network model to the time sequence of the N+1 reconstructions for predicting, thus obtaining N+1 prediction results; and 4) accumulating the N+1 prediction results, so as to obtain the prediction results corresponding to the time sequence of the occupancy of the effective parking space. According to the short-term prediction method disclosed by the invention, a wavelet analysis-wavelet neural network combinational prediction model is raised to perform short-term prediction to the occupancy of the effective parking space of the parking lot according to the short-term variation characteristic of the occupancy of the effective parking space of the parking lot, therefore, the prediction accuracy and the stability are improved.

The invention provides an air conditioning system sensor fault diagnosis method based on a wavelet neural network. The method is characterized by comprising the steps that comprehensive analysis and diagnosis are conducted for defining an air conditioner state control, a diagnosis condition set and a decision event set; sensor fault classification and characteristic description of an air conditioner system are used for classifying fault types; sensor fault diagnosis of the air conditioning system is used for generating a fault diagnosis flow; a network structure and a primary function are determined; selection and pretreatment of training samples are conducted; and an alarming threshold is set. The method has good diagnosis effect on abrupt changing faults such as bias faults and completefaults. Drift biases can be detected as long as the drift distance exceeds the alarming threshold. The method also has good fault detection effect on precision decrease faults of sensors.

The invention relates to a method for predicting the thermal error of a machine tool spindle based on a genetic algorithm wavelet neural network, which belongs to the field of numerical control machine tool processing technology. A temperature sensor is reasonably arranged on a numerical control machine tool, and the temperature of a key temperature measuring point of the machine tool and the temperature data of a proces environment are measured by the temperature sensor, and the thermal error data of a machine tool spindle is obtained by a displacement sensor; the thermal error prediction model of machine tool spindle based on wavelet neural network is established after data processing, combining the advantages of genetic algorithm and wavelet neural network, the thermal error predictionmodel has the advantages of simple calculation, high precision, strong anti-disturbance ability and robustness, and has strong approximation ability and fast network convergence speed. The thermal error of the spindle of the CNC machine tool is effectively reduced, and the machining accuracy of the machine tool is improved.

The present invention provides an application data processing method of a spaceborne microwave radiometer. According to the application processing method, by inputting the global or regional profile data and the radiation brightness-temperature values in a matrix, adopting a wavelet neural network algorithm to train the data, and comparing the outputted atmosphere humiture profile and ground / sea surface parameters with the preset threshold indexes, the back propagation of the atmosphere humiture profile and ground / sea surface parameter errors is implemented; by correcting each layer of weights in an error gradient descent manner, the work of utilizing a data set to correct the inversion parameters locally is realized, the local correction capabilities of the equipment developers and users to the inversion software parameters are improved, and further the detection and inversion precision is improved further.

The invention discloses a method for quality inspection of active fault and diagnosis of intelligent fault of an engine, which corresponds the symptom of the fault with corresponding frequency for quality inspection of the active fault when the engine is not disassembled. The method specifically comprises the steps of: sequentially loading signals with different frequencies from low frequency to high frequency by a vibrating device to an engine system according to the principle of a relationship between the frequencies and the fault; resonating by the signals with the same frequency; determining the symptom of the fault generated by the signals at the frequency or frequency band; collecting corresponding fault information; extracting the eigenvector of the fault for the collected fault information by a wavelet analysis method as an input eigenvector of inheriting and optimizing wavelet neural network input nodes; and obtaining a network training sample to diagnose on-line measured data. A study sample of the invention has the advantages of easy acquirement, strong pertinence, high accuracy, low cost and the like. An optimized network has the characteristics of good nonlinear mapping and high convergence speed, and can effectively diagnose the fault of the engine.

The invention relates to a base station dormancy method based on flow prediction in a heterogeneous network. A conventional base station dormancy method based on fixed flow model can not adapt to the dynamic change of load flow in a base station. A modified wavelet neural network (MWNN) model dynamically predicts the load flow of the base station, and Pico Base Stations (PBSs) instead of a Macro Base Station (MBS)provide services to users during peak hours according to the prediction result. Even if the covering scope of the PMSs is smaller than that of the MBS, the covering scope of the PMSs of certain quantity ensures the service for users when the user number reaches the climax. Since the emission power required by PMSs is much less than that of the MBS, the network energy consumption is reduced, and the purpose of green communication is achieved.

The invention relates to a self-adaption wavelet neural network abnormity detection and fault diagnosis classification system and method, which can be applied to the fields, such as economic management abnormity detection, image recognition analysis, video retrieval, audio retrieval, signal abnormity detection, safety detection, and the like. The system comprises the following seven parts: an acquisition device, a transmitter device, an A / D (Analog / Digital) conversion device, a self-adaption wavelet neural network abnormity detection and fault diagnosis classification processor, a display interaction device, an abnormity alarm device and an abnormity processing device. The abnormity detection and fault diagnosis classification object of the self-adaption wavelet neural network abnormity detection and fault diagnosis classification system is acquired from samples for which a self-adaption mechanism is automatically established by the self-adaption wavelet neural network of a system to be detected, the characteristic information of a signal can be effectively extracted through wavelet transform multi-scale analysis, and a more accurate abnormity detection and fault diagnosis locating result can be obtained. The device adopting the method has the advantages of generalization, high accuracy in the application field, capability of real-time monitoring and low cost.

The invention discloses a preparation method of a ground-based observation air temperature space-time data set. The method comprises the following steps: firstly determining space-time resolution ratio and area range of a precast data set; preprocessing air temperature observation data so as to form a normative driving data set; preparing a macro geographic factor and microcosmic geographic factor grid data set of a research area so as to enable the macro geographic factor and microcosmic geographic factor grid data set to correspond to the space-time resolution ratio of the research area; constructing a wavelet neural network dynamic combined prediction model in the time dimension, performing simulation prediction on the change characteristics of air temperature data so as to form a future air temperature change time sequence set; calculating air temperature numerical values of grids in a target area one by one in the space dimension, so as to generate a spatial data set of air temperature spatial change trends; performing simulation calculation on air temperature in the time dimension and the space dimension simultaneously based on key influence factors so as to generate an air temperature future space-time change scene predication data set. The preparation method has favorable expansibility and can be popularized and applied to preparation and production of space-time data products of other observation elements of a geoscience station.

The invention discloses a time series prediction-based wavelet neural network online PID adjustment method and a system using the same. The method specifically comprises the steps of initiating the parameters, computing the control parameters and rectifying the online adjustment parameters, computing the control amount, computing or acquiring the system output and computing the prediction result; the system specifically comprises a control decision device, an online adjuster, a control executer, a controlled object, an online predictor, a control perturbation source and a prediction perturbation source; the control decision device is used for realizing the parameter initiation; the online adjuster is used for computing the control parameters and rectifying the online adjustment algorithm parameters; the control executer is used for computing the control amount; the online predictor is used for computing the prediction result; and the control decision device is also used for judging whether the algorithm is finished. According to the method, the wavelet neural network and the classic control method are combined to solve the problem of dependence on the parameter configuration work before the system operates in the control field, so that the control system has the effects of prediction, learning, online parameter optimization and self-adaptation.

The invention discloses a big data based power load prediction method. The method comprises the steps of step one, providing data information of N periods, obtaining a first power load predictive value of the (N+1) periods through a reinforcement learning load prediction data model directed at same data information and obtaining a second power load predictive value of the (N+1) periods in a data driving mode; step two, performing information fusion on the first power load predictive value and the second power load predictive value through a D-S evidence theory to obtain a final predictive result of the (N+1) periods. By the aid of the method, directed at a power load prediction system containing multiple dimensions and multiple stages of space, time, attributes and the like, a data driving theory based non-model load prediction controller and wavelet neural network based accumulative learning prediction are combined, information fusion is performed on the predictive values through the information fusion technology to obtain an optimal predictive value, and accordingly, the accuracy and the timeliness of load prediction are improved greatly.

The invention relates to an ultrashort combined predicting method for wind speed of a wind power plant. Firstly, wind speed time series of the wind power plant is acquired, and data pretreatment is carried out on the series to obtain input data of a prediction system; the input data is predicted respectively by a continuous prediction model, an ARMA (Auto Regressive Moving Average) prediction model and a wavelet-neural network prediction model, and three groups of prediction values are obtained through computation; and finally the three groups of prediction values are adopted to obtain the prediction value of the wind speed in the future 0-1 hour by a combined prediction method, and the last two groups of prediction values are adopted to obtain the prediction value of the wind speed in the future 1-4 hours by the combined prediction method. The combined prediction method is adopted, useful information of each single prediction method is fully utilized, the prediction precision of the wind speed in the future 4 hours of the wind power plant is improved, and a reference is provided for reasonably scheduling a power grid.

The invention discloses a nonlinear interference control method and a nonlinear interface control system for a permanent magnet linear synchronous motor (PMLSM). According to the control method, an online estimated compensating voltage value ud of a wavelets neural network (WNN) is added on the basis of up+uf of the conventional composite feedforward proportional differential control, and the sum of up, uf and ud is used as the control voltage U(t) of a stator of the PMLSM, namely the control voltage U(t) is equal to up+uf+ud. The WNN is a three-layer forward network, and ud is the sum of omega1psi1, omega2psi2, ..., omegajpsij, ..., and omegampsim. A learning signal of the WNN is an output value of a proportional differential controller. The control system comprises a digital signal processing controller, a power driving module connected with the stator of the PMLSM, and a rotor displacement sensor arranged on the PMLSM. The WNN is used for effectively compensating interference such as PMLSM thrust fluctuation and frictional force and errors of a fixed parameter model, and tracking accuracy can be improved by more than 2.7 times; and the control system can be implemented by universal hardware, and is convenient to popularize and use.

The invention discloses a hydrological time series prediction method based on a multiple-factor wavelet neural network model. The invention provides a multiple-factor wavelet neural network model used for predicting the hydrological time sequence. The model takes a multiple-time sequence message as input, and the multiple time sequence message not only comprises the current wavelet coefficient of a prediction target time sequence but also comprises the current wavelet coefficient of other time sequences relevant to the time sequence; mutual information between the multiple-time sequence message and the prediction target time sequence serves as a measurement for judging the relevance of the multiple-time sequence message and the prediction target time sequence; other time sequences of strong relevance are selected; and a wavelet function selection criteria based on the a coefficient of weighted correlation is further utilized to select the optimal wavelet function for the model. Compared with the prior art, the method disclosed by the invention has the advantages of higher prediction accuracy and better expandability and practical value.

The invention relates to a wavelet neural network processor based on an SOPC (System On a Programmable Chip), comprising a forward transmission module, an error feedback module and a network updating module, wherein the output end of the forward transmission module is connected to the error feedback module; the output end of the error feedback module is connected to the network updating module; the output end of the network updating module is connected to the forward transmission module; the forward transmission module comprises a forward transmission input layer function module, a forward transmission hidden layer function module and a forward transmission output layer function module; the error feedback module comprises an error feedback output layer function module and an error feedback hidden layer function module; and the network updating module comprises a network updating output layer function module and a network updating hidden layer function module. In the invention, a wavelet neural network is realized on the SOPC, and a wavelet neural network arithmetic is divided into several basic operations; the basic operations are completed by a reconfigurable cell (RC), and the wavelet neural network with different functions can be formed by adopting different RC connection modes.

The invention provides an online partial discharge detection signal recognition method of a cable. The method comprises the following steps: a partial discharge time domain waveform of a known source is acquired; a partial discharge waveform time domain waveform sample library is established; a wavelet packet is used to denoise each partial discharge time domain waveform in the sample library; a self-adaptive wavelet neural network model with a preset number of layers is constructed; according to each denoised partial discharge time domain waveform, a PSO algorithm is used to train the constructed self-adaptive wavelet neural network model; a BP algorithm is then used to train the trained self-adaptive wavelet neural network model for a second training so as to get a well-trained wavelet neural network; and the partial discharge signal of a to-be-recognized source is received, the well-trained wavelet neural network is inputted for recognition, and the source of the to-be-recognized partial discharge signal is obtained. The method of the invention is fast in recognition speed and high in recognition precision.

The invention relates to a mechanical shoulder joint position control method with dynamic friction compensation, which is realized through a global control unit and a local control unit. The global control unit is used for tracking the trajectory of a mechanical shoulder joint in a global large range, the trajectory tracking is realized through a PD (Proportional Differential) controller widely applied in the mechanical shoulder joint, and the input vector of the PD controller comprises the position error of the mechanical arm joint and the change rate of the position error; and the local control unit is used for completing dynamic friction compensation in a local small range, the dynamic friction compensation is realized through a five-layer autoregressive wavelet neutral network controller having an observation layer, and the input vector of the autoregressive wavelet neutral network controller comprises the expected position, the expected speed and the actual position of the mechanical shoulder joint. The actual speed of the mechanical shoulder joint required in the autoregressive wavelet neutral network controller can be calculated through the observation layer. The mechanical shoulder joint position control method provided by the invention can be realized by only installing one position sensor in the mechanical shoulder joint without installing a speed sensor.

The invention discloses a wavelet neural network-based distribution network single-phase short circuit line selection method, which belongs to the technical field of distribution network protection. The method comprises the following steps: 1) db4 wavelets are selected to serve as a wavelet packet basis function to decompose transient zero-sequence current in each feeder line of the distribution network, and the sampling frequency is 10KHz; 2) the modulus maximum of the transient zero-sequence current is calculated; 3) the modulus maximum obtained in the second step and the polarity of the modulus maximum are used for training the neural network; and 4) the BP neural network after the training in the third step is applied to the distribution network with single-phase short circuit, and a fault line is determined according to a different output result. The method of fault line selection by using the wavelet neural network disclosed by the invention has good reliability and practicability, a single-phase grounding fault line can be effectively eliminated, stable operation of the distribution network system is facilitated, and an important role is played in planning and applications of a distributed power supply.

The invention relates to a power transmission line icing prediction method based on a quantum particle swarm and a wavelet nerve network. The power transmission line icing prediction method based on the quantum particle swarm and a wavelet nerve network comprises the following steps of acquiring historical icing meteorological data, namely an ambient temperature, a humidity, a wind speed, a wind direction, an air pressure, a lead temperature and an icing thickness; establishing an icing thickness prediction model by means of the wavelet nerve network; performing initial parameter optimization on the model through adding a quantum particle swarm algorithm of interference factors; and inputting the historical icing data for obtaining a predicated power transmission line icing thickness. The power transmission line icing prediction method has advantages of high prediction precision, high convergence speed, etc. The power transmission line icing prediction method can effectively predicate a line icing change rule and can be applied for power transmission line icing disaster early warning and treatment.

The invention relates to a DVB_RCS satellite channel dynamic distribution method based on predicating of a wavelet neural network. In terms of the multiple-scale features of satellite network flow, firstly, a wavelet neural network flow predicating algorithm is used for predicating real-time arrived flow of user stations of the next cycle and sending the real-time arrived flow to a gateway station; secondly, the gateway station performs distribution as needed according to the real-time access data rate of all the user stations in each channel application distribution cycle. The access data rate in the next cycle predicated by all the user stations serves as weight allocation residual capacitance. The DVB_RCS satellite channel dynamic distribution method based on predicating of the wavelet neural network is characterized in that in terms of the multiple-scale features of satellite service flow such as long relevance, self similarity and the multiple fractal property, the flow of the next moment is predicated with the wavelet neural network, the predication accuracy of the flow is improved, service time slots are reasonably distributed, service time delay is effectively shortened, the service quality of the users is ensured and the channel resource utilization rate is improved.

The invention discloses a multi-step prediction method of parking based on an optimized wavelet neural network. The method comprises a step of processing actually measured effective parking data intoan effective parking time series with a time interval of 5 minutes, and performing multiple-scale decomposition and reconstruction by using a wavelet function 'db32' to, and taking the function as a hidden layer function of the wavelet neural network, a step of adjusting a weight by using a particle swarm algorithm and carrying out gradual iterative update to obtain an optimal value, and a step ofreducing a prediction time of EPWNN by using an ELM algorithm and obtaining a prediction result according to a multi-step prediction strategy. Compared with genetic algorithm optimization neural network, genetic algorithm optimization wavelet neural network, extreme learning machine optimization wavelet transform, extreme learning machine optimization wavelet neural network, particle swarm optimization neural network algorithm, particle swarm optimization wavelet neural network and other algorithms, the prediction error of an EPWNN algorithm is reduced by 89.17%, and the time needed by prediction is reduced by an average of 50.83%.

The embodiment of the invention discloses a traffic-flow forecasting method, device and system based on the wolf-pack algorithm. The traffic-flow forecasting method includes the steps that traffic-flow data is obtained; the traffic-flow data is processed through a pre-established wavelet-neural-network traffic-flow forecasting model to obtain the traffic-flow forecasting result, wherein the wavelet-neural-network traffic-flow forecasting model is trained based on the wolf-pack algorithm, and the training process of the pre-established wavelet-neural-network traffic-flow forecasting model is that an initialized wavelet-neural-network parameter is calculated according to historical data and the wolf-pack algorithm; the initialized wavelet-neural-network parameter is trained through a wavelet neural network and the historical data to obtain the wavelet-neural-network traffic-flow forecasting model. According to the traffic-flow forecasting method, device and system based on the wolf-pack algorithm in the embodiment, when traffic flow is forecasted through the wavelet-neural-network traffic-flow forecasting model trained through the initialized wavelet-neural-network parameter obtained based on the wolf-pack algorithm, the forecasting speed and the forecasting accuracy are increased to a certain degree.