17 results about "Narx neural network" patented technology

The invention discloses a cable production equipment fault diagnosis method based on machine learning. The method comprises the steps that multi-source time sequence data are collected and preprocessed to generate a standardized data set; executing fractional calculus operation to obtain a fractional response sequence and dynamic memory weight distribution; constructing long-term memory features and forming an exogenous input vector sequence; performing joint training on exogenous input and real observation to obtain a convergent improved NARX neural network model; inputting the exogenous input and historical output lagging sequence into the model, outputting a state prediction value and mapping the state prediction value into a fault risk index; generating a fractional order residual energy index and an abnormal score sequence; and judging and outputting the fault state and the degradation trend grade of the cable production equipment. According to the invention, by introducing the fractional calculus algorithm and improving the NARX neural network model, high-precision fault identification and degradation trend intelligent prediction of cable production equipment under complex working conditions are realized.

The invention relates to an underwater electromagnetic detection noise suppression method based on an NARX neural network, and the method comprises the steps: collecting first magnetic field data through a first vector magnetic sensor, and collecting second magnetic field data through a second vector magnetic sensor; training a sea wave noise prediction model based on the first magnetic field data and the second magnetic field data, and repeatedly and synchronously intercepting the first magnetic field data and the second magnetic field data in a preset time range through a sliding window, inputting the intercepted first magnetic field data and the intercepted second magnetic field data into the sea wave noise prediction model for prediction to obtain a corresponding sea wave noise prediction value; and performing differential processing on the first magnetic field data in each sliding window and the sea wave noise prediction value in the corresponding preset time range, and splicing differential results to obtain secondary field information in the first magnetic field data. According to the invention, the accuracy of noise estimation is improved, and the background interference in the signal can be better stripped.

The application provides a method and system for predicting main steam parameters before steam warming of a combined cycle unit, the method comprising the following steps: obtaining historical operation data of a target power generating unit, the historical operation data comprising a plurality of observable input parameters and output parameters closely related to a main steam system; constructing an NARX neural network model, training the neural network model based on the historical operation data, and obtaining a trained main steam parameter prediction model; deploying the main steam parameter prediction model on a general computing device; obtaining initial state parameters under a current working condition before starting the power generating unit, and inputting the initial state parameters into the main steam parameter prediction model; and outputting a predicted trajectory of main steam key parameters before the high-pressure bypass system is put into operation from the main steam parameter prediction model.

The invention discloses a neural network adaptive feedforward method and system for realizing cable constant tension control, and relates to the field of industrial control, and the method comprises the steps: S1, obtaining the latest historical tension data and real-time tension data of a cable; s2, a pre-trained NARX neural network model outputs a cable tension prediction value at a future moment; s3, taking the cable tension prediction value as a feedforward compensation amount; s4, inputting the system control error into a BP neural network to obtain a self-adaptively set PID control parameter; s5, calculating to obtain a feedback control quantity by adopting the self-adaptively set PID control parameters; s6, superposing the feed-forward compensation quantity and the feedback control quantity; according to the method, future tension changes are accurately predicted through the NARX neural network and serve as feed-forward signals, advanced suppression of disturbance is achieved, PID parameters are adjusted online through the BP neural network, and a controller can automatically adapt to changes of complex working conditions such as different sea conditions and loads.

The application discloses a kind of whole vehicle fuel sulfur content calculation methods, belong to fuel sulfur content calculation technical field.The whole vehicle fuel sulfur content calculation method includes the following steps: S1: information collection;S2: training NARX neural network model;S3: report whether SCR catalyst is sulfur poisoning failure;S4: repair to the poisoning SCR catalyst;If driver uses low sulfur content oil, time is very long, possibly all does not cause catalyst poisoning, so adopt the product of sulfur content and time, to express the degree of poisoning comprehensively, using NARX neural network to calculate: the poisoning degree of engine next time, is influenced by its last time poisoning degree, the feature of NARX neural network is to use the output value of last time, as the input of next time participates in calculation, improves the accuracy of whether catalyst is poisoned detection.

An intelligent automobile trajectory tracking control method based on adaptive switching MPC belongs to the technical field of intelligent automobile control, and comprises the following steps: establishing a vehicle mechanism dynamics model; establishing a vehicle dynamics model; designing an MPC controller based on the three types of models; the road curvature, the road adhesion coefficient and the longitudinal vehicle speed serve as fuzzy input, the transverse error serves as a variable universe scaling factor, and self-adaptive switching of the three types of models is achieved through variable universe fuzzy control. According to the method, the modeling precision under the complex working condition is improved through the NARX neural network, the tracking precision and the calculation real-time performance are balanced through the self-adaptive switching mechanism, simulation verification shows that the minimum transverse error peak value can reach 0.102 m, the control time is shortened to 4.56 s, and the requirements of an automatic driving system for high precision and high real-time performance of trajectory tracking are met.

The invention provides a dynamic identification method and system for a GFM inverter dominant power distribution system and a related device. The method comprises the following steps: step 1, establishing a phasor domain mathematical model of a GFM inverter in a phasor domain; step 2, constructing a nonlinear least square optimization model based on the phasor domain mathematical model, obtaining a parameter estimation value corresponding to the GFM inverter, substituting the parameter estimation value into the phasor domain mathematical model, and solving to obtain an optimized output phasor; step 3, optimizing a system dynamic prediction model pre-constructed based on the NARX neural network by using the optimized output phasor obtained in the step 2 to obtain an optimized system dynamic prediction model; and step 4, based on the pre-constructed equivalent model of the GFM inverter in the phasor domain, utilizing the optimized system dynamic prediction model obtained in the step 3 to dynamically predict the dominant power distribution system of the GFM inverter, and through organic combination of physical optimization and data driving, high-precision and high-generalization system dynamic representation is realized.

The embodiment of the application discloses a voice coil motor hysteresis modeling method and a related product, wherein the steps of the method comprise: determining a transfer function relationship between an output displacement and an input current, determining a phase frequency response and an amplitude frequency response relationship according to the transfer function relationship between the output displacement and the input current; and based on the phase frequency response and the amplitude frequency response relationship, fitting a voice coil motor hysteresis trajectory by using a neural network to obtain a voice coil motor hysteresis model. The scheme disclosed in the application can effectively describe the hysteresis phenomenon exhibited by the voice coil motor when the output millimeter-level displacement is output, and the description effect of the model basically remains unchanged with the change of the frequency, and the scheme has a more stable dynamic trajectory tracking effect than a pure NARX neural network model.

The invention discloses an ocean platform mooring cable tension real-time prediction method and application based on a neural network, and the method comprises the following steps: simulating the sea condition through numerical simulation, and generating a high-frequency time sequence data pair of the global displacement and tension of the top end of a mooring cable in an off-line manner, and taking the high-frequency time sequence data pair as the training basis of a model; global displacement data is converted to a local coordinate system with the top end of each cable chain as an original point to achieve motion normalization, and a feature data set which can be directly used for neural network training is constructed through downsampling and standardization processing; constructing a double-hidden-layer NARX neural network taking historical local displacement as input and current tension as output, and performing training and hyper-parameter optimization on the double-hidden-layer NARX neural network by using the generated normalized data set to obtain a prediction model; integrating the trained model into an anchor chain digital twin system, deploying the anchor chain digital twin system to an ocean platform with a motion sensor, predicting tension in real time, and performing early warning and analyzing the fatigue strength of the anchor chain according to a result.

The application discloses an improved NARX ship roll motion multi-step prediction method, which comprises the following steps: step 1, obtaining a plurality of roll angle time series data as training samples; step 2, determining a fixed step length of a multi-step prediction model; step 3, constructing an improved NARX neural network multi-step prediction model; step 4, training the improved NARX neural network multi-step prediction model through the training samples and the fixed step length, and obtaining a trained improved NARX neural network multi-step prediction model; and step 5, predicting the actually collected roll angle time through the trained improved NARX neural network multi-step prediction model, and taking any prediction value in the obtained fixed step number as a prediction result. The application can obtain better prediction effect in the case that the sensor sampling period is long and the system time lag is compensated, and can ensure that the ship equipment can be stably and safely operated.

The invention aims to provide a common rail fuel injection quantity estimation method based on a physical constraint autoregressive network, and belongs to the field of internal combustion engine fuel injection control. Physical parameter neural network training; pre-training the NARX neural network; oil injection stage judgment and physical output calculation; carrying out weighted fusion output; and carrying out integral training on the PI-NARX network. According to the method, time delay preprocessing is innovatively carried out on the training data labels, the time alignment problem of the system is improved, and the real-time estimation accuracy is improved. And a learnable fusion weight is introduced to realize adaptive optimal estimation. The self-adaptive weighted fusion mechanism ensures that the system can always give the most credible estimation result at present.

This invention relates to the field of unmanned aerial vehicle (UAV) swarm control technology, and more particularly to a method, system, medium, and software product for simulating and modeling phase trajectory of a UAV network system. The method includes system and task analysis, construction of interaction rule matrices, state analysis and phase vector space construction, approximate calibration of the stability domain, determination of the ultimate impulse, simulation of disturbance impulse injection, and data segmentation and transition zone marking, realizing a closed-loop process from modeling to disturbance assessment. Trajectory prediction is performed using a NARX neural network, combined with Gaussian white noise augmentation and multi-round simulation to achieve approximate calibration of the stability domain and determination of the ultimate impulse, and automatically outputs a structured trajectory dataset. This invention improves modeling accuracy, simulation efficiency, and system security.

The human lower limb motion intention recognition and exoskeleton robot angle prediction control method comprises the following steps: 1) collecting human surface electromyogram signals, and converting joint angle values into joint signals for recording; 2) pre-processing the collected signals; 3) carrying out noise reduction filtering on the pre-processed signals; 4) intercepting the noise-reduced electromyogram signals to obtain the action segments in the electromyogram signals, and extracting the characteristic values of the action segments; 5) converting the action segments into frequency domain signals, carrying out frequency domain analysis on the action segments, and obtaining the median frequency of the action segments; 6) taking the extracted characteristic values as the index for measuring the accuracy, constructing a regression model by using a BP neural network, and recognizing the lower limb motion intention through the regression model; and 7) constructing an NARX neural network to predict the joint angle. The present application enables the exoskeleton robot to simultaneously realize the estimation and prediction of the human lower limb motion intention and the lower limb joint angle.

The invention provides a carbon emission factor intelligent prediction method and system based on an NARX neural network, and belongs to the field of power system carbon emission prediction. According to the method, an NARX neural network model with historical node carbon emission factors as endogenous autoregression input and historical node load power as exogenous input is adopted, and prediction is carried out in combination with power grid real-time operation data and recent historical data, so that time sequence characteristics of power system data and internal relations among the factors can be effectively mined; the defects that a traditional physical model is difficult to track real-time changes of electric carbon factors and cannot fully consider the influence of market factors and an existing neural network model is weak in generalization ability and insufficient in utilization of time sequence information are overcome, and meanwhile excessive dependence on complete power grid operation parameters is reduced. Therefore, the accuracy, timeliness and adaptability of power grid node electric carbon factor prediction are improved.

The application discloses a kind of based on data driving power battery fault early warning and safety risk assessment method, comprising: data acquisition;The data collected are handled;The data after processing are divided, and the correlation analysis of power battery parameter is carried out, and data set is made;Data set is divided into training set and test set, so that NARX neural network structure model is trained on training set, so that NARX neural network structure model is tested on test set, according to the performance of test set, the hyperparameter adjustment of NARX neural network structure model is carried out, and voltage fault early warning model is established;Temperature fault early warning model is established according to probe temperature;According to voltage fault early warning model and temperature fault early warning model, establish power battery safety risk assessment system.The power battery fault early warning and safety risk assessment method disclosed in the application has the advantages of small amount of calculation, fast calculation speed, early fault early warning, can be more accurately early warning and evaluation to power battery and the like.

The invention discloses a self-adaptive control system and method for plasma ozone products in a complex environment, and belongs to the technical field of ozone generator control. According to the method, multi-dimensional time sequence data of ozone concentration, temperature, humidity and pulse excitation frequency in changing environmental conditions are collected and input into an NARX neural network model with a first-order lagging output characteristic, and the dynamic change trend of the ozone concentration is predicted. And embedding the prediction model into a model prediction control framework, solving an optimal control sequence in real time in a rolling prediction time domain by using a gradient descent algorithm, and adaptively adjusting the excitation frequency. According to the method, observable disturbance is incorporated into a prediction model, the method has the capability of quickly responding to multivariable coupling disturbance, overshoot and steady-state errors caused by environment temperature and humidity changes in ozone concentration control can be inhibited, and the robustness and precision of a control system are remarkably improved. The method is suitable for scenes requiring high-precision ozone control under variable working conditions, and can be popularized to the field of intelligent control of other plasma products.

The present application relates to nuclear power plant fault diagnosis prediction technical field, especially a kind of long-period vibration prediction and regulating system based on NARX modeling, the system includes long-period vibration instability diagnosis model, long-period vibration instability diagnosis model is used to identify the long-period vibration of vibration displacement, determine whether long-period vibration occurs in steam turbine;Vibration displacement prediction NARX neural network model extracts steam turbine bearing pad vibration displacement and operating parameter, and vibration displacement NARX model is obtained by training;Vibration displacement NARX neural network adjustment prediction model is used to determine the best oil temperature and safe oil temperature range of steam turbine not long-period vibration;Oil temperature regulation hysteresis NARX neural network model is driven by the difference between the best oil temperature and current oil temperature, and the output is adjusted oil cooling water flow value.The system realizes accurate and rapid regulation of oil temperature through oil temperature regulation hysteresis NARX neural network model, and makes up for the shortcomings of poor control accuracy caused by hysteresis.