244 results about "Fuzzy adaptive" patented technology

The invention provides a method and device for estimating a state of charge of a battery. The method comprises the following steps: establishing a multi-parameter fusion mathematical model of a lithium iron phosphate battery; using a fuzzy Kalman filtering algorithm based on an index input membership function to optimize and estimate the state of the charge of the battery, wherein the step of establishing the multi-parameter fusion mathematical model comprises establishing a charge-discharge multiplying power-state of charge sub module, a temperature-state of charge sub module, and a cycle index-state of charge sub module. The method for estimating the state of charge of the battery uses the fuzzy self-adaptive Kalman filtering algorithm based the index input membership function, so that the theoretical value of measurement noise in the Kalman filter is adjusted smoother in a self-adpative manner, thereby improving the matching degree of the system measurement noise, and estimating the state of the charge of the battery more accurately.

The invention discloses a constant pressure automatic grinding method based on fuzzy adaptive force control. The grinding method comprises the following steps: detecting the contact force of a grinding head and a workpiece in a real-time manner to feed back force signals; changing an analog voltage value output by a controller according to a preset control algorithm, thereby controlling the output torque of an x-axis servomotor to control the contact force. The invention further provides a grinding device for realizing the grinding method. The grinding device comprises an industrial control main engine, a workbench, motion mechanisms, one-dimensional sensing equipment, a position sensor for acquiring the position coordinate of the workpiece, grinding equipment and workpiece clamping equipment. The grinding device can be used for detecting the normal grinding force in a grinding process based on an intelligent force control technology, and feeding back detection results; through the processing of the controller, driving signals are generated for continuous adjustment of the output torque of a driver, so that the normal grinding force can be controlled in a real-time manner and the constant pressure grinding can be realized.

This invention discloses a dual-feed AC-DC-AC converter control structure for wind power generation, and the rotor exciting magnetic converter using the fuzzy PI adaptive control strategy of current inner loop. When the generator no-load networking, the current inner loop uses the conventional PI regulator, and after networking, as to three modes, sub-synchronization, synchronization, super-synchronous, it uses fuzzy adaptive PI control. The network-side converter uses SPWM control strategy which fixed switching frequency and combining electric network electromotive front-feed. In the invention, both converting processes use four-quadrant operation PWM converters based on full-controlled devices, so it can not only improve the system dynamic response, reduce wastage and impact, achieve the two-way electric transmission, but also it can achieve wind power generation network side unit power factor sinusoidal current networking power generation operating.

The invention discloses a motor non-speed sensor control method for smoothly switching composite rotating speed identification. In the method, the model reference fuzzy self-adaptive rotating speed identification of the magnetizing current reactive power is adopted for identifying the rotating speed of the motor in high speed; the slip angular speed ring opening rotating speed identification is adopted for identifying the rotating speed of the motor in low speed; the method for smoothly switching factors is adopted for smoothly transiting the two motor rotating speed identification methods, thus realizing the composite identification of the high and low rotating speed of the motor in the starting process of the motor. The invention has the beneficial effects that the method solves the contradiction that the indexes of the dynamic property and the steady-state performance in the traditional single rotating speed identification speed can not be simultaneously optimized so that the motor has the good rapidity and strong dynamic track during the starting; in the steady-state process, the overshoot is small and the speed control precision is high; the method has the complete robustness for the state resistance, can implement the vector control of the non-speed sensor in a wider speed range, thus realizing the composite identification of the high and low rotating speed of the motor in the true sense.

The invention provides a bivariate nonlocal average filtering de-noising method for an X-ray image. The method is characterized by comprising the following steps: 1) a selecting method of a fuzzy de-noising window; and 2) a bivariate fuzzy adaptive nonlocal average filtering algorithm. The method has the beneficial effects that in order to preferably remove the influence caused by the unknown quantum noise existing in an industrial X-ray scan image, the invention provides the bivariate nonlocal fuzzy adaptive non-linear average filtering de-noising method for the X-ray image, in the method, a quantum noise model which is hard to process is converted into a common white gaussian noise model, the size of a window of a filter is selected by virtue of fuzzy computation, and a relevant weight matrix enabling an error function to be minimum is searched. A particle swarm optimization filtering parameter is introduced in the method, so that the weight matrix can be locally rebuilt, the influence of the local relevancy on the sample data can be reduced, the algorithm convergence rate can be improved, and the de-noising speed and precision for the industrial X-ray scan image can be improved, so that the method is suitable for processing the X-ray scan image with an uncertain noise model.

The invention provides a fuzzy adaptive variational Bayesian unscented Kalman filter method. The method comprises the steps of estimating a one-step prediction target state as shown in the description and a covariance matrix thereof as shown in the description, iteratively estimating the variance as shown in the description of the measured noise, calculating the true value as shown in the description, the estimate value as shown in the description, the matching degree index as shown in the description and the adjustment quantity as shown in the description of a residual variance matrix at the current moment, and the adjusted measured noise variance as shown in the description, and calculating the estimated value as shown in the description of the target state and the error covariance matrix thereof. The method is capable of estimating the statistic variance capacity of the measured noise on line, and therefore, the filter error caused by unknown time variant of the noise statistical property is reduced and nonlinear filter estimation accuracy is improved. Meanwhile, the fuzzy logic method based on the innovated covariance matching technique is used for adjusting the measured noise variance estimated by the variational Bayesian method in real time, inhibiting the divergence of the filter and enhancing the robustness of the filter method.

The invention discloses an unmanned aerial vehicle attitude fuzzy adaptive predication control method based on nonlinear model and a system thereof. A Cubic-RBF-ARX nonlinear model of the system is established by means of an offline data identification method. Then a fuzzy adaptive prediction controller is designed based on the established nonlinear model. The prediction controller performs online adjustment to the weight coefficient of a target function in the prediction controller according to the real-time state in controlling the attitude of the unmanned aerial vehicle. The fuzzy adaptive prediction controller can ensure a fact that the selected target function accords with a dynamic and stable rule and trend for attitude adjustment in the attitude control process of the unmanned aerial vehicle. Compared with a common unmanned aerial vehicle prediction controller, the whole dynamic and stable process for controlling is considered in setting the parameters of the target function, thereby performing a function of improving an unmanned aerial vehicle attitude control dynamic-and-static response index, and realizing relatively high practical value and good application prospect.

The invention discloses a fuzzy PID adaptive controller design method of an aero-engine based on RBF neural network feedforward and belongs to the aero-engine numerical control system field. The method is used for improving acceleration and deceleration work states of the aero-engine and components under effects of an extraneous condition and internal control so that the engine possesses a good acceleration and deceleration characteristic. The technical scheme is characterized by based on an aero-engine steady state test data sample, using an RBF neural network to carry out off-line training so as to form a feedforward controller; and based on that, adding a fuzzy adaptive PID controller. In the invention, through carrying out acceleration and deceleration control on each state of the aero-engine, the controller possesses a good tracking effect and a good controller parameter online setting capability, control quality of the system in a large range can be further increased and a control difficulty brought by a non-linear characteristic of the engine can be effectively solved.

The invention discloses a variable air volume room temperature control method based on fuzzy PID and a prediction control algorithm. The method comprises the steps that the temperature deviation between an expected room temperature and an actual room temperature is input to a fuzzy adaptive PID controller, PID parameter incremental quantity of the opening degree of a tail end air valve is obtained through calculation, and calibrated opening degree of the tail end air valve is output; the opening degree of the tail end air valve of the fuzzy adaptive PID controller is subtracted from the opening degree of the tail end air valve of a previous moment by utilizing a time delay link to obtain the variable quantity of the opening degree of the tail end air valve; the prediction control algorithm is adopted to obtain the actual opening degree of the tail end air valve through the variable quantity of the opening degree of the tail end air valve, and the room temperature is maintained at a set value and is kept unchanged. The control method combining the fuzzy PID with the prediction control algorithm is adopted to solve the technical problems that parameter setting difficulty is large, working condition adaptive capacity is poor, and adaptive ability is weak in an air conditioning system in the prior art.

The invention relates to an online adjustment method for a PI (proportion integrate) parameter of an asynchronous motor, the method comprises the following steps of formulating a control rule table for changing values of parameters P and I according to certain principle; judging a current operation state of the motor according to an operating speed of the motor; adopting a conventional PI control when the motor operates at a stable speed; when the motor operates at an accelerating speed or a reducing speed, adopting a fuzzy PI control to carry out an online self-adjustment for the PI parameter until the motor operates in a stable state, the self-adjustment comprises that according to a distribution scope of a rotation speed deviation value e and the value of a changing rate ec of the rotation speed deviation, a domain and scale coefficients Ke and Kec of e and ec are defined to obtain a novel control rule table; setting an initial value for the PI parameter; and carrying out the online self-adjustment for the PI parameter. By the adoption of a fuzzy self-adaptive PI control method, the PI parameter is adjusted to enable the current PI parameter to always meet the requirement of current working condition, the reaction time can be accelerated by utilizing the method to adjust the PI parameter, and dynamic performance of the motor is improved.

The invention discloses a speed and current double closed-loop control system and method for a permanent magnet synchronous linear motor. The method comprises the steps of: designing a permanent magnet synchronous linear motor vector control system; designing a fuzzy adaptive sliding-mode speed controller; designing a nonlinear disturbance observation; designing a d-axis fuzzy PID controller; designing a q-axis fuzzy PID controller; obtaining a desired q-axis current reference value as an input of the q-axis fuzzy PID controller according to the designed nonlinear disturbance observer and thedesigned fuzzy adaptive sliding-mode speed controller; setting a d-axis reference current value as 0, and using the id obtained by subtracting park transform from 0 as the input of the d-axis fuzzy PID controller; after the processing by the d-axis fuzzy PID controller and the q-axis fuzzy PID controller, obtaining the ud, uq of the vector control system, and finally outputting the current operating driving voltage of the permanent magnet synchronous linear motor by the anti-park transformation and SVPWM modulation through the vector control and an inverter. The method enhances the robustnessof the system.

The invention discloses a hydro governor parameter optimization method based on the fuzzy self-adaptive DFPSO algorithm. The method comprises the steps of (1), establishing a hydroturbine regulating system mathematical model; (2), determining a fitness function of the fuzzy self-adaptive DFPSO algorithm; (3), implementing fuzziness setting on the speed inertia factor of the algorithm, wherein the linearly-decreasing inertia factor and the current particle optimal performance evaluation index are input fuzzily; (4), calculating the fitness values of particles, maintaining the individual optimal values and the global optimal values of the particles and updating the speeds and positions of the particles; (5), carrying out gene crossover between the particles when the crossover condition is met; (6), judging whether the end condition is met, stopping gene crossover and outputting the optimal value when the end condition is met, and implementing the steps (4) to (6) when the end condition is not met. According to the invention, a system can have good dynamic property on the conditions of frequency disturbance and loading disturbance, small overshoot, short stabilization time and short adjustment time.

The control system includes computer, potentiometric amplifier, A / D converter, D / A converter, filter, position finding system, learning device, optimizing decision maker. Position finding system is setup in ship. Information output of the position finding system is connected to input end of A / D converter. Output end of the A / D converter is connected to computer, which is connected to an input end of filter. Through the learning device and the optimizing decision maker, the output end of the filter is connected to the input end of the potentiometric amplifier. The output end of the potentiometric amplifier is connected to actuator. The invention can adapt to loading change of ship and change of sea condition so as to raise dynamic-positioning precision and lower positioning energy consumption.

The invention provides a multi-UUV (Unmanned Underwater Vehicle) cooperative system underwater target tracking algorithm for a fuzzy adaptive interacting multiple model (FAIMM). Firstly, according to the bearing-only target tracking principle of the multi-UUV cooperative system, a discrete nonlinear state and observation equation for the target tracking system is built; then, according to the characteristics of underwater target motion, in combination of five kinds of target motion modes, analysis is carried out according to the dynamic state transition matrix, the coupled inequality relation among the five modes is put forward, and a motion mode set adapted to underwater target tracking is selected optimally; then, an intermediate Gauss distribution function is adopted as a membership function, a mode probability is used as an evaluation index for filter information and corresponding covariance acquired by each mode, and fuzzy reasoning for the mode transition probability is designed; and finally, the FAIMM algorithm is designed and realized. During the multi-UUV cooperative system bearing-only target tracking process, the least number of target motion sets is selected, adaptive change of the mode transition probability is realized through the fuzzy reasoning, disordered competition among the modes are reduced, the filter accuracy is higher, and demands of multi-UUV cooperative system underwater target tracking can be met.

The invention relates to a control system of an opening-closing die motor and a push-out motor of a fully-electric injection molding machine, comprising a computer controller, a PLC (Programmable Logic Controller), an opening-closing die push-out driver, the opening-closing die motor and the push-out motor which are mutually connected. The invention also analyzes the motion characteristics and the dynamic characteristics of a toggle lever mechanism of a five-hinge die-closing double-toggle die closing mechanism object, provides a die protection implementation method of a die closing mechanismunder the driving of a fully-electric motor as well as a die closing mechanism control method and a field oriented control algorithm on the basis of a fuzzy adaptive theory from the angle of a control theory, meets the control requirements on three aspects of the die closing force, the speed, and the position, ensures the reliable closing of the molded die and realizes the opening-closing motion of the die and accurate and stably push-out of products from the die; and the designed DSP (Data Signal Processing) driver ensures that the system has stable reliability under the condition of high load and simultaneously improves the disturbance compensation.

The present invention discloses an electric vehicle asynchronous motor fuzzy adaptive dynamic surface control method with the consideration of an iron loss. According to the control method, by using the nonlinear function in a fuzzy logic system approximation system, an adaptive backstepping method is used to design a controller, and the dynamic surface control technology and a fuzzy adaptive method are combined. Through introducing a first-order low-pass filter, a 'computing explosion' problem caused by continuous derivation in traditional backstepping control is successfully overcome. In addition, control signals uqs and uds only exist in an adaptive parameter Theta<^>, and the amount of computation is reduced. Through fuzzy adaptive dynamic surface control method adjustment, motor operation can rapidly reaches a stable state and is more suitable for a control object with the need of rapid dynamic response like an electric vehicle driving system, a simulation result shows that the influence of parameter uncertainty is overcome and an ideal control effect is ensured according to the control method, and the rapid and stable response of rotation speed is realized.

The invention relates to a micro fuel gas turbine technology, whose structure includes turbine burning controller and control method. The controller includes PUT, detecting unit, output unit, photo-electricity coupling unit, and driving unit. The control method is realized by computer program module that includes system initialization, self-testing, burner firing, power closed loop adjusting block, stop and turbine capability index real time monitoring module. It adopts neural network PID controller adjusting method. It improves control accuracy, decreases technology cost and increases the stability and safety of the system.

An improved Smith predicting controlling algorithm comprises the following steps: designing a fuzzy adaptive PID controller to replace an original PID controller, wherein the fuzzy adaptive PID controller mainly comprises a PID control regulator and a fuzzy inference engine, and modifying a PID control parameter according to a fuzzy control rule in real time by using an error e and the change rate ec of the error as the input of the controller and the kp, ki and kd of the PID control parameter as the output, so as to meet the requirements of e and ec for the self-regulation of the PID parameter at different moments; adding an inner feedback loop, sending a feedback signal to a predicting model and an input end of an actually-controller object by the inner feedback loop at the same time, comparing the feedback signal with the output of a main controller, using the difference as an input signal of the actually-controlled object and the predicting model and introducing an adaptive regulator. Therefore, the problem of deviation caused by the change of an open-loop gain parameter is solved by changing the value of a proportional controller continuously.

The invention relates to a method for controlling a rotary speed of a magnetically suspended flywheel. The method comprises the following steps of: according to a feedback value of the rotary speed of the magnetically suspended flywheel, performing fuzzy adaptive kalman filtration of the feedback value of the rotary speed, performing rotary speed loop control of the magnetically suspended flywheel according to the rotary speed reference value and the filtered rotary speed value, adaptively adjusting the rotary speed loop control parameters according to the rotary speed reference value and the rotary speed deviation, and outputting the current reference value by the rotary speed loop; performing the current loop control according to the current reference value and the current feedback value; and finally, by adjusting the current value of a motor winding, realizing the high-precision control of the rotary speed of the magnetically suspended flywheel in the full-rotary speed range. The method belongs to the technical field of aerospace control, and can also be applied to the high-precision control of other dc brushless motors.

The invention discloses a high-voltage parallel hybrid active power filter device and an iterative learning control method with a forgetting factor. The device is composed of a passive power filter, an active power filter and a coupling transformer. A steplength-variable fuzzy self-adapting harmonic wave detection method is adopted, so as to coordinate the requirements of steady-state performance and convergence rate of the system, and the detected electric network harmonic current is used as the system control target. An iterative learning control algorithm with the forgetting factor is adopted, and an improved Ziegler-Nichols method is utilized to realize the optimization of control parameters and improve the system control performance. The high-voltage parallel hybrid active power filter device has certain advantages in response time and control accuracy, can satisfy the requirements of large-capacity hybrid active filtering system, not only can well control the harmonic wave in the electric network, but also can compensate large-capacity reactive power, and is reliable in systems operation.

The invention relates to an infrared target detection method based on a space-time cooperation framework. The method comprises the following steps: 1. acquiring a background frame Bg and a current frame Ft of a video, combining the background frame Bg and the current frame Ft to carry out background clutter suppression and acquiring a background suppression graph Gt after the background clutter suppression is performed; 2. for the background suppression graph Gt obtained in the step 1, firstly establishing a space-time background model, and then carrying out target positioning aiming at space-time background model information after the model is established; 3. according to an imaging mechanism of the infrared target, analyzing a space difference of the infrared target and the surrounding background, using a fuzzy adaptive resonance nerve network to carry out local classification aiming at the target which is positioned in the step 2 and then extracting the infrared target. The method has the following advantages that: the method does not depend on any target shapes and motion information priori knowledge; the method is suitable for a complex outdoor scene; a signal to noise ratio can be increased; a target detection rate can be increased and a calculated amount can be reduced; false targets can be effectively removed and a false alarm rate can be reduced; the method is beneficial to follow-up target identification.

Disclosed is a fuzzy adaptive pictorial information digitizer (PID) control method of a motor train unit based on multi-model switch. According to the method, a data driving modeling method is proposed according to collected running process data of the motor train unit and field experience knowledge, the number of braking models is determined by means of subtraction cluster, and then a plurality of partial linear models describing a braking process of the motor train unit are established. The partial linear models adopt the recursive least-squares method to identify model parameters. At each sampling moment, an optimal partial model is selected based on the multi-model switch strategy, and the fuzzy adaptive PID control algorithm is adopted to control the braking process of the motor train unit, and therefore safe, punctual and effective running of the motor train unit is achieved. The multi-model adaptive PID control of the braking process of the motor train unit is simple, practical, capable of achieve multi-purpose control of the braking process of the motor train unit, and suitable for on-line monitoring and automatic control of the braking process of the motor training unit.

The invention discloses a circulating fluidized bedboiler combustion optimizing control method based on fuzzy adaptive inference, and belongs to the technical field of circulating fluidized bed combustion. A circulating fluidized bedboiler combustion optimizing control system comprises a data communication subsystem, a model prediction subsystem and a performance optimization subsystem, wherein the data communication subsystem is in data interaction with OPC server communication software of a DCS; the model prediction subsystem is connected with the data communication subsystem; the performance optimization subsystem is respectively connected with the model prediction subsystem and the data communication subsystem; a boiler efficiency and SO2 and NOx discharge model is established through fuzzy adaptive inference algorithms, the running working condition of a circulating fluidized bedboiler is optimized by selecting a fruit fly optimization algorithm with an optimum reserved strategy, optimal set values of operating variables are provided for a power station DCS basic control layer, and high-efficiency and low-pollutant discharge running of the circulating fluidized bedboiler is realized.

The invention discloses an unmanned ship heading and speed cooperative control method based on a fuzzy adaptive algorithm. In order to solve the problems that in a conventional control method, a motion model is hard to accurately determine, the dynamic response is not ideal, and the anti-jamming capability is low, the fuzzy control algorithm is adopted for improving the fitness of a USV control method under undetermined water conditions, and the ship motion modeling problem in the conventional method is solved. The two-input and two-output improved USV fuzzy control algorithm is put forwards, a range ambiguity domain function is added to serve as the algorithm input, the heading and the speed are brought into the autonomous control range to conduct cooperative control, the convergence rate of USV motion control is increased, and the control robustness is improved; the course angle deviation ratio is adopted as the measurement of the operation environment uncertainty and the running state deviation degree, interval parameters are adjusted and controlled automatically, the control algorithm can adjust control parameters in a self-adaptive mode according to the control environment change, , and the intelligence of an autonomous control system is further improved.

The invention provides a multivariate time series abnormal mode prediction method and a data acquisition monitoring device. The method comprises the steps of obtaining an optimal k value of an MMOD algorithm based on historical data according to a natural neighbor principle; carrying out online expansion on the MMOD algorithm to achieve online identification of a multivariate time sequence abnormal mode; and according to an incremental fuzzy adaptive clustering algorithm, achieving conversion from the multivariate time series sub-sequence to the observation sequence, constructing a hidden Markov model based on a Baum-Welch algorithm and all the observation sequences, and achieving online prediction of the multivariate time sequence abnormal mode based on the constructed hidden Markov model. Through the multivariate time series data acquisition system of the cloud platform, related data needing to be mined can be better acquired, and real-time prediction of the abnormal mode of the multivariate time series can be achieved by utilizing an online density difference anomaly detection algorithm and a Markov prediction model algorithm. A monitoring system APP is constructed, so that real-time monitoring is facilitated.

The invention provides a bearing-only underwater target tracking algorithm based on multilevel information fusion. Three levels are divided. In the first level, a least square method is used for initially tracking and positioning a target, and the estimated state value is calculated as an input value of a second-level AFIMM (Adaptive Fuzzy Interacting Multiple Model). In the second level, the AFIMM is designed, the motion model set of the AFIMM is designed optimally, and a method of a fuzzy adaptive model transition probability is brought forward. In the third level, resampling particle filter is designed as an iterative filtering structure at each step for the AFIMM, and information fusion is realized. During the bearing-only target tracking process for a multi-UUV (underwater unmanned vehicle) cooperative system, through the design of the multilevel information fusion framework, the filter accuracy is higher, and the target of multi-UUV cooperative underwater target tracking can be met.

The invention belongs to the technical field of fault diagnosis and forecasting in the chemical industry production process and relates to a dynamic process fault forecasting method based on fuzzy self-adaptive prediction. Fault parameters and state variables are sequentially determined based on a fuzzy self-adaptive unscented Kalman prediction method, and whether future multi-step prediction is needed or not is determined according to fault parameter standard deviations; then, a future multi-step prediction value is worked out; finally, the fault prediction probability is worked out through sigma points of output variables, and fault prediction is carried out according to the fault prediction probability. The whole technical process is simple, principles are reliable, calculating parameters are accurate, fault state estimation precision is high, the application range is wide, logicality is strong, and the method is environmentally friendly.

The invention discloses a direct torque control method for a brushless direct current motor. By speed and torque double closed loop control, a rotating speed loop is used as an outer loop, and a torque loop is used as an inner loop; motor speed is calculated according to an output signal of a Hall sensor of the brushless direct current motor, and the motor speed is compared with reference speed to obtain speed error; an output value obtained after fuzzy self-adaptation adjustment is used as a torque expectation value; torque error is obtained by subtracting actual torque from expectation torque, wherein the actual torque is calculated by multiplying back-EMF (Electromotive Force) and phase current; the torque error is inputted into a hysteresis comparator; the hysteresis comparator outputs a position signal of a rotor; and an appropriate voltage vector is selected from a formulated switch list to control an inverter bridge connected with a DC power supply so as to make the brushless direct current motor output stable torque. By the method, the defect that current hysteresis control has a poor effect of inhibiting electromagnetic torque is avoided, and the problem that a traditional PID control method has low precision and poor anti-jamming capability in control of BLDCM is overcome.

The invention discloses an electric car permanent magnet synchronous motor command filtering fuzzy control method taking iron loss into account. According to the fuzzy control method, a command filtering technology is introduced in a traditional back-stepping design method, error caused by filtering is reduced by introduction of compensation signals, and a problem of 'calculation explosion' caused by continuous derivation in the traditional back-stepping control is overcome successfully. In the control method provided by the invention, a fuzzy logic system is used for approaching a nonlinear function in the system, a command filtering back-stepping technology and a fuzzy self-adaptive method are combined together to form a fuzzy self-adaptive speed controller; after being regulated by the control method provided by the invention, a motor can quickly run in a steady state, and the control method is more suitable for a control object needing fast dynamic response, such as an electric car drive system; simulation result shows that, with the control method provided by the invention, influence caused by inaccuracy of parameters can be overcome, ideal control effect is guaranteed beneficially, and fast and steady response to revolving speed is realized.

The invention discloses a transformer winding state recognition method based on a fuzzy adaptive resonance neural network. The transformer winding state recognition method comprises the steps of (1) arranging monitoring points of a vibration sensor, setting the sampling frequency and the sampling time of a data acquisition instrument, and acquiring transformer vibration signals; (2) performing wavelet packet decomposition and reconstruction on the transformer vibration signals; (3) extracting vibration signal sub-band energy values; (4) analyzing the vibration signal sub-band energy values of each vibration monitoring point, and selecting feature band energy of valid monitoring points to construct a feature vector so as to act as input of the fuzzy adaptive resonance neural network; (5) building the fuzzy adaptive resonance neural network, adjusting network parameters until set precision is reached; and (6) performing recognition on transformer winding state through the fuzzy adaptive resonance neural network. The transformer winding state recognition method can judge a transformer winding pressing state accurately and quickly, and can be applied to online monitoring and recognition for the transformer winding pressing state.