Long-period vibration prediction and adjustment system based on NARX modeling

Abstract

The invention relates to the technical field of nuclear power station fault diagnosis and prediction, in particular to a long-period vibration prediction and regulation system based on NARX modeling, which comprises a long-period vibration instability diagnosis model, and is characterized in that the long-period vibration instability diagnosis model is used for identifying long-period vibration of vibration displacement and judging whether a steam turbine generates long-period vibration or not; the vibration displacement prediction NARX neural network model extracts the vibration displacement and operation parameters of the steam turbine bearing bush, and a vibration displacement NARX model is obtained through training; the vibration displacement NARX neural network adjustment prediction model is used for determining the optimal oil temperature range and the safe oil temperature range when the steam turbine does not generate long-period vibration; and the oil temperature regulation hysteresis NARX neural network model is driven by the optimal oil temperature and the current oil temperature difference to output a regulation oil cooling water flow value. According to the system, the oil temperature is accurately and rapidly adjusted through the oil temperature adjustment delay NARX neural network model, and the defect of poor control precision caused by delay is overcome.

Description

Technical Field

[0001] The invention relates to the technical field of nuclear power plant fault diagnosis and prediction, in particular to a long-period vibration prediction and regulation system based on NARX modeling. Background Art

[0002] Nuclear power plants include many rotating equipment, and vibration performance is one of the key performance indicators of these equipment. The steam turbine unit and its steam supply system are the key systems responsible for converting thermal energy into mechanical energy in the power production process of nuclear power plants. The steam turbine is one of the main equipment in the conventional island of a nuclear power plant. The steam turbine unit is composed of sophisticated and complex high-speed rotating moving parts. During operation, the steam turbine unit is prone to abnormal shaft vibration and other problems, which affects the unit's output and service life. Vibration failures in steam turbine units not only directly affect the production of nuclear power, but are also closely related to the safe and efficient operation of nuclear power plant units. Therefore, steam turbine units are important maintenance objects in the daily operation and maintenance of nuclear power plants.

[0003] In the prior art, the NARX (nonlinear autoregressive external input) modeling of rotating machinery faults uses the harmonic product spectrum to determine the real-time rotation frequency of the rotor radial vibration signal, and uses the forward orthogonal least squares method to identify a customized NARX model. This method is based on the direct vibration signal of the rotor or related equipment components for data processing and NARX model training. This method can only determine whether a certain fault has occurred in the rotor, but cannot determine the cause of the fault, nor can it quickly adjust to suppress the occurrence of vibration. Therefore, this method cannot completely solve the technical problems of intelligent prediction and efficient adjustment and suppression of long-period vibration. Summary of the invention

[0004] The present invention provides a long-period vibration prediction and regulation system based on NARX modeling, which is used to solve the problem in the prior art that the customized NARX model identified by forward orthogonal least squares method cannot realize the prediction of long-period vibration displacement of the steam turbine and the identification of long-period vibration.

[0005] The technical solution of the present invention is as follows:

[0006] The present invention discloses a long-period vibration prediction and regulation system based on NARX modeling, the system comprises a long-period vibration instability diagnosis model, a vibration displacement prediction NARX neural network model, a vibration displacement NARX neural network regulation prediction model and an oil temperature regulation hysteresis NARX neural network model, the long-period vibration instability diagnosis model identifies the long-period vibration of the vibration displacement of the steam turbine, extracts the vibration displacement and main frequency of the steam turbine bearing by continuous wavelet change, and judges whether the steam turbine has long-period vibration;

[0007] The vibration displacement prediction NARX neural network model extracts the vibration displacement and operating parameters of the turbine bearing, takes the operating parameters as input and the vibration displacement as output, and trains to obtain the vibration displacement NARX model;

[0008] The vibration displacement NARX neural network regulation prediction model uses the vibration displacement prediction NARX neural network model to predict the vibration displacement of the steam turbine, and the long-period vibration instability diagnosis model determines whether the prediction result is in the long-period vibration, thereby determining the optimal oil temperature and safe oil temperature range for the steam turbine to avoid long-period vibration;

[0009] The oil temperature regulation hysteresis NARX neural network model is driven by the difference between the optimal oil temperature and the current oil temperature, and outputs the oil cooling water flow value, which is used by the turbine oil temperature regulation system to regulate the oil temperature.

[0010] In some embodiments, the continuous wavelet variation mathematical expression of the long-period vibration instability diagnosis model is formula (1);

[0011]

[0012] Where x(t) is the signal to be analyzed; t is time; ψ(t)* a,b is the selected wavelet function; a is the scale parameter, which characterizes the degree of expansion and contraction of the wavelet function; b is the position parameter, which characterizes the moving position of the wavelet function; W(a,b) is the result of continuous wavelet transform under the scale parameter a and the position parameter b.

[0013] In some embodiments, the long-period vibration instability diagnostic model obtains the spectrum of the vibration displacement signal after performing continuous wavelet transform on the turbine vibration displacement signal, and performs logical judgment on the main frequency of the vibration displacement signal. When the vibration displacement signal has a main frequency and the main frequency falls within the frequency region of long-period vibration, it is determined that long-period vibration occurs.

[0014] In some embodiments, the operating parameters extracted by the vibration displacement prediction NARX neural network model include time-course data of oil temperature, power, hydrogen temperature, oil pressure, and hydrogen pressure parameters.

[0015] In some embodiments, the vibration displacement prediction NARX neural network model takes the operating parameters as input and the vibration displacement as output, and the vibration displacement NARX model is obtained by training with the goal of minimizing the mean square error through the Bayesian regularization method. The output function of the vibration displacement NARX model is as shown in formula (2);

[0016] y(t)=f(y(t-1), y(t-2),…,y(td),u(t),u(t-1),…,u(td)) (2)

[0017] Wherein, t is time; y(t) is the output function at time t; f is a nonlinear function; y(t-1), y(t-2),…, y(td) are the vibration displacements of the output at the past d time points, and u(t), u(t-1),…, u(td) are the operating parameters of the external input at the current and past d time points.

[0018] In some embodiments, the vibration displacement prediction NARX neural network model performs empirical mode decomposition, filtering and noise reduction on the external input operating parameters, and decomposes the operating parameter signal into an intrinsic mode function IMF and a residual term r N , as shown in formula (3);

[0019]

[0020] Where t is the time and N is the number of intrinsic mode functions (IMFs) obtained by decomposition.

[0021] In some embodiments, the vibration displacement prediction NARX neural network model performs empirical mode decomposition on the external input operating parameters, specifically including: identifying all local maxima and local minima in the operating parameter signal; connecting all local maxima and local minima by spline interpolation method, constructing the upper envelope and the lower envelope, and calculating the mean of the upper envelope and the lower envelope; subtracting the mean of the upper envelope and the lower envelope from the operating parameter signal, and checking whether it satisfies the local symmetry and the local zero mean, if so, taking it as an intrinsic mode function IMF, and subtracting the new signal of the intrinsic mode function IMF from the operating parameter signal as the operating parameter signal, and repeating the above steps until the operating parameter signal no longer contains oscillation components.

[0022] In some embodiments, the vibration displacement NARX neural network adjustment prediction model sets the oil temperature range according to the characteristics of the turbine system and evenly divides the oil temperature range into several oil temperature intervals, takes the endpoints of the oil temperature intervals as the input values ​​of the operating parameters, and uses the vibration displacement NARX model obtained by training the vibration displacement prediction NARX neural network model to predict the vibration displacement of the turbine, and uses the long-period vibration instability diagnosis model to determine whether the vibration displacement prediction result of this oil temperature occurs long-period vibration; the oil temperature value with the minimum vibration displacement value is selected as the optimal oil temperature based on the prediction results of all oil temperature endpoints; at the same time, the oil temperature range in which long-period vibration does not occur is determined, and the oil temperature range is the safe oil temperature range.

[0023] In some embodiments, the oil temperature regulation hysteresis NARX neural network model records the cooling water flow regulation amount, cooling water temperature, average cooling water flow, turbine power, initial oil temperature, oil flow, and regulated oil temperature amount corresponding to the regulated oil cooling water flow to establish a sample database, with cooling water temperature, cooling water flow, turbine power, initial oil temperature, oil flow, and target oil temperature as input, and cooling water flow regulation amount as output, using the Bayesian regularization method, and training with the goal of minimizing the mean square error to obtain the oil temperature regulation hysteresis NARX neural network model; the input value of the cooling water flow includes data from the past n steps, and through training they have different weights.

[0024] In some embodiments, during the training process, the Bayesian regularization method introduces the negative logarithmic probability of the prior distribution as a penalty term and adds it to the original loss function. The loss function is expressed as formula (4);

[0025]

[0026] Where L(w) is the total loss function, p(D|w) is the probability of the data set D given the model with weight parameter w; -logp(D|w) is the negative log-likelihood; is the regularization term; λ is the regularization coefficient.

[0027] The implementation of the present invention has the following beneficial effects:

[0028] The present invention proposes a long-period vibration prediction and regulation system based on NARX modeling. The system uses the vibration displacement prediction NARX to perform training prediction and judgment of the long-period vibration instability diagnosis model, thereby realizing the prediction of the long-period vibration displacement of the steam turbine and the identification of the long-period vibration; the system uses the oil temperature regulation hysteresis NARX neural network model to realize accurate and rapid regulation of the oil temperature, thereby compensating for the disadvantage of poor control accuracy caused by hysteresis; the system realizes active control to suppress the long-period vibration of the steam turbine by coupling the vibration displacement prediction NARX neural network model with the oil temperature regulation hysteresis NARX neural network model. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A schematic diagram of a long-period vibration prediction and regulation system based on NARX modeling proposed in an embodiment of the present invention;

[0030] Figure 2 A schematic diagram of a vibration displacement NARX neural network adjustment prediction model of a long-period vibration prediction and adjustment system based on NARX modeling proposed in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The technical solution of the present invention is described clearly and completely below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] like Figure 1 to Figure 2 As shown, the present invention proposes a long-period vibration prediction and regulation system based on NARX modeling, characterized in that the system includes a long-period vibration instability diagnosis model, a vibration displacement prediction NARX neural network model, a vibration displacement NARX neural network regulation prediction model and an oil temperature regulation hysteresis NARX neural network model.

[0033] The long-period vibration instability diagnosis model identifies the long-period vibration of the vibration displacement, extracts the vibration displacement and main frequency of the turbine bearing through continuous wavelet transformation, and determines whether the turbine has long-period vibration. Wavelet transform decomposes the signal into components of different frequencies and can perform localized analysis in the time domain and frequency domain. Unlike Fourier transform, wavelet transform has multi-resolution analysis capabilities and can provide time and frequency information at the same time, so it has advantages in processing non-stationary signals. Continuous wavelet transform convolves the signal with the scaled and translated mother wavelet to obtain wavelet coefficients of different scales and positions. Its mathematical expression is formula (1).

[0034]

[0035] Where x(t) is the signal to be analyzed; t is time; ψ(t)* a,b is the selected wavelet function; a is the scale parameter, which characterizes the degree of expansion and contraction of the wavelet function; b is the position parameter, which characterizes the moving position of the wavelet function; W(a,b) is the result of continuous wavelet transform under scale parameter a and position parameter b. After continuous wavelet transform of the vibration displacement signal, the spectrum of the vibration signal is obtained, and the main frequency of the vibration is logically judged. When the main frequency exists and the main frequency frequency falls in the frequency region of long-period vibration, it is considered that long-period vibration occurs.

[0036] The vibration displacement prediction NARX neural network model extracts the long-period vibration displacement of the turbine bearing and the time history data of parameters such as oil temperature, power, hydrogen temperature, oil pressure, and hydrogen pressure, and performs noise cleaning on the data. With oil temperature, power, hydrogen temperature, oil pressure, and hydrogen pressure as input and vibration displacement as output, the vibration displacement NARX model is obtained through Bayesian regularization method and training with the goal of minimizing the mean square error.

[0037] The functional structure of the NARX model for vibration displacement prediction is:

[0038] y(t)=f(y(t-1), y(t-2),…,y(td),u(t),u(t-1),…,u(td)) (2)

[0039] Where t is time; y(t) is the output function at time t; f is a nonlinear function; y(t-1), y(t-2), …, y(td) are the output vibration displacements at the past d time points, and u(t), u(t-1), …, u(td) are the operating parameters of the external inputs at the current and past d time points. The external inputs are mainly operating parameters closely related to long-period vibrations, including turbine operating power, oil temperature, hydrogen temperature, oil pressure, hydrogen pressure, etc.

[0040] For the external input operating parameters, filtering and noise reduction are performed through empirical mode decomposition; the complex signal is decomposed into a series of intrinsic mode functions IMF and a residual term rN. Each IMF is a single oscillation mode with local symmetry and local zero mean.

[0041]

[0042] The main steps of Empirical Mode Decomposition are:

[0043] Step 1: Extreme point detection, specifically identifying all local maxima and local minima in the operating parameter signal.

[0044] Step 2: Envelope construction, specifically connecting all local maxima and local minima through spline interpolation method to construct upper envelope and lower envelope respectively.

[0045] Step 3: Mean calculation, specifically calculating the mean of the upper envelope and the lower envelope.

[0046] Step 4: Extract IMF, specifically, subtract the mean from the original signal of the operating parameters to obtain a preliminary IMF candidate. If the IMF conditions of local symmetry and local zero mean are met, it is regarded as an IMF. Otherwise, repeat steps 1-4 until the IMF is extracted.

[0047] Step 5: Signal updating, specifically, subtracting the extracted IMF from the original signal of the operating parameters to obtain a new signal.

[0048] Step 6: Repeat steps 1-5 until the operating parameter signal no longer contains oscillation components.

[0049] After the external input operation parameter signal is decomposed, the high-frequency noise or low-frequency disturbance of the operation parameter signal is filtered and denoised.

[0050] In the training process of the vibration displacement prediction NARX model, in order to prevent the neural network from overfitting, the loss function uses Bayesian regularization. Bayesian regularization can effectively avoid overfitting of machine learning models. By introducing the idea of ​​Bayesian statistics into the regularization process, Bayesian regularization not only improves the generalization ability of the model, but also provides uncertainty estimates of model parameters. During the training process, the negative logarithmic probability of the prior distribution is introduced as a penalty term and added to the original loss function. The loss function is expressed as formula (4).

[0051]

[0052] Among them, L(w) is the total loss function, p(D|w) is the probability of data set D given the model with weight parameter w, which is usually calculated by the probability distribution predicted by the model. -log p(D|w) is the negative log-likelihood, which measures the degree of fit of the model to data set D under the weight parameter w. is the regularization term, which is a penalty term for model complexity. It prevents overfitting by penalizing the sum of squares of parameters. λ is the regularization coefficient, which is used to control the strength of the regularization term. A larger λ value will increase the impact of regularization, making the model parameters smaller, thereby reducing overfitting. A smaller λ value will reduce the impact of regularization.

[0053] The purpose of the vibration displacement NARX neural network regulation prediction model is to determine the optimal oil temperature and safe oil temperature range. When long-period vibration occurs in the current state, the model outputs the optimal oil temperature as the target oil temperature value, and the regulation system operates at this target oil temperature value, and the system will suppress the occurrence of long-period vibration. When long-period vibration does not occur in the current state, the model outputs a safe oil temperature range, and the system operates in this oil temperature range to prevent the occurrence of long-period vibration in the system.

[0054] like Figure 2As shown in the figure, the vibration displacement NARX neural network adjustment prediction model is constructed based on the transfer learning vibration displacement prediction NARX neural network model. First, according to the system characteristics of the steam turbine, the adjustable range of the system oil temperature, the minimum oil temperature Tmin and the maximum oil temperature Tmax are determined. The range is evenly divided into m oil temperature intervals, and Tmin, Tmin+(Tmax-Tmin) / m, Tmin+(Tmax-Tmin)*2 / m, …Tmin+(Tmax-Tmin)*i / m, .., Tmax, a total of m+1 oil temperature points are taken. Each oil temperature value and the current value of other input parameters are used as input in turn. The vibration displacement prediction NARX neural network model is used to predict the vibration displacement of rotating equipment such as steam turbines, and the long-period vibration instability diagnosis model is used to determine whether the oil temperature has long-period vibration. Comparing the vibration displacement values ​​corresponding to each oil temperature value, the minimum vibration displacement value is the optimal oil temperature. If the system is in long-period vibration under the current working condition, the optimal oil temperature is used as the adjustment target to achieve long-period vibration suppression. At the same time, the oil temperature range in which long-period vibration does not occur is determined. This range is the safe oil temperature range. The operation of the system within this oil temperature range can prevent the occurrence of long-period vibration.

[0055] The oil temperature regulation hysteresis NARX neural network model is driven by the difference between the optimal oil temperature and the current oil temperature, and outputs the adjustment oil cooling water flow value. The turbine oil temperature regulation system achieves oil temperature regulation by adjusting the oil cooling water flow, and the adjustment oil cooling water flow value is used for the turbine oil temperature regulation system to adjust the oil temperature. The training of the oil temperature regulation hysteresis NARX neural network model first builds a high-quality database. By adjusting the oil cooling water flow, the cooling water flow adjustment amount, cooling water temperature, average cooling water flow, turbine power, initial oil temperature, oil flow, and oil temperature adjustment data are recorded to establish a database sample. The cooling water temperature, cooling water flow, turbine power, initial oil temperature, oil flow, and target oil temperature are used as inputs, and the cooling water flow adjustment amount is output. Due to the hysteresis problem of the influence of water temperature on oil temperature, the cooling water flow Q input by the model contains the data of the past n steps. Q(t-△t), Q(t-2△t),…, Q(tn△t) are all used as input values, and they have different weights through training. The training adopts Bayesian regularization method and takes the minimization of mean square error as the training target to obtain the oil temperature regulation hysteresis NARX neural network model. The model predicts the current water flow regulation value in real time, thereby achieving accurate control of oil temperature and preventing overshoot.

[0056] In general, the present invention extracts the vibration displacement of the turbine bearing, extracts its vibration frequency and amplitude, and the long-period vibration instability diagnosis model determines whether the turbine shaft system has long-period vibration according to the vibration frequency and amplitude. When long-period vibration of the shaft system occurs, the vibration displacement NARX neural network adjustment prediction model and the vibration displacement prediction NARX neural network model are called, and the optimal oil temperature for suppressing long-period vibration is determined by the optimization method. The difference between the optimal oil temperature and the current oil temperature is used as the driving force, and the oil cooling water flow is adjusted through the output of the NARX oil temperature adjustment hysteresis neural network model, thereby adjusting the oil temperature of the turbine to avoid long-term vibration.

[0057] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the attached claims.

Claims

1. A long-period vibration prediction and regulation system based on NARX modeling, characterized in that: The system includes a long-period vibration instability diagnosis model, a vibration displacement prediction NARX neural network model, a vibration displacement NARX neural network regulation prediction model and an oil temperature regulation hysteresis NARX neural network model. The long-period vibration instability diagnosis model identifies the long-period vibration of the turbine vibration displacement, extracts the vibration displacement and main frequency of the turbine bearing through continuous wavelet changes, and determines whether the turbine has long-period vibration. The vibration displacement prediction NARX neural network model extracts the vibration displacement and operating parameters of the turbine bearing, takes the operating parameters as input and the vibration displacement as output, and trains to obtain the vibration displacement NARX model; The vibration displacement NARX neural network adjustment prediction model uses the vibration displacement prediction NARX neural network model to predict the vibration displacement of the steam turbine, and the long-period vibration instability diagnosis model determines whether the prediction result is in long-period vibration, thereby determining the optimal oil temperature and safe oil temperature range for the steam turbine not to have long-period vibration; The oil temperature regulation hysteresis NARX neural network model is driven by the difference between the optimal oil temperature and the current oil temperature, and outputs a regulated oil cooling water flow value, which is used by the turbine oil temperature regulation system to regulate the oil temperature.

2. A long-period vibration prediction and regulation system based on NARX modeling according to claim 1, characterized in that: The continuous wavelet variation mathematical expression of the long-period vibration instability diagnosis model is formula (1); Where x(t) is the signal to be analyzed; t is time; ψ(t)* a,b is the selected wavelet function; a is the scale parameter, which characterizes the degree of expansion and contraction of the wavelet function; b is the position parameter, which characterizes the moving position of the wavelet function; W(a,b) is the result of continuous wavelet transform under the scale parameter a and the position parameter b.

3. A long-period vibration prediction and regulation system based on NARX modeling according to claim 2, characterized in that: The long-period vibration instability diagnostic model obtains the spectrum of the vibration displacement signal after performing continuous wavelet transform on the turbine vibration displacement signal, and performs logical judgment on the main frequency of the vibration displacement signal. When the vibration displacement signal has a main frequency and the main frequency falls within the frequency region of long-period vibration, it is determined that long-period vibration occurs.

4. The long-period vibration prediction and regulation system based on NARX modeling according to claim 1 is characterized in that: The operating parameters extracted by the vibration displacement prediction NARX neural network model include time-course data of oil temperature, power, hydrogen temperature, oil pressure, and hydrogen pressure parameters.

5. A long-period vibration prediction and regulation system based on NARX modeling according to claim 4, characterized in that: The vibration displacement prediction NARX neural network model takes the operating parameters as input and the vibration displacement as output. The vibration displacement NARX model is obtained by training with the goal of minimizing the mean square error through the Bayesian regularization method. The output function of the vibration displacement NARX model is as shown in formula (2); y(t)=f(y(t-1), y(t-2),…,y(td),u(t),u(t-1),…,u(td)) (2) Wherein, t is time; y(t) is the output function at time t; f is a nonlinear function; y(t-1), y(t-2),…, y(td) are the vibration displacements of the output at the past d time points, and u(t), u(t-1),…, u(td) are the operating parameters of the external input at the current and past d time points.

6. A long-period vibration prediction and regulation system based on NARX modeling according to claim 4, characterized in that: The vibration displacement prediction NARX neural network model performs empirical mode decomposition, filtering and noise reduction on the external input operating parameters, and decomposes the operating parameter signal into an intrinsic mode function IMF and a residual term r N , as shown in formula (3); Where t is the time and N is the number of intrinsic mode functions (IMFs) obtained by decomposition.

7. A long-period vibration prediction and regulation system based on NARX modeling according to claim 6, characterized in that: The vibration displacement prediction NARX neural network model performs empirical mode decomposition on the external input operating parameters, specifically including: identifying all local maxima and local minima in the operating parameter signal; connecting all local maxima and local minima by spline interpolation method, constructing the upper envelope and the lower envelope, and calculating the mean of the upper envelope and the lower envelope; subtracting the mean of the upper envelope and the lower envelope from the operating parameter signal, and checking whether it satisfies the local symmetry and the local zero mean, if so, taking it as an intrinsic mode function IMF, and subtracting the new signal of the intrinsic mode function IMF from the operating parameter signal as the operating parameter signal, and repeating the above steps until the operating parameter signal no longer contains oscillation components.

8. The long-period vibration prediction and regulation system based on NARX modeling according to claim 1 is characterized in that: The vibration displacement NARX neural network adjustment prediction model sets the oil temperature range according to the characteristics of the steam turbine system and evenly divides the oil temperature range into several oil temperature intervals. The endpoints of the oil temperature intervals are used as input values ​​of the operating parameters. The vibration displacement NARX model obtained by training the vibration displacement prediction NARX neural network model is used to predict the vibration displacement of the steam turbine, and the long-period vibration instability diagnosis model is used to determine whether the vibration displacement prediction result of this oil temperature occurs long-period vibration; the oil temperature value with the minimum vibration displacement value is selected as the optimal oil temperature based on the prediction results of all oil temperature endpoints; at the same time, the oil temperature range in which no long-period vibration occurs is determined, and the oil temperature range is the safe oil temperature range.

9. The long-period vibration prediction and regulation system based on NARX modeling according to claim 1 is characterized in that: The oil temperature regulation hysteresis NARX neural network model records the cooling water flow regulation amount, cooling water temperature, average cooling water flow, turbine power, initial oil temperature, oil flow, and regulated oil temperature amount corresponding to the fuel-saving cooling water flow to establish a sample database, with cooling water temperature, cooling water flow, turbine power, initial oil temperature, oil flow, and target oil temperature as input, and cooling water flow regulation amount as output, using the Bayesian regularization method, and training with the goal of minimizing the mean square error to obtain the oil temperature regulation hysteresis NARX neural network model; the input value of the cooling water flow includes data from the past n steps, and through training, they have different weights.

10. The long-period vibration prediction and regulation system based on NARX modeling according to claim 5, characterized in that: The Bayesian regularization method introduces the negative logarithmic probability of the prior distribution as a penalty term during the training process and adds it to the original loss function. The loss function is expressed as formula (4); Where L(w) is the total loss function, p(D|w) is the probability of the dataset D given the model with weight parameter w; -log p(D|w) is the negative log-likelihood; is the regularization term; λ is the regularization coefficient.

Images