Deformation prediction method of rockfill concrete dam based on hysteresis analysis of environmental factors

Abstract

The present invention provides a deformation prediction method for rockfill concrete dam based on hysteresis analysis of environmental factors, comprising the following steps: Step 1: Perform partial autocorrelation analysis on deformation monitoring values ​​of dam deformation time series to determine deformation factor lag time; Step 2: Perform CEEMDAN decomposition on deformation measurement value sequence to obtain decomposition sequence, calculate sequence sample entropy value, and remove noise sequence with high sample entropy value; Step 3: Generate environmental quantity lag sequence according to different time sliding window lag lengths; Step 4: Calculate the Pearson correlation coefficient between deformation measurement value decomposition sequence and environmental quantity lag sequence, and select environmental quantity lag sequence with the highest correlation with each decomposition sequence as lag factor; Step 5: Construct deformation prediction input data set according to deformation factor and lag factor, and use IPSO-GRU model to predict the deformation value of rockfill concrete dam. Application of this technical solution can effectively improve the prediction accuracy of the model.

Description

Technical Field

[0001] The invention relates to the technical field of dam deformation prediction, in particular to a method for predicting deformation of a rockfill concrete dam based on hysteresis analysis of environmental factors. Background Art

[0002] As a new type of dam-building material, rockfill concrete has been widely used in recent years due to its good durability, ease of construction, and gentle hydration heat temperature rise. At present, there are more than 100 rockfill concrete dams under construction or already built in China. How to establish an effective safety monitoring method will lay an important foundation for the promotion and application of this type of dam.

[0003] The dam deformation prediction and analysis method is the core project of concrete dam safety monitoring. Studies have shown that the environmental factors affecting dam deformation, such as water level and temperature, have significant hysteresis effects on dam deformation. The inventors found that the existing concrete dam deformation prediction model lacks targeted analysis methods, especially quantitative analysis methods, for the hysteresis of environmental factors, and the corresponding research on rockfill concrete materials as a new type of dam construction material is even weaker. Therefore, in view of the needs of rockfill concrete dam safety monitoring, how to quantitatively analyze the hysteresis effect of environmental factors on dam deformation and build a more effective deformation prediction method on this basis is the problem to be solved at present. Summary of the invention

[0004] In view of this, the object of the present invention is to provide a deformation prediction method for rockfill concrete dam based on hysteresis analysis of environmental factors, so as to effectively improve the prediction accuracy of the model.

[0005] To achieve the above object, the present invention adopts the following technical solution: A deformation prediction method for rockfill concrete dam based on hysteresis analysis of environmental factors comprises the following steps:

[0006] Step 1: Conduct partial autocorrelation analysis on the deformation monitoring values ​​of the dam deformation time series, and determine the deformation factor lag time based on the maximum interval time when the partial autocorrelation coefficient falls within the specified confidence interval and remains stable;

[0007] Step 2: Perform CEEMDAN decomposition on the deformation measurement sequence to obtain the decomposed sequence, calculate the sequence sample entropy value, and remove the noise sequence with higher sample entropy value;

[0008] Step 3: Generate environmental quantity lag series according to different time sliding window lag lengths;

[0009] Step 4: Calculate the Pearson correlation coefficient between the deformation measurement value decomposition sequence and the environmental quantity lag sequence, and select the environmental quantity lag sequence with the highest correlation with each decomposition sequence as the lag factor;

[0010] Step 5: Construct a deformation prediction input data set based on the deformation factor and hysteresis factor, and apply the IPSO-GRU model to predict the deformation value of the rockfill concrete dam.

[0011] In a preferred embodiment, the specified confidence interval of step 1 is obtained by the specified confidence probability of the PACF value sequence of the deformation sequence training set, and the deformation factor lag time is determined by the maximum time that the partial autocorrelation coefficient of the deformation measurement time series stably falls into the specified probability confidence interval; the following steps are included:

[0012] Step 11: Calculate the training set scale factor split according to the deformation sequence and select the training set sequence:

[0013] disp[:]=DISP[:]*split (1)

[0014] Step 12: Calculate the partial autocorrelation coefficient of the training set sequence and obtain the corresponding training set partial autocorrelation coefficient sequence value:

[0015] PACF_disp[:]=pacf(disp[:],length(disp[:])) (2)

[0016] Step 13: Calculate the range of the confidence interval of the partial autocorrelation coefficient based on the specified confidence probability PV:

[0017]

[0018] lower_limit=interval(PV,loc=PACF_disp[:].mean,

[0019] scale=sqrt(∑(PACF_disp[:]-PACF_disp[:].mean) / length(PACF_disp[:]))[0]

[0020] Step 14: Calculate the maximum lag days D_lag that stably falls into the specified confidence interval. i The pacf value is calculated according to the range of partial autocorrelation coefficient calculated in step 13, and the following formula is calculated:

[0021] interval=(PACF_disp[D i ]-upper_limit)*(PACF_disp[D i ]-lower_limit) (4)

[0022] Let D be the lag days with interval values ​​greater than 0. j , the maximum value is the lag days of deformation factor D_lag.

[0023] In a preferred embodiment, in step 2, the deformed sequence is decomposed by CEEMDAN to obtain the number of decomposed sequences num_imf, and by calculating the entropy value of each sequence sample, the maximum entropy value of the decomposed sequence sample is regarded as a noise sequence, which is eliminated to obtain the feature sequence matrix imf_mat[:,:], thereby realizing denoising of the decomposed sequence.

[0024] In a preferred embodiment, step 3 generates an environmental quantity hysteresis matrix according to different time sliding window hysteresis lengths, and generates an environmental quantity hysteresis sequence according to the set hysteresis period lag_day, hysteresis sliding group number lag_set, and environmental factors H, H2, H3, and T; comprising the following steps:

[0025] Step 31: Calculate the maximum lag time timewin = lag_day*lag_set, and initialize the hysteresis matrix of each environmental quantity according to the length from the maximum lag time timewin to the target measurement value sequence number dest_number:

[0026]

[0027]

[0028]

[0029]

[0030] Step 32: Add the raw value of each environmental quantity to the first column of the hysteresis matrix:

[0031]

[0032] Step 33: Assemble the environmental quantity hysteresis matrix according to the hysteresis period lag_day and the number of hysteresis slip groups lag_set:

[0033]

[0034] In a preferred embodiment, step 4 calculates the Pearson correlation coefficient between the deformation measurement value decomposition sequence and the environmental quantity lag sequence, and selects the environmental quantity lag sequence with the highest correlation among the decomposition sequences as the environmental quantity lag factor, which includes the following steps:

[0035] Step 41: Generate a correlation matrix between the deformation feature sequence and the environmental quantity hysteresis sequence; according to the deformation measured value imf sequence matrix imf_mat[:,:] calculated in step 2, calculate the Pearson correlation estimate between each feature sequence in imf_mat[:,:] and the corresponding sequence of each environmental quantity hysteresis matrix calculated in step 3, and save it in the correlation matrix cov_mat[:,:]:

[0036]

[0037] Step 42: Determine the environmental quantity hysteresis sequence with the highest correlation with each deformation feature sequence. For each deformation feature sequence, calculate the environmental quantity factor num_fac[i] and the sliding time window period num_lag[i] corresponding to the item with the highest absolute value in the correlation matrix cov_mat[:,i]:

[0038]

[0039] Step 43: Output the hysteresis factor with the highest correlation for each deformation feature sequence and the corresponding hysteresis slip time window period to obtain a set of environmental quantity hysteresis factors.

[0040] In a preferred embodiment, the prediction model of step 5 is based on the deformation lag factor D_lag determined in step 1, the environmental lag factor determined in step 4, H1-n, H2-n, H3-n, Tn..., n represents the lag slip time window period, combined with water level, temperature, and aging factor to form a training set input matrix:

[0041] δ=δ H +δ T +δ θ +(H1,H2,H3,H4,T) 环境量滞后因子集 +D_lag 变形滞后因子 (10)

[0042] In a preferred embodiment, the IPSO-GRU prediction model of step 5 includes the following steps:

[0043] Step 51: Initialize the calculation, perform sensitivity analysis based on the deformation sequence characteristics of the rockfill concrete dam, generate the learning rate learning_rate, the number of hidden nodes hidden_node, the number of iterations num_iter value range lr_range[:], hn_range[:], nit_range[:], and set the initial optimal objective function value best_fit to satisfy best_fit>0;

[0044] Step 52: Calculate the objective function value cal_fit. If cal_fit> the optimal objective function value best_fit, perform the assignment operation best_fit=cal_fit.

[0045] Step 53: Use the IPSO method to update the population position corresponding to the learning rate learning_rate, the number of hidden nodes hidden_node, and the number of iterations num_iter and calculate cal_fit and best_fit;

[0046] Step 54: Determine whether the number of iterations num_iter is less than the maximum number of iterations max_iter. If not, stop the calculation and output the optimal parameters of the learning rate, number of hidden nodes, number of iterations and the deformation measurement sequence prediction result y_predict[:] calculated by the GRU model. Otherwise, repeat steps 52 to 53.

[0047] Compared with the prior art, the present invention has the following beneficial effects: the present invention considers the hysteresis of the influence of the factors affecting dam deformation on the dam, obtains the original monitoring sequence features after stripping through a signal decomposition method, obtains the hysteresis factors corresponding to each feature by calculating the correlation between each feature and the hysteresis period of different factors, constructs a prediction model factor set in combination with the hysteresis influence of the displacement measurement itself, applies the IPSO-GRU method to predict the deformation of the rockfill concrete dam, and quantifies the selection of hysteresis factors, which can effectively improve the prediction accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A method flow chart of a preferred embodiment of the present invention;

[0049] Figure 2 The historical measured values ​​of a deformation monitoring point of a rockfill concrete dam in a preferred embodiment of the present invention;

[0050] Figure 3 This is a graph of interval coefficients corresponding to different lag days in a preferred embodiment of the present invention;

[0051] Figure 4 A schematic diagram of removing noise sequences according to sample entropy in a preferred embodiment of the present invention;

[0052] Figure 5 is the environmental quantity hysteresis factor corresponding to the deformation measurement value characteristic sequence of the preferred embodiment of the present invention;

[0053] Figure 6 The model deformation prediction result of the preferred embodiment of the present invention is compared with the actual value. DETAILED DESCRIPTION

[0054] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0055] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0056] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0057] The technical problem to be solved by the present invention is to overcome the problem of low accuracy caused by the lack of quantitative analysis of the hysteresis of environmental factors in the deformation prediction method of rockfill concrete dams, and propose a deformation prediction method of rockfill concrete dams based on quantitative analysis of the hysteresis of environmental factors. First, the lag time length of the deformation factor is determined by partial autocorrelation analysis of the dam deformation measurement time series and the specified confidence interval; then, feature extraction is achieved by decomposing the deformation measurement sequence, and on this basis, correlation analysis is performed with the generated environmental quantity lag sequence with different time slip window lag lengths, and the environmental quantity lag sequence with the highest correlation with each decomposition sequence is selected as the lag factor; finally, a deformation prediction input data set is constructed based on the deformation factor and the lag factor, and the IPSO-GRU model is used to predict the deformation value of the rockfill concrete dam. The specific implementation flow chart is shown in the attached figure. Figure 1 As shown. It mainly includes the following 5 steps:

[0058] Step 1: Conduct partial autocorrelation analysis on the deformation monitoring values ​​of the dam deformation time series, and determine the deformation factor lag time based on the maximum interval time when the partial autocorrelation coefficient falls within the specified confidence interval and remains stable;

[0059] Step 2: Perform CEEMDAN decomposition on the deformation measurement sequence to obtain the decomposed sequence, calculate the sequence sample entropy value, and remove the noise sequence with higher sample entropy value;

[0060] Step 3: Generate environmental quantity lag series according to different time sliding window lag lengths;

[0061] Step 4: Calculate the Pearson correlation coefficient between the deformation measurement value decomposition sequence and the environmental quantity lag sequence, and select the environmental quantity lag sequence with the highest correlation with each decomposition sequence as the lag factor;

[0062] Step 5: Construct a deformation prediction input data set based on the deformation factor and hysteresis factor, and apply the IPSO-GRU model to predict the deformation value of the rockfill concrete dam.

[0063] Take a rockfill concrete dam deformation monitoring point as an example. The historical monitoring values ​​of this point are as follows: Figure 2 As shown, now follow the method flow ( Figure 1 ) for a detailed introduction and description:

[0064] Step 1: Perform partial autocorrelation analysis on the deformation monitoring values ​​of the dam deformation time series, and determine the deformation factor lag time according to the maximum interval time when the partial autocorrelation coefficient falls within the specified confidence interval and remains stable, which includes the following steps:

[0065] ① Calculate the training set ratio factor split=0.6 according to the deformation sequence and select the training set sequence:

[0066] disp[:]=DISP[:]*split (1)

[0067] ② Calculate the partial autocorrelation coefficient of the training set sequence and obtain the corresponding training set partial autocorrelation coefficient sequence value:

[0068] PACF_disp[:]=pacf(disp[:],length(disp[:])) (2)

[0069] ③Calculate the range of the confidence interval of the partial autocorrelation coefficient based on the specified confidence probability PV:

[0070]

[0071] Calculation shows that upper_limit = 0.04472066, lower_limit = -0.03487634.

[0072] ④ Calculate the maximum lag days D_lag that stably falls into the specified confidence interval. For each lag day D i The pacf value is calculated according to the range of partial autocorrelation coefficient calculated in step ③, and the following formula is calculated:

[0073] interval=(PACF_disp[D i ]-upper_limit)*(PACF_disp[D i ]-lower_limit)(4)

[0074] Let D be the lag days with interval values ​​greater than 0. j (Calculation results see Figure 3 ), the maximum value is the lag days of deformation factor D_lag.

[0075] Step 2: Perform CEEMDAN decomposition on the deformation measurement sequence to obtain the decomposed sequence, and calculate the sequence sample entropy. The feature sequences whose sample entropy exceeds the first 95% of the sequence are regarded as noise sequences and removed to obtain the feature sequence matrix imf_mat[:,:]. The feature sequences before and after denoising are as follows: Figure 4 shown.

[0076] Step 3: Generate an environmental quantity hysteresis matrix according to the hysteresis length of different time sliding windows, and generate an environmental quantity hysteresis sequence according to the set hysteresis period lag_day=7, hysteresis sliding group number lag_set=24, and environmental factors H, H2, H3, and T. It includes the following steps:

[0077] S3.1: Calculate the maximum lag time timewin=lag_day*lag_set=168, and initialize the hysteresis matrix of each environmental quantity according to the length from the maximum lag time timewin to the target measurement value sequence number dest_number=2539:

[0078]

[0079] S3.2: Add the raw value of each environmental quantity to the first column of the hysteresis matrix:

[0080]

[0081] S3.3: Assemble the environmental quantity hysteresis matrix according to the hysteresis period lag_day=7 and the number of hysteresis slip groups lag_set=24:

[0082]

[0083]

[0084]

[0085]

[0086] Step 4: Calculate the Pearson correlation coefficient between the deformation measurement value decomposition sequence and the environmental quantity lag sequence, and select the environmental quantity lag sequence with the highest correlation among the decomposition sequences as the environmental quantity lag factor. The following steps are included:

[0087] S4.1: Generate the correlation matrix of deformation feature sequence and environmental quantity hysteresis sequence. According to the deformation measured value imf sequence matrix imf_mat[:,:] calculated in step 2, calculate the Pearson correlation estimate between each feature sequence in imf_mat[:,:] and the corresponding sequence of each environmental quantity hysteresis matrix calculated in step 3, and save it in the correlation matrix cov_mat[:,:]:

[0088]

[0089] S4.2: Determine the environmental quantity hysteresis sequence with the highest correlation with each deformation feature sequence. For each deformation feature sequence, calculate the environmental quantity factor num_fac[i] and the sliding time window period num_lag[i] corresponding to the item with the highest absolute value in the correlation matrix cov_mat[:,i]:

[0090]

[0091] S4.3: Output the hysteresis factor with the highest correlation for each deformation feature sequence and the corresponding hysteresis slip time window period, and obtain the environmental quantity hysteresis factor set (the environmental quantity hysteresis factor set corresponding to the deformation measurement feature sequence is as follows Figure 5 shown).

[0092] Step 5: Based on the deformation lag factor D_lag determined in step 1, the environmental lag factors (H1(0-7), H1(14-21), H2(0-7), H2(7-14), H3(7-14), H3(14-21)) determined in step 4, combined with water level, temperature, and aging factors, the training set input matrix is ​​formed:

[0093]

[0094] On this basis, the IPSO-GRU prediction model is used to calculate the deformation prediction value, which includes the following steps:

[0095] S5.1: Initialize the calculation, perform sensitivity analysis based on the deformation sequence characteristics of the rockfill concrete dam, generate the learning rate learning_rate, the number of hidden nodes hidden_node, the number of iterations num_iter value range lr_range[0.1:0.001], hn_range[0:100], nit_range[0-100], and set the initial optimal objective function value best_fit=1, satisfying best_fit>0;

[0096] S5.2: Calculate the objective function value cal_fit. If cal_fit> the optimal objective function value best_fit, perform the assignment operation best_fit=cal_fit;

[0097] S5.3: Use the IPSO method to update the population position corresponding to the learning rate learning_rate, the number of hidden nodes hidden_node, and the number of iterations num_iter, and calculate cal_fit and best_fit;

[0098] S5.4: Determine whether the number of iterations num_iter is less than the maximum number of iterations max_iter. If not, stop the calculation and output the optimal parameters of the learning rate, number of hidden nodes, number of iterations and the deformation measurement sequence prediction result y_predict[:] calculated by the GRU model. Otherwise, repeat ②-③. The final prediction result of the model is shown in Figure 6 .

[0099] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations within the meaning and scope of the elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0100] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A deformation prediction method for rockfill concrete dam based on hysteresis analysis of environmental factors, characterized in that: The following steps are involved: Step 1: Conduct partial autocorrelation analysis on the deformation monitoring values ​​of the dam deformation time series, and determine the deformation factor lag time based on the maximum interval time when the partial autocorrelation coefficient falls within the specified confidence interval and remains stable; Step 2: Perform CEEMDAN decomposition on the deformation measurement sequence to obtain the decomposed sequence, calculate the sequence sample entropy value, and remove the noise sequence with higher sample entropy value; Step 3: Generate environmental quantity lag series according to different time sliding window lag lengths; Step 4: Calculate the Pearson correlation coefficient between the deformation measurement value decomposition sequence and the environmental quantity lag sequence, and select the environmental quantity lag sequence with the highest correlation with each decomposition sequence as the lag factor; Step 5: Construct a deformation prediction input data set based on the deformation factor and hysteresis factor, and use the IPSO-GRU model to predict the deformation value of the rockfill concrete dam; In step 2, the deformed sequence is decomposed by CEEMDAN to obtain the number of decomposed sequences num_imf, and the sample entropy value of each sequence is calculated. The maximum entropy value of the decomposed sequence sample is regarded as the noise sequence, which is eliminated to obtain the feature sequence matrix imf_mat[:,:], so as to realize the denoising of the decomposed sequence.

2. The method for predicting deformation of rockfill concrete dam based on hysteresis analysis of environmental factors according to claim 1 is characterized in that: The specified confidence interval of step 1 is obtained by the specified confidence probability of the PACF value sequence of the deformation sequence training set, and the deformation factor lag time is determined by the maximum time when the partial autocorrelation coefficient of the deformation measurement time series stably falls into the specified probability confidence interval; the following steps are included: Step 11: Calculate the training set scale factor split according to the deformation sequence and select the training set sequence: disp[:]=DISP[:]*split (1) Step 12: Calculate the partial autocorrelation coefficient of the training set sequence and obtain the corresponding training set partial autocorrelation coefficient sequence value: PACF_disp[:]=pacf (disp[:], length(disp[:])) (2) Step 13: Calculate the range of the confidence interval of the partial autocorrelation coefficient based on the specified confidence probability PV: Step 14: Calculate the maximum lag days D_lag that stably falls into the specified confidence interval. i The pacf value is calculated according to the range of partial autocorrelation coefficient calculated in step 13, and the following formula is calculated: interval=(PACF_disp[D i ]-upper_limit)*(PACF_disp[D i ]-lower_limit) (4) Let D be the lag days with interval values ​​greater than 0. j , the maximum value is the lag days of deformation factor D_lag.

3. The method for predicting deformation of rockfill concrete dam based on hysteresis analysis of environmental factors according to claim 1 is characterized in that: Step 3 generates an environmental quantity hysteresis matrix according to the hysteresis lengths of different time sliding windows, and generates an environmental quantity hysteresis sequence according to the set hysteresis period lag_day, the number of hysteresis sliding groups lag_set, and the environmental factors H, H2, H3, and T; including the following steps: Step 31: Calculate the maximum lag time timewin = lag_day*lag_set, and initialize the hysteresis matrix of each environmental quantity according to the length from the maximum lag time timewin to the target measurement value sequence number dest_number: Step 32: Add the raw value of each environmental quantity to the first column of the hysteresis matrix: Step 33: Assemble the environmental quantity hysteresis matrix according to the hysteresis period lag_day and the number of hysteresis slip groups lag_set:

4. The method for predicting deformation of rockfill concrete dam based on hysteresis analysis of environmental factors according to claim 1 is characterized in that: Step 4: Calculate the Pearson correlation coefficient between the deformation measurement value decomposition sequence and the environmental quantity lag sequence, and select the environmental quantity lag sequence with the highest correlation among the decomposition sequences as the environmental quantity lag factor. The following steps are included: Step 41: Generate a correlation matrix between the deformation feature sequence and the environmental quantity hysteresis sequence; according to the deformation measured value imf sequence matrix imf_mat[:,:] calculated in step 2, calculate the Pearson correlation estimate between each feature sequence in imf_mat[:,:] and the corresponding sequence of each environmental quantity hysteresis matrix calculated in step 3, and save it in the correlation matrix cov_mat[:,:]: Step 42: Determine the environmental quantity hysteresis sequence with the highest correlation with each deformation feature sequence; for each deformation feature sequence, calculate the environmental quantity factor num_fac[i] and the sliding time window period num_lag[i] corresponding to the item with the highest absolute value in the correlation matrix cov_mat[:,i]: Step 43: Output the hysteresis factor with the highest correlation for each deformation feature sequence and the corresponding hysteresis slip time window period to obtain a set of environmental quantity hysteresis factors.

5. The method for predicting deformation of rockfill concrete dam based on hysteresis analysis of environmental factors according to claim 1, characterized in that: The prediction model of step 5 is based on the deformation lag factor D_lag determined in step 1, the environmental lag factor determined in step 4, H1-n, H2-n, H3-n, Tn…, where n represents the lag slip time window period, combined with water level, temperature, and aging factor to form the training set input matrix: d=d H +d T +d θ +(H1,H2,H3,H4,T) 环境量滞后因子集 +D_lag 变形滞后因子 (10)。 6. The method for predicting deformation of rockfill concrete dam based on hysteresis analysis of environmental factors according to claim 1, characterized in that: The IPSO-GRU prediction model in step 5 includes the following steps: Step 51: Initialize the calculation, perform sensitivity analysis based on the deformation sequence characteristics of the rockfill concrete dam, generate the learning rate learning_rate, the number of hidden nodes hidden_node, the number of iterations num_iter value range lr_range[:], hn_range[:], nit_range[:], and set the initial optimal objective function value best_fit to satisfy best_fit>0; Step 52: Calculate the objective function value cal_fit. If cal_fit> the optimal objective function value best_fit, perform the assignment operation best_fit=cal_fit. Step 53: Use the IPSO method to update the population position corresponding to the learning rate learning_rate, the number of hidden nodes hidden_node, and the number of iterations num_iter and calculate cal_fit and best_fit; Step 54: Determine whether the number of iterations num_iter is less than the maximum number of iterations max_iter. If not, stop the calculation and output the optimal parameters of the learning rate, number of hidden nodes, number of iterations and the deformation measurement sequence prediction result y_predict[:] calculated by the GRU model. Otherwise, repeat steps 52 to 53.

Images