DeepAR-based diversion canal slope displacement dynamic interval prediction and anomaly identification method

Abstract

The invention relates to the technical field of hydraulic engineering safety monitoring, in particular to a DeepAR-based diversion canal slope displacement dynamic interval prediction and anomaly recognition method, which comprises the following steps of: selecting an input variable; calculating correlation; a DeepAR model is constructed; training and predicting the model; evaluating a prediction effect; carrying out abnormal identification on displacement prediction; the method has the advantages that displacement probability distribution at different time points is obtained, the mean value and variance of displacement prediction are dynamically calculated, prediction intervals changing along with time under different confidence coefficients are calculated through the probability distribution, an interval prediction model of bandwidth changing along with time is constructed, prediction precision is improved, and meanwhile the slope displacement state is obtained more comprehensively.

Description

Technical Field

[0001] The present invention relates to the field of water conservancy project safety monitoring, and specifically to a method for dynamic interval prediction and anomaly identification of water diversion channel slope displacement based on DeepAR. Background Art

[0002] As an important part of the water diversion project, the stability of the water diversion channel slope is directly related to the safety and normal operation of the water diversion project. Therefore, it is necessary to strengthen the safety monitoring and management of the water diversion channel slope to prevent slope accidents. Displacement is an important basis for judging slope instability. Predicting the future change trend of displacement is of great significance to ensure the safety of the water diversion channel slope.

[0003] Models commonly used in slope displacement prediction, such as back propagation neural network

[0004] The output of the Backpropagation Neural Network (BP), Support Vector Machine (SVM), Recurrent Neural Network (RNN), etc. is usually a single displacement prediction value, i.e., a point prediction result. Due to the limitations of the knowledge of the diversion channel slope system, the noise interference in the monitoring data, and the influence of other uncertain factors, the uncertainty of displacement prediction is inevitable.

[0005] Therefore, a dynamic interval prediction and anomaly identification method for diversion channel slope displacement based on DeepAR is needed to solve the above technical problems. Summary of the invention

[0006] The purpose of the present invention is to provide a dynamic interval prediction and anomaly identification method for water diversion channel slope displacement based on DeepAR, obtain the displacement probability distribution at different time points, dynamically calculate the mean and variance of the displacement prediction, calculate the prediction interval that changes with time under different confidence levels through the probability distribution, construct an interval prediction model whose bandwidth changes with time, improve the prediction accuracy, and obtain the slope displacement state more comprehensively to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: a method for dynamic interval prediction and abnormality identification of water diversion channel slope displacement based on DeepAR, the method comprising the following steps:

[0008] Selection of input variables;

[0009] Correlation calculation;

[0010] Construction of DeepAR model;

[0011] Model training and prediction;

[0012] Evaluation of prediction effect;

[0013] Anomaly identification for displacement prediction.

[0014] Preferably, the selection of the input variables specifically includes: the influencing factors of the displacement of the water diversion channel slope mainly include water level, time effect and rainfall.

[0015] Preferably, the correlation calculation specifically includes:

[0016] The displacement has a certain correlation with the water level and the higher power of the water level. 2 , H 3 As the water level influencing factors, the Pearson correlation coefficient method is used to calculate the correlation between the previous rainfall and time effect and the displacement, and the rainfall and time effect influencing factors with strong correlation are selected, as follows:

[0017] Suppose two variable sequences are X=(x1,x2…x n ), Y=(y1,y2…y n ), the Pearson correlation coefficient is calculated using the following formula:

[0018]

[0019] in are the means of the sequences of the corresponding variables, x t ,y t are the monitoring values ​​of different variables at different time points, and n is the sample size.

[0020] Preferably, the construction of the DeepAR model specifically includes:

[0021] The hidden layer output h at the current moment t As shown below:

[0022] h t =h(h t-1 , z t-1 , x t )=f(W hz z t-1 +W hx x t +W hh h t-1 +b) (2)

[0023] Where: z t represents the displacement value at time t, X t Represents the influencing factor at time t. At time t, the input of the network includes the influencing factor x t , the value z at the previous moment t-1 , and the hidden layer output h at the previous momentt-1 ,h t is the output of the hidden layer at time t, W hz is the weight corresponding to the displacement data from the input layer to the hidden layer, W hx is the weight corresponding to the input layer impact factor to the hidden layer, W hh is the weight from the hidden layer output at the previous time point to the hidden layer output at the next time point, b is the bias term from the input layer to the hidden layer, and f(x) is the hyperbolic tangent activation function, which is specifically expressed as follows:

[0024]

[0025] Get the hidden layer output h t Finally, calculate the mean and standard deviation of the Gaussian distribution:

[0026] μ t =W μh h t +b μ (4)

[0027] σ t =ln[1+exp(W σh h t +b σ )] (5)

[0028] Where: W μh , b μ Calculate the output layer μ t The corresponding weight and bias term, W σh , b σ Calculate the output layer σ t The corresponding weights and bias terms are obtained by calculating the mean and variance at each time point to obtain the Gaussian likelihood function as follows:

[0029]

[0030] Preferably, the training and prediction of the model specifically includes:

[0031] In the training phase, the network input includes the influencing factor and the displacement value of the previous moment, as well as the hidden layer output of the previous moment, so as to calculate the parameter mean μ and standard deviation σ of the likelihood function. It is believed that the closer the fitting distribution of the network distribution output is to the actual situation, that is, the closer the fitting distribution is to the actual situation, the better the probability is. t There is a great probability value at z t Get the maximum value and construct the training objective function as follows:

[0032]

[0033] Where t0 is the time when prediction starts. In the DeepAR model, the negative form of the objective function is Train the loss function. In actual training, when the loss function tends to be stable, the network training is considered to be completed, that is, the objective function reaches the maximum;

[0034] In the prediction stage, DeepAR aims to find the probability distribution of the current moment based on the displacement values ​​and impact factor values ​​at the historical moments. The joint distribution consists of the product of Gaussian likelihood functions at different time points.

[0035]

[0036] Where: T is the prediction end time, represents the displacement value of the prediction time period, represents the displacement value of the training period, x1:T is the influencing factor; Q(z t |z 1：t-1 , x 1：T ) is the conditional probability distribution, that is, given z 1：t-1 and x 1：T Probability distribution of displacement under conditions; is the joint distribution of the model, which is the conditional probability Q(z t |z 1：t-1 , x 1：T )

[0037] Preferably, the prediction effect evaluation specifically includes:

[0038] The root mean square error E is used as the evaluation index of point prediction RMSE , mean absolute percentage error E MAPE , and the interval evaluation index interval coverage probability P PICP , the normalized average width of the prediction interval P PINAW As an evaluation index of model performance, the corresponding calculation formula is as follows:

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] Where: N is the number of prediction samples, z t is the measured displacement value, Output displacement values ​​for the model, is the minimum and maximum value of the prediction interval at time t, z max , z min is the maximum and minimum displacement within the prediction period; P PICP The larger the value, the more true values ​​fall within the prediction interval. PICP It is necessary to achieve the corresponding confidence level, namely the nominal confidence level, as much as possible, otherwise the obtained prediction interval is considered to be insufficiently reliable; P PINAW The smaller the value, the narrower the prediction interval. PICP Same as P PINAW The smaller it is, the better the prediction effect of the model.

[0045] Preferably, the abnormality identification of the displacement prediction specifically includes:

[0046] The interval prediction result with a confidence level of 90% is selected as the standard for judging abnormalities. When the actual displacement value falls within the prediction interval corresponding to this confidence level, it is determined that the actual displacement value is in a normal state under this confidence level. On the contrary, if the actual displacement value exceeds the range of the prediction interval, it is determined that the actual displacement value is abnormal under this confidence level, and the alarm mechanism is triggered accordingly.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The dynamic interval prediction and anomaly identification method of water diversion channel slope displacement based on DeepAR proposed in the present invention obtains the displacement probability distribution at different time points, dynamically calculates the mean and variance of the displacement prediction, calculates the prediction interval that changes with time under different confidence levels through the probability distribution, and constructs an interval prediction model whose bandwidth changes with time, thereby improving the prediction accuracy and obtaining a more comprehensive slope displacement state. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is the displacement and rainfall data graph of the present invention;

[0050] Figure 2 This is the DeepAR training period loss function curve of the present invention

[0051] Figure 3 This is the displacement prediction effect diagram of the DeepAR model in 2015 of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose and technical solution of the present invention clearly and completely described, and the advantages more clearly understood, the embodiments of the present invention are further described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of the embodiments, and are only used to explain the embodiments of the present invention, and are not used to limit 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.

[0053] For example, see Figure 1-Figure 2 The present invention provides a technical solution: a method for dynamic interval prediction and abnormal identification of water diversion channel slope displacement based on DeepAR, the method comprising the following steps:

[0054] Selection of input variables: The factors affecting the displacement of the diversion channel slope mainly include water level, time and rainfall.

[0055] Correlation calculation: displacement has a certain correlation with water level and its higher power. H, H 2 , H 3 As the water level influencing factors, the Pearson correlation coefficient method is used to calculate the correlation between the previous rainfall and time effect and the displacement, and the rainfall and time effect influencing factors with strong correlation are selected, as follows:

[0056] Suppose two variable sequences are X=(x1,x2…x n ), Y=(y1,y2…y n ), the Pearson correlation coefficient is calculated using the following formula:

[0057]

[0058] in are the means of the sequences of the corresponding variables, x t ,y t are the monitoring values ​​of different variables at different time points, and n is the sample size.

[0059] Construction of DeepAR model; the hidden layer output h at the current moment t As shown below:

[0060] h t =h(h t-1 , z t-1 , x t )=f(W hz z t-1 +W hx x t +W hh h t-1 +b) (2)

[0061] Where: zt represents the displacement value at time t, X t Represents the influencing factor at time t. At time t, the input of the network includes the influencing factor x t , the value z at the previous moment t-1 , and the hidden layer output h at the previous moment t-1 ,h t is the output of the hidden layer at time t, W hz is the weight corresponding to the displacement data from the input layer to the hidden layer, W hx is the weight corresponding to the input layer impact factor to the hidden layer, W hh is the weight from the hidden layer output at the previous time point to the hidden layer output at the next time point, b is the bias term from the input layer to the hidden layer, and f(x) is the hyperbolic tangent activation function, which is specifically expressed as follows:

[0062]

[0063] Get the hidden layer output h t Finally, calculate the mean and standard deviation of the Gaussian distribution:

[0064] μ t =W μh h t +b μ (4)

[0065] σ t =ln[1+exp(W σh h t +b σ )] (5)

[0066] Where: W μh , b μ Calculate the output layer μ t The corresponding weight and bias term, W σh , b σ Calculate the output layer σ t The corresponding weights and bias terms are obtained by calculating the mean and variance at each time point to obtain the Gaussian likelihood function as follows:

[0067]

[0068] Model training and prediction; in the training stage, the network input includes the influencing factor and the displacement value of the previous moment, as well as the hidden layer output of the previous moment, so as to calculate the parameter mean μ and standard deviation σ of the likelihood function. It is believed that the closer the fitting distribution of the network distribution output is to the actual situation, that is, the closer the fitting distribution is to the actual situation, the closer the fitting distribution is to the actual situation. t There is a great probability value at z t Get the maximum value and construct the training objective function as follows:

[0069]

[0070] Where t0 is the time when prediction starts. In the DeepAR model, the negative form of the objective function is Train the loss function. In actual training, when the loss function tends to be stable, the network training is considered to be completed, that is, the objective function reaches the maximum;

[0071] In the prediction stage, DeepAR aims to find the probability distribution of the current moment based on the displacement values ​​and impact factor values ​​at the historical moments. The joint distribution consists of the product of Gaussian likelihood functions at different time points.

[0072]

[0073] Where: T is the prediction end time, represents the displacement value of the prediction time period, represents the displacement value of the training period, x 1：T is the impact factor; Q(z t |z 1：t-1 , x 1：T ) is the conditional probability distribution, that is, given z 1：t-1 and x 1：T Probability distribution of displacement under conditions; is the joint distribution of the model, which is the conditional probability Q(z t |z 1：t-1 , x 1：T )

[0074] Evaluation of prediction effect; use the point prediction evaluation index root mean square error E RMSE , mean absolute percentage error E MAPE , and the interval evaluation index interval coverage probability P PICP , the normalized average width of the prediction interval P PINAW As an evaluation index of model performance, the corresponding calculation formula is as follows:

[0075]

[0076]

[0077] Where: N is the number of prediction samples, z t is the measured displacement value, Output displacement values ​​for the model, is the minimum and maximum value of the prediction interval at time t, z max , z min is the maximum and minimum displacement within the prediction period; P PICP The larger the value, the more true values ​​fall within the prediction interval.PICP It is necessary to achieve the corresponding confidence level, namely the nominal confidence level, as much as possible, otherwise the obtained prediction interval is considered to be insufficiently reliable; P PINAW The smaller the value, the narrower the prediction interval. PICP Same as P PINAW The smaller it is, the better the prediction effect of the model.

[0078] For abnormal identification of displacement prediction, the interval prediction result with a confidence level of 90% is selected as the standard for judging abnormalities. When the actual displacement value falls within the prediction interval corresponding to this confidence level, it is determined that the actual displacement value is in a normal state under this confidence level. On the contrary, if the actual displacement value exceeds the range of the prediction interval, it is determined that the actual displacement value is abnormal under this confidence level, and the alarm mechanism is triggered accordingly.

[0079] Embodiment 2, refer to the attached Figure 3 As shown, based on the first embodiment, the slope of a channel on the middle line is taken as the research object. The channel is located from the north of Zhanghe River to the southern section of the ancient canal. The designed water depth is 6m, the water-passing section is trapezoidal, and the designed slope coefficient is 2.35. A total of 8 monitoring sections are arranged on the channel slope. The displacement data of the monitoring point QDBM503 of the 103+658 section of the high fill section is taken as an example for analysis. Considering the uncertainty of displacement, a displacement interval prediction model based on DeepAR is constructed, and the mean and variance of the displacement prediction value at different time points are calculated to obtain the probability distribution of displacement with time difference. The range of possible changes in future values ​​is predicted through confidence intervals under a certain confidence level, and the dynamic changes of uncertainty are analyzed to achieve dynamic interval prediction of displacement, as follows:

[0080] Select rainfall influencing factors Q 1 , Q 3 , Q 5 , Q 7 , Q 9 , Q 11 , Q 13 , time-effect factor T, T 0.5 , T -0.5 , lnT, 1-e -T / 100 , T / (T+1), and the correlation between different influencing factors and displacement is calculated using formula (1), where Q 1 represents the previous day's rainfall and Q 3 Represents the rainfall in the previous three days, and the others are similar. T is the number of days from the observation date to the start date of measurement. A total of 29 sets of data on displacement, rainfall, and time series from August 2014 to May 2015 are used to calculate the correlation between different influencing factors and displacement. The displacement and rainfall data of this period are as follows: Figure 1 The correlation calculation results are shown in Table 1.

[0081] Table 1 Calculation results of Pearson correlation coefficients of rainfall, time effect and displacement

[0082]

[0083] When the absolute value of the Pearson correlation coefficient is closer to 1, the correlation between the two variables is higher. 5 , Q 11 , Q 13 ,T,T 0.5 , lnT has a high correlation with displacement, so they are used as the influencing factors of rainfall and time effect into the model.

[0084] A DeepAR model with a three-layer recurrent neural network hidden layer was constructed, with a learning rate of 0.001, 15 neurons, and 100 iterations. The displacement data from August 2014 to May 2015 was used as the training set, and the model loss function was calculated. The change process is shown in the following figure. Figure 2 shown.

[0085] It can be seen that the model has stabilized when the number of iterations is 20. At this time, the model training is completed. The trained model is used to predict the displacement in the next three months. The prediction results are as follows: Figure 3 shown.

[0086] Depend on Figure 3 It can be seen that the model prediction results are consistent with the actual displacement trend. PICP The maximum bandwidth is 0.35 mm and the minimum bandwidth is 0.24 mm. PICP The maximum bandwidth is 0.96mm and the minimum bandwidth is 0.67mm. PICP They are all greater than their corresponding nominal confidence levels, the reliability of interval prediction is high, and the interval widths at different confidence levels change dynamically over time.

[0087] The DeepAR interval measurement model is used to predict the displacement interval values ​​from May 14, 2015 to August 27, 2015, and further identify the abnormality of slope displacement. Taking the interval prediction results at 90% confidence level as an example, the slope displacement is analyzed. The displacement interval prediction results at 90% confidence level are shown in Table 2.

[0088] Table 2 Displacement interval prediction results at 90% confidence level

[0089]

[0090] Taking the minimum predicted bandwidth in the table as an example to identify abnormal slope displacement, the displacement prediction interval bandwidth on May 23, 2015 in the table is the smallest, which is 0.67mm, and the displacement prediction interval is [17.86mm, 18.53mm]. The actual measured displacement value of the slope on that day was 17.9mm, and the actual displacement value fell within the prediction interval. Therefore, at a confidence level of 90%, the displacement at this time point is within the normal range and there is no abnormality.

[0091] Taking the maximum bandwidth within the prediction time period as an example, the interval bandwidth on July 11, 2015 was 0.96mm, which is the maximum bandwidth within the prediction time period. The prediction interval is [18.90mm, 19.86mm]. The actual displacement value of the slope on that day was 19.3mm, which also fell within the prediction interval. Therefore, at a confidence level of 90%, the predicted displacement value at this time point is also within the normal range and there is no abnormality.

[0092] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for dynamic interval prediction and anomaly identification of diversion channel slope displacement based on DeepAR, characterized by: The method comprises the following steps: Selection of input variables; Correlation calculation; Construction of DeepAR model; Model training and prediction; Evaluation of prediction effect; Anomaly identification for displacement prediction.

2. The method for dynamic interval prediction and abnormal identification of water diversion channel slope displacement based on DeepAR according to claim 1 is characterized by: The selection of the input variables specifically includes: the influencing factors of the displacement of the water diversion channel slope mainly include water level, time effect and rainfall.

3. The method for dynamic interval prediction and abnormal identification of diversion channel slope displacement based on DeepAR according to claim 1 is characterized in that: The correlation calculation specifically includes: The displacement has a certain correlation with the water level and the higher power of the water level. 2 , H 3 As the water level influencing factors, the Pearson correlation coefficient method is used to calculate the correlation between the previous rainfall and time effect and the displacement, and the rainfall and time effect influencing factors with strong correlation are selected, as follows: Suppose two variable sequences are X=(x1,x2…x n ), Y=(y1,y2…y n ), the Pearson correlation coefficient is calculated using the following formula: in are the means of the sequences of the corresponding variables, x t ,y t are the monitoring values ​​of different variables at different time points, and n is the sample size.

4. The method for dynamic interval prediction and abnormal identification of diversion channel slope displacement based on DeepAR according to claim 1 is characterized in that: The construction of the DeepAR model specifically includes: The hidden layer output h at the current moment t As shown below: h t =h(h t-1 ,z t-1 ,x t )=f(W hz z t-1 +W hx x t +W hh h t-1 +b)(2) Where: z t represents the displacement value at time t, x t Represents the influencing factor at time t. At time t, the input of the network includes the influencing factor x t , the value z at the previous moment t-1 , and the hidden layer output h at the previous moment t-1 ,h t is the output of the hidden layer at time t, W hz is the weight corresponding to the displacement data from the input layer to the hidden layer, W hx is the weight corresponding to the input layer influencing factor to the hidden layer, W hh is the weight from the hidden layer output at the previous time point to the hidden layer output at the next time point, b is the bias term from the input layer to the hidden layer, and f(x) is the hyperbolic tangent activation function, which is specifically expressed as follows: Get the hidden layer output h t Finally, calculate the mean and standard deviation of the Gaussian distribution: m t =W μh h t +b μ (4) σ t =ln[1+exp(W σh h t +b σ )] (5) Where: W μh , b μ Calculate the output layer μ t The corresponding weight and bias term, W σh , b σ Calculate the output layer σ t The corresponding weights and bias terms are obtained by calculating the mean and variance at each time point to obtain the Gaussian likelihood function as follows:

5. The method for dynamic interval prediction and abnormal identification of water diversion channel slope displacement based on DeepAR according to claim 1 is characterized in that: The training and prediction of the model specifically include: In the training phase, the network input includes the influencing factor and the displacement value of the previous moment, as well as the hidden layer output of the previous moment, so as to calculate the parameter mean μ and standard deviation σ of the likelihood function. It is believed that the closer the fitting distribution of the network distribution output is to the actual situation, that is, the closer the fitting distribution is to the actual situation, the better the probability is. t There is a great probability value at z t Get the maximum value and construct the training objective function as follows: Where t0 is the time when prediction starts. In the DeepAR model, the negative form of the objective function is Train the loss function. In actual training, when the loss function tends to be stable, the network training is considered to be completed, that is, the objective function reaches the maximum; In the prediction stage, DeepAR aims to find the probability distribution of the current moment based on the displacement values ​​and impact factor values ​​at the historical moments. The joint distribution consists of the product of Gaussian likelihood functions at different time points. Where: T is the prediction end time, represents the displacement value of the prediction time period, represents the displacement value of the training period, x 1:T is the impact factor; Q(z t |z 1:t-1 ,x 1:T ) is the conditional probability distribution, that is, given z 1:t-1 and x 1:T Probability distribution of displacement under conditions; is the joint distribution of the model, and is the conditional probability Q(z t |z 1:t-1 ,x 1:T ) 6. The method for dynamic interval prediction and abnormal identification of water diversion channel slope displacement based on DeepAR according to claim 1 is characterized by: The prediction effect evaluation specifically includes: The root mean square error E is used as the evaluation index of point prediction RMSE , mean absolute percentage error E MAPE , and the interval evaluation index interval coverage probability P PICP , the normalized average width of the prediction interval P PINAW As an evaluation index of model performance, the corresponding calculation formula is as follows: Where: N is the number of prediction samples, z t is the measured displacement value, Output displacement values ​​for the model, is the minimum and maximum value of the prediction interval at time t, z max ,z min is the maximum and minimum displacement within the prediction period; P PICP The larger the value, the more true values ​​fall within the prediction interval. PICP It is necessary to achieve the corresponding confidence level, namely the nominal confidence level, as much as possible, otherwise the obtained prediction interval is considered to be insufficiently reliable; P PINAW The smaller the value, the narrower the prediction interval. PICP Same as P PINAW The smaller it is, the better the prediction effect of the model.

7. The method for dynamic interval prediction and abnormal identification of diversion channel slope displacement based on DeepAR according to claim 1 is characterized by: The abnormal identification of the displacement prediction specifically includes: The interval prediction result with a confidence level of 90% is selected as the standard for judging abnormalities. When the actual displacement value falls within the prediction interval corresponding to this confidence level, it is determined that the actual displacement value is in a normal state under this confidence level. On the contrary, if the actual displacement value exceeds the range of the prediction interval, it is determined that the actual displacement value is abnormal under this confidence level, and the alarm mechanism is triggered accordingly.

Images