Pulsating pressure value prediction method and system, storage medium and electronic equipment

Abstract

The invention belongs to the technical field of hydraulic engineering, and provides a pulsating pressure value prediction method and system, a storage medium and electronic equipment, and the method comprises the steps: obtaining a pulsating pressure value in a force pool in real time; constructing a pressure prediction model based on an adaptive lifting algorithm; setting a sliding window, and constructing a training set based on the pulsating pressure value; training the pressure prediction model by using the training set to obtain a trained pressure prediction model; and predicting a pulsating pressure value in the stilling pool by using the trained pressure prediction model to obtain a predicted pulsating pressure value. According to the method, the change condition of the pulsating pressure value is accurately captured and reflected, the technical problem of poor data monitoring and prediction precision is solved, and the operation efficiency and stability of the stilling pool are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of hydraulic engineering, and in particular relates to a method, system, storage medium and electronic equipment for predicting pulsating pressure values. Background Art

[0002] The stilling basin plays a vital role in the hydraulic system. It is mainly used to reduce the energy of water flow and prevent damage to hydraulic equipment and buildings. The change of pulsating pressure in the stilling basin directly reflects the working status and efficiency of the stilling basin. The accurate monitoring and prediction of this variable is very important.

[0003] In traditional methods, the monitoring of pulsating pressure mainly relies on physical measurement. However, due to the harsh monitoring environment, the pre-buried sensors in this type of method often have some sensor anomalies, signal interference and other undesirable phenomena during measurement. They cannot directly provide real-time pressure information of the energy dissipation pool, and cannot accurately reflect the actual pressure conditions inside the energy dissipation pool container at a specific time and place.

[0004] In addition, due to the complexity of pressure changes in the stilling basin, simple physical monitoring methods cannot generate accurate prediction results. In terms of data processing, previous methods are mainly based on statistical theory, such as using representative values ​​such as mean, median, and variance to process and analyze data. These methods have high requirements for data, such as being sensitive to the existence of outliers and abnormal values, which can easily lead to errors. In addition, traditional data processing methods have limited processing capabilities for time series data, making it difficult to capture the dynamic changes of pulsating pressure over time. Summary of the invention

[0005] The present invention provides a pulsating pressure value prediction method, system, storage medium and electronic equipment.

[0006] A method for predicting a pulsating pressure value of the present invention comprises:

[0007] Obtain the pulsating pressure value in the stilling pool in real time;

[0008] Construct a pressure prediction model based on the adaptive boosting algorithm;

[0009] Setting a sliding window and constructing a training set based on the pulsating pressure value;

[0010] Using the training set to train the pressure prediction model to obtain a trained pressure prediction model;

[0011] The trained pressure prediction model is used to predict the pulsating pressure value in the stilling basin to obtain the predicted pulsating pressure value.

[0012] Further,

[0013] After the real-time acquisition of the pulsating pressure value in the stilling pool, the method further includes:

[0014] Extract features of pulsating pressure values.

[0015] Further,

[0016] The method further comprises: predicting the pulsating pressure value in the stilling pool by using the trained pressure prediction model to obtain the predicted pulsating pressure value;

[0017] The characteristics of the pulsating pressure value are used to determine whether the predicted pulsating pressure value is abnormal.

[0018] Further,

[0019] The feature of extracting the pulsating pressure value includes:

[0020] Performing real-time monitoring of the pulsating pressure in the stilling pool to obtain original monitoring data;

[0021] Preprocessing the original monitoring data to obtain preprocessed monitoring data;

[0022] The characteristics of the pulsating pressure value are extracted from the preprocessed monitoring data.

[0023] Further,

[0024] The pressure prediction model is constructed based on the adaptive boosting algorithm, including:

[0025] A new adaptive boosting algorithm TimeSeriesBoost is built on the basis of the adaptive boosting algorithm AdaBoost. The new adaptive boosting algorithm TimeSeriesBoost is more suitable for the prediction of time series data. A pressure prediction model is constructed based on the new adaptive boosting algorithm TimeSeriesBoost.

[0026] Further,

[0027] The setting of the sliding window and constructing a training set based on the pulsating pressure value include:

[0028] Extract a sliding window of a time series, assuming that the length of the sliding window is l, with [{P i-l ,……,P i-1},P i ] as the sample points in the training set, where {P i-l ,……,P i-1} is the input value, P i Output value for the target.

[0029] Further,

[0030] The step of training the pressure prediction model using the training set to obtain the trained pressure prediction model includes:

[0031] Specify the training rounds;

[0032] Assigning the same weight to each sample point in the training set to obtain the weight of each sample point;

[0033] Based on the weight of each of the sample points, a weak learner is trained to obtain a trained weak learner and a predicted pulsating pressure value of each of the sample points output by the weak learner;

[0034] Obtaining the actual pulsating pressure value and the prediction threshold of each of the sample points, identifying the prediction error sample points based on the predicted pulsating pressure value of each of the sample points, and calculating the error rate of the weak learner;

[0035] Calculating the weight of the weak learner based on the error rate of the weak learner;

[0036] Update the weight of each of the sample points;

[0037] The weak learners and their weights in all the training rounds are summed up to obtain a trained pressure prediction model.

[0038] Further,

[0039] The trained pressure prediction model is:

[0040] F(x)=∑αt*ht(x),

[0041] Where F(x) refers to the trained pressure prediction model; αt refers to the weight of the weak learner obtained after the t-th round of training; ht(x) refers to the t-th round of weak learner.

[0042] The present invention also provides a pulsating pressure value prediction system for implementing the above-mentioned pulsating pressure value prediction method, the system comprising:

[0043] An acquisition module is used to obtain the pulsating pressure value in the stilling pool in real time;

[0044] A model building module, used to build a pressure prediction model based on an adaptive boosting algorithm;

[0045] A training set construction module, used for setting a sliding window and constructing a training set based on the pulsating pressure value;

[0046] A training module, used for training the pressure prediction model using the training set to obtain a trained pressure prediction model;

[0047] The prediction module is used to predict the pulsating pressure value in the stilling basin by using the trained pressure prediction model to obtain the predicted pulsating pressure value.

[0048] The present invention also provides a computer-readable storage medium storing a program or instruction. When the program or instruction is executed on a computer, the computer executes the aforementioned pulsating pressure value prediction method.

[0049] The present invention also provides an electronic device, comprising a processor coupled to a memory; the processor is used to read and execute a computer program stored in the memory to implement the aforementioned pulsating pressure value prediction method.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] The present invention uses high-performance equipment to obtain accurate monitoring data of the energy stilling pool, uses advanced data processing methods to preprocess and extract features of the monitoring data, combines an adaptive boosting algorithm to build a pressure prediction model, predicts the pulsating pressure value, accurately captures and reflects the changes in the pulsating pressure value, solves the technical problem of poor data monitoring and prediction accuracy, and improves the operating efficiency and stability of the energy stilling pool. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 This is a flow chart of the pulsating pressure value prediction method of the present invention;

[0054] Figure 2 It is a schematic diagram of the structure of the pulsating pressure value prediction system of the present invention;

[0055] Figure 3 It is a schematic diagram of the electronic device structure of the present invention.

[0056] Description of reference numerals:

[0057] 201 - acquisition module, 202 - model construction module, 203 - training set construction module, 204 - training module, 205 - prediction module, 301 - processor, 302 - memory. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not 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.

[0059] In addition, in order to have a deeper understanding of the present invention, some of the terms are explained as follows:

[0060] Figure 1 : is a flow chart of a method for predicting a pulsating pressure value provided by an embodiment of the present invention. In one embodiment, the method specifically includes the following steps:

[0061] S101: Obtain the pulsating pressure value in the stilling pool in real time and extract the characteristics of the pulsating pressure value.

[0062] S101-1: Monitor the pulsating pressure in the stilling basin in real time and obtain original monitoring data.

[0063] Select a suitable pressure sensor to monitor the pulsating pressure in the stilling pool in real time. The suitable pressure sensor should have high sensitivity, high accuracy and fast response so as to accurately and in real time monitor the pulsating pressure in the stilling pool.

[0064] Since the pulsating pressure in the stilling pool is unevenly distributed, pressure sensors are installed at multiple suitable locations in the stilling pool, such as the top, bottom, left and right parts of the stilling pool, to ensure multi-point collection and obtain more accurate original monitoring data.

[0065] Set an appropriate data acquisition frequency, which should be high enough to capture the rapid changes in pulsating pressure.

[0066] After obtaining the original monitoring data, use special data acquisition equipment and software to record the original monitoring data, including the pulsating pressure value of the data point, the acquisition time and the acquisition location. The original monitoring data is stored in a safe storage device for subsequent data processing and analysis. If the original monitoring data needs to be analyzed remotely, it can be sent to the data processing center through the data transmission device.

[0067] S101-2: Preprocess the original monitoring data to obtain preprocessed monitoring data.

[0068] Preprocess the original monitoring data. There are three ways to preprocess:

[0069] 1. Process the outliers in the original monitoring data.

[0070] Outliers refer to data that significantly deviate from other pulsating pressure values. Outliers may be caused by measurement errors, equipment failures or other unusual events.

[0071] If the difference between the pulsating pressure value at a certain time point and the average pulsating pressure value exceeds three times the standard deviation of the pulsating pressure value, the pulsating pressure value at that time point is identified as an abnormal value, and the abnormal value is warned or deleted. The judgment formula is shown in formula (1):

[0072] |P i -m|>3σ (1),

[0073] Where P i Refers to the pulsating pressure value at a certain time point; m refers to the average pulsating pressure value; σ refers to the standard deviation of the pulsating pressure value.

[0074] 2. Standardize the original monitoring data.

[0075] If the dimensions or units of the original monitoring data are different, the original monitoring data need to be standardized so that all data fall within the same numerical range.

[0076] The Z-Score standardization method is used for processing. The standardization formula is shown in formula (2):

[0077] P new =(Pm) / σ (2),

[0078] Where P new refers to the pulsating pressure value after standardization; P refers to the pulsating pressure value before standardization; m refers to the average pulsating pressure value; σ refers to the standard deviation of the pulsating pressure value.

[0079] 3. Process the missing values ​​in the original monitoring data.

[0080] If the pulsating pressure values ​​at some time points in the original monitoring data are missing, the time points can be deleted according to the actual situation, or filled in by interpolation, regression, mean replacement and other methods.

[0081] After the original monitoring data are processed according to some or all of the above methods, the obtained data are all pre-processed monitoring data.

[0082] S101-3: Extracting features of the pulsating pressure value from the preprocessed monitoring data.

[0083] The features of the pulsating pressure value are extracted from the preprocessed monitoring data. The features include peak value, mean value, median value, variance, standard deviation and fluctuation range. The extraction process of the above three features is introduced below:

[0084] 1. Peak value of pulsating pressure.

[0085] The peak value of the pulsating pressure value refers to the maximum value of the pulsating pressure value within a certain monitoring period. The extraction formula is shown in formula (3):

[0086] P peak =max(P i ) (3),

[0087] Where P peak Refers to the peak value of pulsating pressure; P i Refers to the pulsating pressure value at a certain point in time.

[0088] 2. The average value of pulsating pressure.

[0089] The mean value of the pulsating pressure value refers to the average value of all pulsating pressure values ​​within a certain monitoring period. The extraction formula is shown in formula (4):

[0090] P mean =∑P i / N (4),

[0091] Where P mean Refers to the mean value of pulsating pressure; P i Refers to the pulsating pressure value at a certain time point; ∑P i Refers to the sum of the pulsating pressure values ​​at all time points; N refers to the total number of all time points.

[0092] 3. The median value of the pulsating pressure.

[0093] The peak value of the pulsating pressure value refers to the median value of all pulsating pressure values ​​within a certain monitoring period. The extraction formula is shown in formula (5):

[0094] P median =median(P1,……,P i ,……,P N ) (5),

[0095] Where P median Refers to the median value of pulsating pressure; P i Refers to the pulsating pressure value at a certain point in time.

[0096] 4. Variance of pulsating pressure value.

[0097]

[0098] In the formula, S 2 refers to the mean of the pulsating pressure value; N refers to the total number of all time points; P i Refers to the pulsating pressure value at a certain point in time; P mean Refers to the mean value of pulsating pressure.

[0099] 5. Standard deviation of pulsating pressure values.

[0100]

[0101] Where, σ refers to the standard deviation of the pulsating pressure value; N refers to the total number of all time points; P i Refers to the pulsating pressure value at a certain time point; P mean Refers to the mean value of pulsating pressure.

[0102] 6. Fluctuation range of pulsating pressure value.

[0103] The fluctuation range of the pulsating pressure value refers to the difference between the maximum and minimum values ​​of the pulsating pressure value within a certain monitoring period. The calculation formula is shown in formula (8):

[0104] P pp =max(P i )-min(P i ) (8),

[0105] Where P pp Refers to the fluctuation range of pulsating pressure value; max(P i ) refers to the maximum value of the pulsating pressure; min(P i ) refers to the minimum value of pulsating pressure.

[0106] S102: Constructing a pressure prediction model based on an adaptive boosting algorithm.

[0107] The adaptive boosting algorithm AdaBoost is an iterative ensemble learning method. The main idea is to adjust the sample weights so that the learner focuses on samples that are difficult to classify, then train a weak learner in each round of iterative training, and then integrate the trained weak learners into a strong learner.

[0108] A new adaptive boosting algorithm TimeSeriesBoost is built based on the adaptive boosting algorithm AdaBoost. The new adaptive boosting algorithm TimeSeriesBoost is more suitable for the prediction of time series data.

[0109] A pressure prediction model is constructed based on the new adaptive boosting algorithm TimeSeriesBoost.

[0110] S103: Setting a sliding window and constructing a training set based on the pulsating pressure value.

[0111] The training set is used to train the pressure prediction model, specifically to train the weak learner.

[0112] Set the pulsating pressure value of the stilling pool to P = {P1, P2, P3, ..., P N}, where P iRefers to the pulsating pressure value at a certain point in time.

[0113] Extract a sliding window of a time series, and assume that the length of the sliding window is l, that is, the pulsating pressure value at the next time point can be predicted through l pulsating pressure values.

[0114] [{P i-l ,……,P i-1},P i ] as the sample points in the training set, where {P i-l ,……,P i-1} is the input value, P i is the target output value, also called the actual pulsating pressure value, assuming that there are n sample points as the training set.

[0115] S104: Using the training set to train the pressure prediction model to obtain a trained pressure prediction model.

[0116] The pressure prediction model is trained using the training set. Assuming t rounds of training, a total of T weak learners are trained in t rounds.

[0117] The training process is as follows:

[0118] S104-1: Assign a weight to each sample point in the training set to obtain the weight of each sample point.

[0119] In the first round, each sample point in the training set is given the same weight. The weighting formula is shown in formula (9):

[0120] D1(j)=1 / n (9),

[0121] Where j refers to a single sample point in the training set; D1(j) refers to the weight of a single sample point in the first round of training; and n refers to the total number of sample points in the training set.

[0122] S104-2: Train the weak learner based on the weight of each sample point to obtain a trained weak learner, thereby obtaining a predicted pulsating pressure value of each sample point output by the trained weak learner.

[0123] The weak learner uses the autoregressive model (abbreviated as "AR"), and the calculation formula is shown in formula (10):

[0124]

[0125] In the formula, ht(x j ) refers to the actual output value of the weak learner obtained after the t-th round of training, that is, the predicted pulsating pressure value of a single sample point; l refers to the length of the sliding window; ar l and β refer to the model parameters of the weak learner; x jlRefers to the input value of a single sample point.

[0126] In the AdaBoost algorithm, for each round t, the model parameters ar of the weak learner in formula (10) need to be obtained through training. l And β, thus obtaining the trained weak learner. In different rounds, the model parameters are considered separately. The model parameters of each round are trained based on the weights of the sample points in the current round, relying on the least squares optimization algorithm to minimize the weighted error, as follows:

[0127]

[0128] Where arg min represents the model parameter that is solved to make the function output as small as possible; n refers to the total number of sample points in the training set; Dt(j) refers to the weight of a single sample point in the tth round of training; y j refers to the target output value of a single sample point; l refers to the length of the sliding window; ar l and β refer to the model parameters of the weak learner; x jl Refers to the input value of a single sample point.

[0129] By solving the above optimization problem, the model parameters of this round of weak learners can be trained, and then the pulsating pressure value can be predicted using this round of weak learners to obtain the predicted pulsating pressure value of each sample point.

[0130] S104-3: Obtain the target output value and prediction threshold of each sample point, identify the prediction error sample points based on the predicted pulsating pressure value of each sample point, and calculate the error rate of the weak learner.

[0131] The target output value and prediction threshold of each sample point are obtained. If the difference between the target output value and the predicted pulsating pressure value of a sample point exceeds the prediction threshold, the sample point is identified as a prediction error sample point. The error rate of the weak learner is calculated based on the prediction error sample point. The calculation formula is shown in formula (12):

[0132] εt=∑Dt(j)*I(y j ≠ht(x j )) (12),

[0133] Where εt refers to the error rate of the weak learner obtained after the tth round of training; Dt(j) refers to the weight of a single sample point in the tth round of training; I() refers to the indicator function, which takes the value of 1 if the conditions in the brackets are met, otherwise it takes the value of 0; y j Refers to the target output value of a single sample point; ht(x j ) refers to the actual output value of the weak learner obtained after the t-th round of training in the t-th round of training, that is, the predicted pulsating pressure value of a single sample point.

[0134] S104-4: Calculate the weight of the weak learner based on the error rate of the weak learner.

[0135] The calculation formula of the weak learner weight is shown in formula (13):

[0136]

[0137] Where αt refers to the weight of the weak learner obtained after the tth round of training; εt refers to the error rate of the weak learner obtained after the tth round of training.

[0138] S104-5: Update the weight of each sample point.

[0139] Based on the weight of each sample point, the weight of the weak learner, the target output value of each sample point, and the predicted pulsating pressure value of each sample point output by the weak learner, the weight of each sample point in the training set is updated for use in the next round of training. The calculation formulas are shown in equations (14) and (15):

[0140]

[0141] Zt=∑Dt(j)*exp(-αt*y j *ht(x j )) (15),

[0142] Where Dt+1(j) refers to the weight of a single sample point in the t+1th round of training; Zt refers to the normalization factor in the tth round of training; Dt(j) refers to the weight of a single sample point in the tth round of training; αt refers to the weight of the weak learner obtained after the tth round of training; y j Refers to the target output value of a single sample point; ht(x j ) refers to the actual output value of the weak learner obtained after the t-th round of training in the t-th round of training, that is, the predicted pulsating pressure value of a single sample point.

[0143] In summary, through steps S104-2 to S104-5, t rounds of training are completed to obtain T weak learners and their corresponding weights. The T weak learners and their weights are added together to obtain the trained pressure prediction model. The calculation formula is shown in formula (16):

[0144] F(x)=∑αt*ht(x) (16),

[0145] Where F(x) refers to the trained pressure prediction model; αt refers to the weight of the weak learner obtained after the tth round of training; ht(x) refers to the weak learner obtained after the tth round of training.

[0146] The pressure prediction model trained according to the above method can better handle the time series characteristics of the pulsating pressure of the stilling basin and improve the accuracy of the prediction of the pulsating pressure value. At the same time, the above method utilizes the characteristics of adaptive adjustment of sample weights in the adaptive boosting algorithm AdaBoost, so that the trained pressure prediction model focuses on the sections that are more difficult to predict, thereby improving the robustness of the model.

[0147] S105: Predicting the pulsating pressure value in the stilling basin using the trained pressure prediction model to obtain a predicted pulsating pressure value.

[0148] The trained pressure prediction model is used to predict the pulsating pressure value to obtain the predicted pulsating pressure value.

[0149] After obtaining the predicted pulsating pressure value, the characteristics of the pulsating pressure value are used to determine whether the predicted pulsating pressure value is abnormal. For example, if the predicted pulsating pressure value frequently shows peaks or minimum values, it can be inferred that the stilling pool may have pressure stability problems. Using the above method, a more accurate prediction of future pressure changes can be made, which is conducive to timely adjustment and optimization of the stilling pool to ensure stable and effective operation of the stilling pool.

[0150] At the same time, corresponding decisions can be made based on the predicted pulsating pressure value. If the predicted pulsating pressure value exceeds the pressure safety threshold that needs to be warned, it can be decided to reduce the water flow to reduce the impact of the water flow on the stilling pool; if the predicted pulsating pressure value continues to be lower than a certain threshold, check whether there is a problem with the equipment, such as blockage; if the predicted pulsating pressure value suddenly drops, check whether some abnormality has occurred, such as rupture, etc. If an abnormality occurs, immediate countermeasures need to be taken. The safety threshold is obtained through specifications, experience or calculation.

[0151] Verification Example

[0152] The pulsating pressure values ​​of a stilling pool at five time points are: P1=10, P2=15, P3=7, P4=9, and P5=11.

[0153] For the pulsating pressure values ​​at the above five time points, the average pulsating pressure value was calculated to be 10.4, and the standard deviation of the pulsating pressure value was 2.65. According to the formula |P i -m|>3σ calculation shows that the absolute deviations of the pulsating pressure values ​​at the five time points from the mean are all less than three times of 2.65, so there are no abnormal values.

[0154] The pulsating pressure values ​​at the above five time points have the same dimension and do not need to be processed.

[0155] The pulsating pressure values ​​at the above five time points are intact and do not require any processing.

[0156] The characteristic values ​​of the pulsating pressure value are calculated as follows: peak value is 15, mean value is 10.4, median value is 10, variance is 8.8, standard deviation is 2.65, and fluctuation range is 8.

[0157] Assume that the length of the sliding window is l = 2, that is, the pulsating pressure value at the next time point can be predicted by two pulsating pressure values. Therefore, there are three sample points as the training set for training the weak learner, namely: [(10,15),7], [(15,7),9] and [(7,9),11]. For [(10,15),7], (10,15) is the input value and 7 is the target output value; for [(15,7),9], (15,7) is the input value and 9 is the target output value; for [(7,9),11], (7,9) is the input value and 11 is the target output value.

[0158] A total of t rounds of training are performed, and T weak learners are trained in t rounds.

[0159] Assign the same weight, i.e. 1 / 3, to the three sample points in the training set.

[0160] Start the first iteration:

[0161] The model parameters obtained by the first round of weak learner optimization training are ar1=0.5, ar2=0.2, and β=0.

[0162] The predicted pulsating pressure values ​​are output by the trained weak learner. The predicted pulsating pressure values ​​of the three sample points are 0.5*10+0.2*15=8, 0.5*15+0.2*7=8.9, and 0.5*7+0.2*9=5.3 respectively.

[0163] The target output values ​​of the three sample points are obtained, which are 7, 9 and 11 respectively; the prediction threshold is 2; the predicted pulsating pressure values ​​of the three sample points are 8, 8.9 and 5.3. It can be seen that the second and third sample points are predicted wrong sample points. Based on the second and third sample points, according to the formula εt=∑Dt(j)*I(y j ≠ht(x j ))Calculate the error rate of the weak learner, specifically ε1=0.66.

[0164] According to the formula Calculate the weight of the weak learner, specifically α1 = -0.41.

[0165] According to the formula The weights of the three sample points in the update are specifically: the weight of the first sample point D2(1)=0.27, the weight of the second sample point D2(2)=0.37, and the weight of the third sample point D2(3)=0.36.

[0166] The first round of iteration is completed through the above method.

[0167] The above method is used to perform subsequent rounds of iterations until the set iteration round t is reached or the error rate of the weak learner no longer decreases.

[0168] The embodiment of the present invention also provides a pulsating pressure value prediction system, such as Figure 2 As shown, including:

[0169] The acquisition module 201 is used to acquire the pulsating pressure value in the stilling basin in real time and extract the characteristics of the pulsating pressure value.

[0170] The model building module 202 is used to build a pressure prediction model based on an adaptive boosting algorithm.

[0171] The training set construction module 203 is used to set a sliding window and construct a training set based on the pulsating pressure value.

[0172] The training module 204 is used to train the pressure prediction model using the training set to obtain a trained pressure prediction model.

[0173] The prediction module 205 is used to predict the pulsating pressure value in the stilling basin by using the trained pressure prediction model to obtain the predicted pulsating pressure value.

[0174] It should be noted here that the above-mentioned acquisition module 201, model construction module 202, training set construction module 203, training module 204 and prediction module 205 correspond to steps S101 to S105 in the embodiment of the pulsating pressure value prediction method, and the examples and application scenarios implemented by the above-mentioned modules and corresponding steps are the same, but are not limited to the contents disclosed in the above-mentioned embodiments.

[0175] An embodiment of the present invention further provides a computer-readable storage medium storing a program or instruction. When the program or instruction is executed on a computer, the computer executes the pulsating pressure value prediction method as described in the above method embodiment.

[0176] like Figure 3 As shown, an embodiment of the present invention further provides an electronic device, comprising: a processor 301, the processor 301 is coupled to a memory 302, and the processor 301 is used to read and execute a computer program stored in the memory 302 to implement the pulsating pressure value prediction method as described in the above method embodiment.

[0177] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting pulsating pressure value, characterized in that: include: Obtain the pulsating pressure value in the stilling pool in real time; Construct a pressure prediction model based on the adaptive boosting algorithm; Setting a sliding window and constructing a training set based on the pulsating pressure value; Using the training set to train the pressure prediction model to obtain a trained pressure prediction model; The trained pressure prediction model is used to predict the pulsating pressure value in the stilling basin to obtain the predicted pulsating pressure value.

2. The method according to claim 1, characterized in that After the real-time acquisition of the pulsating pressure value in the stilling pool, the method further includes: Extract features of pulsating pressure values.

3. The method according to claim 2, characterized in that The method further comprises: predicting the pulsating pressure value in the stilling pool by using the trained pressure prediction model to obtain the predicted pulsating pressure value; The characteristics of the pulsating pressure value are used to determine whether the predicted pulsating pressure value is abnormal.

4. The method according to claim 2, characterized in that: The feature of extracting the pulsating pressure value includes: Performing real-time monitoring of the pulsating pressure in the stilling pool to obtain original monitoring data; Preprocessing the original monitoring data to obtain preprocessed monitoring data; The characteristics of the pulsating pressure value are extracted from the preprocessed monitoring data.

5. The method according to claim 1, characterized in that The pressure prediction model is constructed based on the adaptive boosting algorithm, including: A new adaptive boosting algorithm TimeSeriesBoost is built on the basis of the adaptive boosting algorithm AdaBoost, and a pressure prediction model is constructed based on the new adaptive boosting algorithm TimeSeriesBoost.

6. The method according to claim 1, characterized in that The setting of the sliding window and constructing a training set based on the pulsating pressure value include: Extract a sliding window of a time series, assuming that the length of the sliding window is l, and the sample points in the training set are [{P i-l ,……,P i-1 },P i ], where {P i-l ,……,P i-1 } is the input value, P i Output value for the target.

7. The method according to claim 6, characterized in that The step of training the pressure prediction model using the training set to obtain the trained pressure prediction model includes: Specify the training rounds; Assigning the same weight to each sample point in the training set to obtain the weight of each sample point; Training a weak learner based on the weight of each sample point to obtain a trained weak learner, thereby obtaining a predicted pulsating pressure value of each sample point output by the trained weak learner; Obtaining a target output value and a prediction threshold for each of the sample points, identifying prediction error sample points based on the predicted pulsating pressure value of each of the sample points, and calculating an error rate of the weak learner; Calculating the weight of the weak learner based on the error rate of the weak learner; Update the weight of each of the sample points; The weak learners and their weights in all the training rounds are summed up to obtain a trained pressure prediction model.

8. The method according to claim 7, characterized in that The trained pressure prediction model is: F(x)=∑αt*ht(x), Where F(x) refers to the trained pressure prediction model; αt refers to the weight of the weak learner obtained after the tth round of training; ht(x) refers to the weak learner obtained after the tth round of training.

9. A pulsating pressure value prediction system, characterized in that: include: An acquisition module is used to obtain the pulsating pressure value in the stilling pool in real time; A model building module, used to build a pressure prediction model based on an adaptive boosting algorithm; A training set construction module, used for setting a sliding window and constructing a training set based on the pulsating pressure value; A training module, used for training the pressure prediction model using the training set to obtain a trained pressure prediction model; The prediction module is used to predict the pulsating pressure value in the stilling basin by using the trained pressure prediction model to obtain the predicted pulsating pressure value.

10. A computer-readable storage medium, characterized in that: A program or instruction is stored, and when the program or instruction is run on a computer, the computer is enabled to execute the pulsating pressure value prediction method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: comprising a processor coupled to a memory; The processor is used to read and execute the computer program stored in the memory to implement the pulsating pressure value prediction method according to any one of claims 1-8.

Images