Waste heat boiler state monitoring model operation method, system, equipment and medium

Abstract

The invention discloses a waste heat boiler state monitoring model operation method, system, equipment and medium, and relates to the technical field of boiler state monitoring, and the method comprises the following steps: collecting historical healthy operation data of a power plant, and preprocessing the healthy operation data to obtain model input parameters; establishing an LSTM prediction model based on the model input parameters; optimizing hyper-parameters of the LSTM prediction model through an SSA algorithm to obtain optimal values of the hyper-parameters; and establishing a waste heat boiler state monitoring model of the PCA-SSA-LSTM neural network through the optimal value, and setting an alarm threshold value to realize waste heat boiler state monitoring. According to the method, the neural network model is used, the data processing algorithm is combined, the correlation among the variables is removed, the dimension reduction of the data is realized, the training time of the model can be shortened, and meanwhile, the rule among the historical data can be better learned.

Description

Technical Field

[0001] The present invention relates to the technical field of boiler status monitoring, and in particular to a method, system, equipment and medium for operating a waste heat boiler status monitoring model. Background Art

[0002] Waste heat boiler is an important hub in the combined cycle system. Once a waste heat boiler fails, the impact and loss are huge. It is necessary to monitor the operating status of the waste heat boiler in real time and judge whether the equipment is running healthily to improve the reliability and availability of the unit. At present, there are relatively few studies on the fault monitoring of waste heat boilers. The main research methods for fault monitoring of waste heat boilers are as follows: Fault tree analysis method (offline diagnosis): A fault tree method for waste heat boilers is developed based on offline static analysis. This method cannot provide online monitoring services for waste heat boiler related parameters and cannot immediately identify the occurrence of faults. Physical model method (mechanism modeling): A waste heat boiler fault diagnosis method based on a physical model. This model is mainly used to study the performance degradation of waste heat boilers. The simplified mathematical model cannot be used to predict the degree of current equipment performance degradation. Data-driven method (intelligent diagnosis direction): A fault diagnosis method based on data-driven methods monitors whether the equipment is in healthy operation by comparing the error between the parameter value under actual operating conditions and the predicted value, and can well predict the state parameters of the equipment during operation. Summary of the invention

[0003] In view of the problems existing in the existing waste heat boiler state monitoring model operation method, system, equipment and medium, the present invention is proposed.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, an embodiment of the present invention provides a method for operating a waste heat boiler state monitoring model, which comprises the following steps: Collect the historical healthy operation data of the power plant and pre-process the historical healthy operation data of the power plant to obtain the model input parameters; Establish LSTM prediction model based on model input parameters; The hyperparameters of the LSTM prediction model are optimized through the SSA algorithm to obtain the optimal values ​​of the hyperparameters; The waste heat boiler state monitoring model of PCA-SSA-LSTM neural network is established through the optimal value, and the alarm threshold is set to realize the waste heat boiler state monitoring.

[0005] As a preferred scheme of the waste heat boiler status monitoring model operation method described in the present invention, the step of preprocessing the historical healthy operation data of the power plant includes the selection of input parameters, the selection of output parameters, the removal of shutdown data, the screening of steady-state data, and the dimensionality reduction of input parameters.

[0006] As a preferred solution of the waste heat boiler state monitoring model operation method of the present invention, when preprocessing the historical healthy operation data of the power plant, the principal component analysis method is used, which specifically includes: Assuming that there are n groups of historical healthy operation data of the power plant, each group of data contains m parameters, the output matrix X is: , Normalize the matrix X to get the matrix , specifically expressed as: , In the formula, X represents the original data matrix, Represents the column vector of the i-th parameter, that is, the vector composed of the i-th parameter values ​​of all samples, represents the data of the jth parameter in the i-th group of data; Z represents the standardized data matrix, represents the standardized value of the jth parameter in the i-th group of data, represents the mean of the jth parameter, represents the standard deviation of the jth parameter, n represents the number of groups, and m represents the number of parameters; in: , Then calculate the covariance matrix C of the matrix Z, which is specifically expressed as: , In the formula, C represents the covariance matrix of the standardized matrix Z, which reflects the correlation between parameters. The product of the standardized matrix and its transpose, n-1 represents the degree of freedom correction term, which is used for unbiased estimation; Finally, the eigenvalues ​​of the covariance matrix C are calculated l i and the corresponding eigenvectors; Set the contribution rate of the i-th principal component to k i , the cumulative contribution rate of the first i principal components is p i , and finally get: , are the eigenvalues ​​of the covariance matrix C, is the eigenvalue of the i-th covariance matrix C, x represents a natural number, that is, Represents the contribution rate of the xth principal component.

[0007] As a preferred solution of the method for operating the waste heat boiler state monitoring model of the present invention, when establishing the LSTM prediction model, the following steps are included: Selectively forget the information of the previous moment through the forget gate; The sigmoid layer of the input gate performs selective memory, and then the tanh layer is combined with the sigmoid layer to update the state of the unit; Afterwards, the old unit state is updated, and finally the final output of the LSTM unit is obtained through the output gate.

[0008] As a preferred solution of the method for operating the waste heat boiler state monitoring model of the present invention, the method for obtaining the optimal value of the hyperparameter includes: The SSA algorithm is used to optimize the three hyperparameters of the LSTM layer of the LSTM model: the number of hidden units, the initial learning rate, and the L2 regularization coefficient. The optimization model of LSTM using SSA algorithm is established, and finally the optimal value of hyperparameters is obtained.

[0009] As a preferred solution of the method for operating the waste heat boiler state monitoring model of the present invention, the method for setting the alarm threshold includes: The actual healthy operation data of the power plant is input into the waste heat boiler status monitoring model to calculate the parameter prediction values; The 0.99 quantile of the error distribution is selected as the alarm threshold of the parameter prediction value.

[0010] As a preferred solution of the waste heat boiler state monitoring model operation method of the present invention, the waste heat boiler state monitoring method includes: When the absolute value of the absolute error between the parameter prediction value and the true value exceeds the alarm threshold, it indicates that a suspected failure of the waste heat boiler has occurred, and an alarm is issued to indicate the failure.

[0011] In a second aspect, an embodiment of the present invention provides a waste heat boiler state monitoring model operation system, which includes a data acquisition and preprocessing module, an LSTM prediction model module, an SSA algorithm optimization module, a PCA dimension reduction module, and a state monitoring and alarm module; The data acquisition and preprocessing module is responsible for collecting the historical healthy operation data of the power plant and preprocessing the data to obtain parameters suitable for model input; The LSTM prediction model module establishes an LSTM prediction model based on the preprocessed data to predict the operating state of the waste heat boiler; The SSA algorithm optimization module uses the sparrow search algorithm to optimize the hyperparameters of the LSTM prediction model to obtain the optimal value of the hyperparameters, thereby improving the prediction accuracy of the model; The PCA dimension reduction module integrates principal component analysis technology in the model to perform dimension reduction processing on the original data to reduce redundant information and noise in the data and improve the calculation efficiency and prediction accuracy of the model; The state monitoring and alarm module sets an alarm threshold and realizes the state monitoring of the waste heat boiler by comparing the error between the parameter prediction value and the actual value.

[0012] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, any step of the above-mentioned waste heat boiler state monitoring model operation method is implemented.

[0013] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned waste heat boiler state monitoring model operation method is implemented.

[0014] The beneficial effects of the present invention are: The neural network model is used in combination with the data processing algorithm to remove the correlation between the variables and achieve data dimensionality reduction, which can not only reduce the training time of the model, but also better learn the rules between historical data, and select the 0.99th percentile of the error distribution as the alarm threshold of this parameter, so as to achieve the purpose of accurately predicting and monitoring each parameter. Through the verification of the test set and related indicators, it can be seen that the model has high prediction accuracy and strong generalization ability, and has strong practical application value.

[0015] A waste heat boiler status monitoring model is established, the error between the predicted value and the true value is calculated, and the alarm threshold is obtained based on the error distribution. When the absolute value of the absolute error between the parameter predicted value and the true value exceeds the alarm threshold, it means that the waste heat boiler is suspected to have a fault, thereby achieving the purpose of monitoring the waste heat boiler status and improving the reliability and availability of the unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0017] Figure 1 Flow chart of the operation method of the waste heat boiler condition monitoring model.

[0018] Figure 2 A diagram showing the composition of the heating surface of a waste heat boiler for the operation method of the waste heat boiler condition monitoring model.

[0019] Figure 3 This is the LSTM network unit structure diagram of the waste heat boiler state monitoring model operation method.

[0020] Figure 4 This is the absolute error distribution diagram of the reheat steam temperature of the waste heat boiler condition monitoring model operation method. DETAILED DESCRIPTION

[0021] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention. Example

[0022] Reference Figure 1~Figure 4 , which is the first embodiment of the present invention, and provides a method for operating a waste heat boiler state monitoring model, comprising the following steps: S1. Collect the historical healthy operation data of the power plant and pre-process the healthy operation data to obtain the model input parameters.

[0023] The steps of preprocessing healthy operation data include selection of input parameters, selection of output parameters, removal of shutdown data, screening of steady-state data, and dimensionality reduction of input parameters.

[0024] In this embodiment, the data used comes from the gas-steam combined cycle unit. Combined with the measurement points of the waste heat boiler in the power plant and the properties of the waste heat boiler itself, 22 input parameters are determined, namely unit load / MW, relative humidity / %, ambient temperature / ℃, gas entering boiler temperature / ℃, gas entering boiler pressure / kPa, medium-pressure feed water to medium-pressure drum flow / t / h, feed water main pipe pressure / kPa, reheater inlet temperature / ℃, reheater inlet pressure / kPa, high-pressure cylinder exhaust temperature / ℃, medium-pressure steam temperature / ℃, medium-pressure steam flow / t / h, medium-pressure steam pressure / kPa, reheat desuperheating water flow / t / h, reheat desuperheating water temperature / ℃, high-pressure desuperheating water flow / t / h, high-pressure desuperheating water temperature / ℃, high-pressure economizer flow / t / h, medium-pressure feed water flow / t / h, low-pressure feed water to drum temperature / ℃, low-pressure feed water to drum pressure / kPa, and low-pressure feed water to drum flow / t / h.

[0025] When preprocessing the health operation data, the principal component analysis method is used, which includes: Assume that there are n sets of healthy operation data, each set of data contains m parameters, so the output matrix is: , Standardize the matrix X to get the matrix Z, which is specifically expressed as: , In the formula, X represents the original data matrix, represents the column vector of the i-th parameter, that is, the vector composed of the i-th parameter values ​​of all samples; Z represents the standardized data matrix, eliminating the dimension effect, represents the standardized value of the jth parameter in the i-th group of data, represents the mean of the jth parameter, represents the standard deviation of the j-th parameter.

[0026] in: , Then calculate the covariance matrix C of the matrix Z, which is specifically expressed as:

[0027] In the formula, C represents the covariance matrix of the standardized matrix Z, which reflects the correlation between parameters. The product of the standardized matrix and its transpose, n-1 represents the degree of freedom correction term, which is used for unbiased estimation; according to Calculate the eigenvalues ​​of the covariance matrix C l i And the corresponding eigenvector, that is ; Where E is the identity matrix of the same dimension as the covariance matrix C; Calculate the principal components , setting the contribution rate of the i-th principal component to k i , the cumulative contribution rate of the first i principal components is p i , and finally get: , , is the eigenvalue of the covariance matrix C, indicating the variance of the principal component, E is the unit matrix, ui is the eigenvector corresponding to the i-th eigenvalue, and F i is the i-th principal component, which is a linear combination of the original parameters, X m is the normalized parameter column vector.

[0028] S2. Establish an LSTM prediction model based on model input parameters.

[0029] Selectively forget the information of the previous moment through the forget gate; Previous state input information ht-1 With the current input information x t Input sigmoid function together : ; F t Is the output of the forget gate, which is used to determine how much information in the cell state at the previous moment is retained in the cell state at the current moment. is the bias vector of the forget gate, Weight matrix of the forget gate.

[0030] The sigmoid layer of the input gate performs selective memory, and then the tanh layer is combined with the sigmoid layer to update the state of the unit: , ;I t is the output of the input gate, b i is the bias vector of the input gate, and is the weight matrix of the input gate and subsequent units, b g is the bias vector of the candidate unit state; Then update the old unit status: , and finally the final output of the LSTM unit is obtained through the output gate: , is the weight matrix of the output gate, b y is the bias vector of the output gate, h t is the hidden state at the current moment, Y t is the output of the output gate.

[0031] Where W is the weight, b is the bias matrix, t is the time, C is the unit state, is the sigmoid activation function, and tanh is the tanh activation function.

[0032] S3. Use the SSA algorithm to optimize the hyperparameters of the LSTM prediction model and obtain the optimal values ​​of the hyperparameters.

[0033] Use SSA to optimize the three hyperparameters of the LSTM layer of the LSTM model: the number of hidden units, the initial learning rate, and the L2 regularization coefficient; An SSA-LSTM optimization model is established for the three main steam temperature parameters, and finally the optimal value of the hyperparameter is obtained.

[0034] Assume that a sparrow population consists of n sparrows and the parameter dimension to be optimized is m. The population can be expressed as follows: , the finder is responsible for finding food for the group, so it has a larger food search range.

[0035] In each iteration, the finder's position is updated as follows: , t represents the current iteration number, item max represents the maximum number of iterations, X ij Represents the position information of the i-th sparrow in the j-th parameter dimension.

[0036] is a random number, Indicates the alarm value. represents the safety value; Q is a random number that follows a normal distribution; L is a A matrix in which all elements are equal to 1.

[0037] The position of the joiner is updated as follows: , where X p is the best position currently occupied by the discoverer; Xworst is the current global worst position.

[0038] A represents a 1×d matrix, each element in the matrix is ​​randomly assigned 1 or -1. A + = A T ( AA T ) -1 .

[0039] In the sparrow population, 10% to 20% of the sparrows can identify danger. The initial positions of these sparrows are randomly generated: , X best is the current global optimal position; β is a random number that follows a normal distribution with a mean of 0 and a variance of 1; K is a random number in the interval [-1,1]; f i is the fitness value of the current sparrow individual, f g is the global optimal fitness value, f i is the global worst fitness value; is a constant.

[0040] S4. A waste heat boiler status monitoring model of the PCA-SSA-LSTM neural network is established through the optimal value, and the alarm threshold is set to realize the waste heat boiler status monitoring.

[0041] The operating data generated by actual healthy operation is input into the waste heat boiler condition monitoring model to calculate the parameter prediction values; The 0.99 quantile of the error distribution is selected as the alarm threshold of the parameter prediction value.

[0042] Methods for monitoring the status of waste heat boilers include: When the absolute value of the absolute error between the parameter prediction value and the true value exceeds the alarm threshold, it indicates that a suspected failure of the waste heat boiler has occurred, and an alarm is issued to indicate the failure.

[0043] In summary, the use of neural network models, combined with data processing algorithms, removes the correlation between variables and achieves data dimensionality reduction, which not only reduces the training time of the model, but also better learns the rules between historical data, and selects the 0.99th percentile of the error distribution as the alarm threshold of this parameter, so as to achieve the purpose of accurately predicting and monitoring each parameter. Through the verification of the test set and related indicators, it can be seen that the model has high prediction accuracy and strong generalization ability, and has strong practical application value.

[0044] A waste heat boiler status monitoring model is established, the error between the predicted value and the true value is calculated, and the alarm threshold is obtained based on the error distribution. When the absolute value of the absolute error between the parameter predicted value and the true value exceeds the alarm threshold, it means that the waste heat boiler is suspected to have a fault, thereby achieving the purpose of monitoring the waste heat boiler status and improving the reliability and availability of the unit. Example

[0045] Reference Figure 1~Figure 4 As shown, Figure 1 The following is a flow chart of the method. Using the operation data of the waste heat boiler of a combined cycle unit in a gas turbine power plant, the technical solution is described below.

[0046] The first is the data collection and preprocessing part. The data is collected from the operating data of a combined cycle unit in a gas turbine power plant. The time span is two months, the data collection time interval is 20s, and the data is mainly the data of the unit during steady-state operation.

[0047] Data preprocessing includes the selection of input and output parameters, removal of shutdown data, screening of steady-state data, and dimensionality reduction of input parameters. After setting the initial hyperparameters and establishing the LSTM model, SSA is used to optimize the hyperparameters of the LSTM model corresponding to each output parameter, and then the optimal hyperparameters are substituted into different output parameter prediction models.

[0048] Figure 2 This is the composition diagram of the heating surface of the waste heat boiler. The heating surface of the boiler is divided into three pressure levels: high, medium and low. The final destination of the steam generated by these three pressure level systems is as follows: the low-pressure main steam generated by the second low-pressure superheater will be mixed with the exhaust steam of the intermediate pressure cylinder of the steam turbine and enter the low-pressure cylinder to do work.

[0049] The intermediate pressure steam generated by the intermediate pressure superheater is mixed with the exhaust steam of the high pressure cylinder and enters the first reheater for heating. Then, the steam temperature is adjusted by the reheater desuperheater and enters the second reheater. The reheated steam produced by the second reheater enters the intermediate pressure cylinder to perform work. The steam generated by the first high-pressure superheater enters the second high-pressure superheater after the temperature is adjusted by the high-pressure steam desuperheater. The high-pressure main steam generated by the second high-pressure superheater enters the high-pressure cylinder to do work. The heat source of the flow medium on each heating surface is the exhaust gas of the combustion engine.

[0050] Figure 3 This is the LSTM network unit structure diagram. First, LSTM selectively forgets the information of the previous moment through the forget gate. The previous state input information h t-1 With the current input information x t Input sigmoid function together .

[0051] Then, the sigmoid layer of the input gate performs selective memory, and then the tanh layer is combined with the sigmoid layer to update the state of the unit. After that, LSTM updates the old unit state, and finally, the final output of the LSTM unit is obtained through the output gate.

[0052] The steam temperature parameters are trained to establish the corresponding prediction models, and the prediction results of the models are verified using the test set. The absolute error distribution of the reheat steam temperature model is as follows: Figure 4 As shown, the same treatment is adopted for the high-pressure main steam temperature and the low-pressure main steam temperature.

[0053] The distribution of these absolute errors is approximately normal distribution, so the 0.99 quantile of these three distributions is selected as the alarm threshold of these three parameters. The alarm thresholds of these parameters are obtained by calculation. When the absolute value of the absolute error between the predicted value and the true value of the parameter exceeds the alarm threshold, it means that the waste heat boiler is suspected to have a fault. Example

[0054] On the basis of the first embodiment, this embodiment further provides a waste heat boiler state monitoring model operation system, including a data acquisition and preprocessing module, an LSTM prediction model module, an SSA algorithm optimization module, a PCA dimension reduction module, and a state monitoring and alarm module; The data acquisition and preprocessing module is responsible for collecting the historical healthy operation data of the power plant and preprocessing the data to obtain parameters suitable for model input; The LSTM prediction model module establishes an LSTM prediction model based on the preprocessed data to predict the operating state of the waste heat boiler; The SSA algorithm optimization module uses the sparrow search algorithm to optimize the hyperparameters of the LSTM prediction model to obtain the optimal value of the hyperparameters, thereby improving the prediction accuracy of the model; The PCA dimension reduction module integrates principal component analysis technology in the model to perform dimension reduction processing on the original data to reduce redundant information and noise in the data and improve the calculation efficiency and prediction accuracy of the model; The state monitoring and alarm module sets an alarm threshold and realizes the state monitoring of the waste heat boiler by comparing the error between the parameter prediction value and the actual value.

[0055] This embodiment also provides a computer device, which is suitable for the operation method of the waste heat boiler state monitoring model, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the operation method of the waste heat boiler state monitoring model proposed in the above embodiment.

[0056] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0057] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for implementing the waste heat boiler state monitoring model operation method proposed in the above embodiment is implemented.

[0058] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0059] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for operating a waste heat boiler state monitoring model, characterized in that: The following steps are included: Collect the historical healthy operation data of the power plant and pre-process the historical healthy operation data of the power plant to obtain the model input parameters; Establish a long short-term memory network LSTM prediction model based on the model input parameters; The hyperparameters of the LSTM prediction model are optimized through the sparrow search SSA algorithm to obtain the optimal value of the hyperparameters; The waste heat boiler state monitoring model of principal component analysis PCA-SSA-LSTM neural network is established through the optimal value, and the alarm threshold is set to realize the waste heat boiler state monitoring.

2. The method for operating a waste heat boiler state monitoring model according to claim 1, characterized in that: The steps of preprocessing the historical healthy operation data of the power plant include the selection of input parameters, the selection of output parameters, the removal of shutdown data, the screening of steady-state data, and the dimensionality reduction of input parameters.

3. The method for operating a waste heat boiler state monitoring model according to claim 2, characterized in that: When preprocessing the historical healthy operation data of the power plant, the principal component analysis method is used, which includes: Assuming that there are n groups of historical healthy operation data of the power plant, each group of data contains m parameters, the output matrix X is: , Normalize the matrix X to get the matrix , specifically expressed as: , In the formula, X represents the original data matrix, Represents the column vector of the i-th parameter, that is, the vector composed of the i-th parameter values ​​of all samples, Represents the data of the jth parameter in the i-th group of data, represents the column vector of the i-th parameter of the n-th group; Z represents the standardized data matrix, represents the standardized value of the jth parameter in the i-th group of data, represents the mean of the jth parameter, represents the standard deviation of the jth parameter, n represents the number of groups, and m represents the number of parameters; in: , , Then calculate the covariance matrix C of the matrix Z, which is specifically expressed as: ； In the formula, C represents the covariance matrix of the standardized matrix Z, reflecting the correlation between parameters, the product of the standardized matrix and its transpose, and n-1 represents the degree of freedom correction term, which is used for unbiased estimation; Finally, the eigenvalues ​​of the covariance matrix C are calculated l i and the corresponding eigenvectors; Set the contribution rate of the i-th principal component to k i , the cumulative contribution rate of the first i principal components is p i , and finally get: ， are the eigenvalues ​​of the covariance matrix C, is the eigenvalue of the i-th covariance matrix C, x represents a natural number, that is, Represents the contribution rate of the xth principal component.

4. The method for operating a waste heat boiler state monitoring model according to claim 3, characterized in that: When building an LSTM prediction model, the following steps are included: Selectively forget the information of the previous moment through the forget gate; The sigmoid layer of the input gate performs selective memory, and then the tanh layer is combined with the sigmoid layer to update the state of the unit; Afterwards, the old unit state is updated, and finally the final output of the LSTM unit is obtained through the output gate.

5. The method for operating a waste heat boiler state monitoring model according to claim 4, characterized in that: Methods for obtaining the optimal values ​​of hyperparameters include: The SSA algorithm is used to optimize the three hyperparameters of the LSTM layer of the LSTM model: the number of hidden units, the initial learning rate, and the L2 regularization coefficient. The optimization model of LSTM using SSA algorithm is established, and finally the optimal value of hyperparameters is obtained.

6. A method for operating a waste heat boiler state monitoring model according to claim 5, characterized in that: The method for setting the alarm threshold comprises: The actual healthy operation data of the power plant is input into the waste heat boiler status monitoring model to calculate the parameter prediction values; The 0.99 quantile of the error distribution is selected as the alarm threshold of the parameter prediction value.

7. A method for operating a waste heat boiler state monitoring model according to claim 6, characterized in that: Methods for monitoring the status of waste heat boilers include: When the absolute value of the absolute error between the parameter prediction value and the true value exceeds the alarm threshold, it indicates that a suspected failure of the waste heat boiler has occurred, and an alarm is issued to indicate the failure.

8. A waste heat boiler state monitoring model operation system, based on a waste heat boiler state monitoring model operation method according to any one of claims 1 to 7, characterized in that: It includes data collection and preprocessing module, LSTM prediction model module, SSA algorithm optimization module, PCA dimension reduction module, and status monitoring and alarm module; The data acquisition and preprocessing module is responsible for collecting the historical healthy operation data of the power plant and preprocessing the data to obtain parameters suitable for model input; The LSTM prediction model module establishes an LSTM prediction model based on the preprocessed data to predict the operating state of the waste heat boiler; The SSA algorithm optimization module uses the sparrow search algorithm to optimize the hyperparameters of the LSTM prediction model to obtain the optimal value of the hyperparameters, thereby improving the prediction accuracy of the model; The PCA dimension reduction module integrates principal component analysis technology in the model to perform dimension reduction processing on the original data, reduce redundant information and noise in the data, and improve the calculation efficiency and prediction accuracy of the model; The state monitoring and alarm module sets an alarm threshold and realizes the state monitoring of the waste heat boiler by comparing the error between the parameter prediction value and the actual value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the waste heat boiler state monitoring model operation method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for operating a waste heat boiler state monitoring model as described in any one of claims 1 to 7 are implemented.

Images