Power plant coordinated control monitoring method and system based on state deviation slow features

Abstract

The application provides a fire power generation coordinated control monitoring method and system based on state deviation slow characteristics, the method comprises the following steps: constructing a regulation and control model and a state deviation slow characteristic model of a fire power generation coordinated control system, determining a regulation and control monitoring threshold, inputting historical state data of a unit generator into the regulation and control model to obtain current state estimation data, inputting current state estimation deviation obtained according to current state real data and the current state estimation data into the state deviation slow characteristic model to obtain current state deviation slow characteristics, obtaining normal working condition change statistical indexes and real abnormal statistical indexes according to the current state deviation slow characteristics, and obtaining a coordinated control monitoring result according to the normal working condition change statistical indexes, the real abnormal statistical indexes and the regulation and control monitoring threshold. The application effectively realizes global state monitoring of the coordinated control, automatically and accurately judges normal working condition changes and real process abnormalities, and provides reliable protection for fire power generation.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for thermal power generation, and in particular to a method and system for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation. Background Technology

[0002] In the field of thermal power generation, coordinated control must ensure that the unit responds quickly to load demands while maintaining relative stability of the unit's key parameter, the inlet pressure, during load changes. It is a crucial component of the unit's automatic control system. Furthermore, in actual production applications, faults, disturbances, and other anomalies severely impact the normal operation of the boiler-turbine coordinated control process. Therefore, to ensure the safe, stable, and efficient operation of the power generation process, real-time monitoring of the overall control process and timely detection of these anomalies are of paramount importance.

[0003] In current production processes, unit generators often employ monitoring schemes that set multi-level thresholds for key variables, and the selection of key variables and thresholds largely relies on expert experience. However, during coordinated control, it is usually necessary to repeatedly change operating conditions to adjust loads to meet grid dispatch requirements. Multi-level threshold monitoring cannot distinguish between normal operating condition changes and anomalies that truly affect the characteristics of the control process, leaving operators exposed to a large number of invalid false alarms. This leads to overall failure of process monitoring, increases the burden on operators, and affects their focus on addressing genuine safety hazards in the production process.

[0004] Therefore, there is an urgent need to provide a monitoring method for coordinated control of thermal power generation that can monitor the overall state of coordinated control without requiring manual analysis by experts, and can automatically and accurately determine changes in normal operating conditions and actual process anomalies. Summary of the Invention

[0005] The purpose of this invention is to provide a monitoring method for coordinated control of thermal power generation based on the slow characteristic of state deviation. By analyzing the state variables of all coordinated control of thermal power generation using the slow characteristic of state deviation, statistical indicators of normal operating condition changes and true anomaly statistical indicators are obtained. Combined with the set monitoring thresholds for normal operating condition changes and true anomaly monitoring thresholds, the monitoring results of coordinated control are given. This method solves the application defects of existing multi-level threshold monitoring methods, realizes global state monitoring of coordinated control, and automatically and accurately judges normal operating condition changes and true process anomalies, thus providing a strong guarantee for the reliable and stable operation of thermal power generation systems.

[0006] To achieve the above objectives, it is necessary to provide a method and system for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation, addressing the aforementioned technical problems.

[0007] In a first aspect, embodiments of the present invention provide a method for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation, the method comprising the following steps:

[0008] A control model and a slow characteristic model of state deviation for a coordinated control system of thermal power generation are constructed respectively, and the corresponding control monitoring thresholds are determined; the control monitoring thresholds include normal operating condition change monitoring thresholds and true anomaly monitoring thresholds.

[0009] Historical status data of the unit is acquired, and the historical status data is input into the control model to obtain current status estimation data;

[0010] Based on the obtained current state real data and the current state estimated data, the current state estimation deviation is obtained;

[0011] The current state estimation deviation is input into the state deviation slow feature model to obtain the current state deviation slow feature, and based on the current state deviation slow feature, the normal operating condition change statistical index and the true abnormality statistical index are obtained.

[0012] Based on the statistical indicators of changes in normal operating conditions, the statistical indicators of true anomalies, and the control monitoring threshold, the coordinated control monitoring results are obtained.

[0013] Furthermore, the step of constructing the regulation model includes:

[0014] Obtain the real state data of the thermal power generation coordination and control system at a first preset number of sampling times, and construct the current state dataset.

[0015] The real state data of the second preset number of sampling times before each sampling time are obtained respectively, and the historical state dataset corresponding to the current state dataset is constructed.

[0016] Calculate the current state difference dataset and the historical state difference dataset corresponding to the current state dataset and the historical state dataset, respectively;

[0017] Based on the current state difference dataset and the historical state difference dataset, a preset black-box model is trained to obtain the control model; the preset black-box model is represented as follows:

[0018]

[0019] In the formula,

[0020] X(T)=[x(T) T ...x(T-(N-1)ΔT) T ] T

[0021]

[0022]

[0023] ΔX(T)=X(T)-X(T-DΔT)

[0024]

[0025] Where, ΔX(T) and Let x(T) and N represent the current state difference dataset and the historical state dataset, respectively; A represents the coefficient matrix of the control model; N represents the number of data points in the current state dataset and the historical state dataset collected at time T; x(T) and N(T) represent the coefficient matrix of the control model. Let represent the m-dimensional state data at time T and the state data composed of the q sampled times before time T, respectively; q represents the order of the coordinated control dynamic model; ΔT represents the sampling interval; and D represents the hysteresis characteristic of the coordinated control process.

[0026] Furthermore, the step of constructing the slow characteristic model of the state deviation of the thermal power generation coordinated control system includes:

[0027] Based on the historical state dataset and the control model, the current state estimation dataset is obtained;

[0028] Based on the current state dataset and the current state estimation dataset, a state estimation bias dataset is obtained;

[0029] Based on the state estimation bias dataset, a preset bias slow feature model is trained to obtain the state bias slow feature model; the preset bias slow feature model is expressed as:

[0030] S(T)=U(T)W

[0031] In the formula,

[0032]

[0033]

[0034] Where S(T) represents the slow bias feature; U(T) represents the state estimation bias dataset; W represents the coefficient matrix of the state bias slow feature model; X(T), and Let A represent the current state dataset, the current state estimation dataset, and the historical state dataset, respectively; A represents the coefficient matrix of the control model.

[0035] Further, the step of training a preset bias slow feature model based on the state estimation bias dataset includes:

[0036] Based on the state estimation bias dataset, the estimation bias covariance matrix is ​​obtained;

[0037] The estimated bias covariance matrix is ​​subjected to singular value decomposition to obtain the singular decomposition matrix and the eigenvalue diagonal matrix;

[0038] Based on the singular decomposition matrix and the eigenvalue diagonal matrix, the first slow eigenvalue coefficient decomposition matrix is ​​obtained; the first slow eigenvalue coefficient decomposition matrix is ​​expressed as:

[0039] Q = Λ -0.5 V T

[0040] In the formula,

[0041] C = VΛV T

[0042] C = Cov(U(T))

[0043] Where Q represents the first slow eigenvalue coefficient decomposition matrix; C represents the estimation bias covariance matrix; U(T) represents the state estimation bias dataset; V and Λ represent the singular decomposition matrix and eigenvalue diagonal matrix, respectively; Cov(·) represents covariance calculation;

[0044] Based on the first slow feature coefficient decomposition matrix and the state estimation bias dataset, a whitening matrix is ​​obtained, and based on the whitening matrix, the corresponding whitening derivative matrix is ​​obtained; the whitening matrix is ​​expressed as:

[0045] Z(T)=U(T)Q

[0046] Where Z(T) and U(T) represent the whitening matrix and the state estimation bias dataset, respectively; Q represents the first slow eigenvalue coefficient decomposition matrix;

[0047] Based on the whitening derivative matrix, the corresponding whitening covariance matrix is ​​obtained, and singular value decomposition is performed on the whitening covariance matrix to obtain the second slow eigenvalue decomposition matrix.

[0048] Based on the first slow feature coefficient decomposition matrix and the second slow feature coefficient decomposition matrix, the coefficient matrix of the state deviation slow feature model is obtained; the coefficient matrix of the state deviation slow feature model is expressed as follows:

[0049] W = PQ

[0050] Where W represents the coefficient matrix of the slow feature model of state deviation; Q and P represent the first and second slow feature coefficient decomposition matrices, respectively.

[0051] Furthermore, the regulation and monitoring threshold is expressed as:

[0052]

[0053]

[0054] in, and These represent the true anomaly monitoring threshold and the normal operating condition change monitoring threshold, respectively. This represents the χ² values ​​with degrees of freedom m and nm at a significance level of α. 2 Distribution; F m,-1,α Let f(x) represent the F-distribution with (m, nm-1) degrees of freedom at a significance level of α, where n is the number of samples in the state estimation bias dataset and m is the dimension of the variables in the state estimation bias dataset.

[0055] Furthermore, the step of obtaining the statistical indicators of normal operating condition changes and the statistical indicators of true anomalies based on the slow deviation characteristics of the current state includes:

[0056] Based on the slow deviation characteristic of the current state, the statistical index of normal operating condition change is obtained; the statistical index of normal operating condition change is expressed as:

[0057] T 2 =s(t)s(t) T

[0058] Among them, T 2 This represents the statistical indicators of changes under normal operating conditions; s(t) represents the slow deviation characteristic of the current state; (·) T Represents the transpose of a matrix;

[0059] Obtain the diagonal matrix of the slow deviation feature covariance eigenvalues ​​corresponding to the coefficient matrix of the slow deviation feature model, and obtain the true anomaly statistical index based on the derivative of the current slow deviation feature and the diagonal matrix of the slow deviation feature covariance eigenvalues; the true anomaly statistical index is expressed as:

[0060]

[0061] Among them, S 2 Indicates truly abnormal statistical indicators; Λ represents the derivative of the current state's slow deviation feature; Λ represents the diagonal matrix of the slow deviation feature covariance eigenvalues; (·) -1 This represents the inverse of a matrix.

[0062] Furthermore, the step of obtaining the coordinated control monitoring results based on the normal operating condition change statistical indicators, the true anomaly statistical indicators, and the regulation monitoring threshold includes:

[0063] Determine whether the statistical index of true anomaly is greater than the monitoring threshold of true anomaly. If so, determine that the coordinated control monitoring result is true anomaly. Otherwise, further determine whether the statistical index of normal operating condition change is greater than the monitoring threshold of normal operating condition change.

[0064] If the statistical index of normal operating condition change is greater than the monitoring threshold of normal operating condition change, the coordinated control monitoring result is determined to be a normal operating condition change; otherwise, the coordinated control monitoring result is determined to be a normal operation.

[0065] Secondly, embodiments of the present invention provide a coordinated control and monitoring system for thermal power generation based on the slow characteristics of state deviation, the system comprising:

[0066] The preprocessing module is used to construct the control model and the slow characteristic model of state deviation of the thermal power generation coordinated control system, and to determine the corresponding control monitoring thresholds; the control monitoring thresholds include normal operating condition change monitoring thresholds and true anomaly monitoring thresholds.

[0067] The state estimation module is used to acquire historical state data of the unit and input the historical state data into the control model to obtain current state estimation data;

[0068] The deviation calculation module is used to obtain the current state estimation deviation based on the acquired current state real data and the current state estimation data;

[0069] The indicator calculation module is used to input the current state estimation deviation into the state deviation slow feature model to obtain the current state deviation slow feature, and based on the current state deviation slow feature, to obtain the normal operating condition change statistical indicator and the true abnormal statistical indicator.

[0070] The result acquisition module is used to obtain the coordinated control monitoring results based on the normal operating condition change statistical indicators, the true abnormality statistical indicators, and the regulation monitoring threshold.

[0071] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0072] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0073] This application provides a method and system for monitoring coordinated control of thermal power generation based on the slow characteristic of state deviation. The method constructs a control model and a slow characteristic model of state deviation for the coordinated control system of thermal power generation. After determining the control monitoring threshold, historical state data of the acquired unit is input into the control model to obtain current state estimation data. The current state estimation deviation, obtained from the actual current state data and the current state estimation data, is input into the slow characteristic model to obtain the current state deviation slow characteristic. Based on the current state deviation slow characteristic, statistical indicators of normal operating conditions and statistical indicators of true anomalies are obtained. Finally, based on the statistical indicators of normal operating conditions, the statistical indicators of true anomalies, and the control monitoring threshold, the coordinated control monitoring results are obtained. Compared with existing technologies, this method for monitoring coordinated control of thermal power generation based on the slow characteristic of state deviation effectively achieves global state monitoring of coordinated control by using the slow characteristic of state deviation analysis on all state variables of coordinated control of thermal power generation. It automatically and accurately judges normal operating condition changes and true process anomalies, providing a strong guarantee for the reliable and stable operation of thermal power generation systems. Attached Figure Description

[0074] Figure 1 This is a schematic diagram illustrating the application scenario of the thermal power generation coordinated control and monitoring method based on the slow characteristic of state deviation in this embodiment of the invention.

[0075] Figure 2 This is a flowchart illustrating the coordinated control and monitoring method for thermal power generation based on the slow characteristic of state deviation in an embodiment of the present invention.

[0076] Figure 3 This is a schematic diagram of the structure of the thermal power generation coordinated control and monitoring system based on the slow state deviation characteristic in an embodiment of the present invention;

[0077] Figure 4 This is an internal structural diagram of the computer device in an embodiment of the present invention. Detailed Implementation

[0078] To make the objectives, technical solutions, and beneficial effects of this application clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the embodiments described below are only part of the embodiments of the present invention and are used to illustrate the present invention, but are not intended to limit the scope of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0079] Coordinated control of thermal power generation involves regulating the boiler and turbine as a unified whole, ensuring that state data (involving the steam-water system, pulverizing system, combustion system, etc.) remains in dynamic equilibrium as the power grid's load commands fluctuate drastically. However, state data often exhibits characteristics that are notoriously difficult to monitor, such as strong coupling of multiple variables, dynamic nature, and time-varying characteristics. Existing multi-threshold monitoring schemes utilize expert experience to set multiple threshold levels (high-high-high, high-high, high, low, low, low-low, low-low-low) for coordinated control of certain state variables. These thresholds remain fixed despite drastic load changes driven by power grid commands. Whether the change in coal consumption leads to operational variations or feedwater pump blockage causes anomalies, alarms are triggered, requiring significant time from experts to analyze and identify the changes and anomalies. Considering the many inconveniences of existing multi-threshold monitoring schemes for coordinated control of thermal power generation, a method for monitoring the coordinated control process is proposed. This method uses state deviation slow feature analysis on all state variables of coordinated control of thermal power generation and combines only two monitoring indicators to effectively achieve global state monitoring of coordinated control, and automatically and accurately judges changes in normal operating conditions and actual process anomalies.

[0080] The method for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation provided by this invention can be applied to... Figure 1 The terminal and server shown are described. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be a standalone server or a server cluster consisting of multiple servers. The server can use the method of this invention to perform real-time monitoring and automatic evaluation of the coordinated control process of thermal power generation, and use the obtained coordinated control monitoring results for subsequent research on the server or send them to the terminal for users to view and analyze. The following embodiments will provide a detailed description of the thermal power generation coordinated control monitoring method based on the slow characteristics of state deviation of this invention.

[0081] In one embodiment, such as Figure 2 As shown, a method for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation is provided, including the following steps:

[0082] S11. Construct the control model and the slow characteristic model of state deviation of the thermal power generation coordinated control system respectively, and determine the corresponding control monitoring thresholds; the control monitoring thresholds include the normal operating condition change monitoring threshold and the true anomaly monitoring threshold.

[0083] The control model can be understood as a coordinated control process model identified using a differential black-box method. The corresponding construction process is as follows: First, the coordinated control status data after the unit's trial operation is retrieved from the Distributed Control System (DCS) database through the unit's Safety Instrumented System (SIS) as a dataset; second, the difference in the dataset is calculated; finally, the dynamic model of the coordinated control process is obtained using a black-box identification method. Specifically, the steps for constructing the control model include:

[0084] The real state data of the thermal power generation coordinated control system at a first preset number of sampling times are obtained to construct the current state dataset; wherein, the first preset number can be selected according to the application requirements, for example, to obtain the real state data of N (31.536 million) sampling times for a period of 1 year during the trial operation or formal commissioning of the unit, so as to obtain the current state dataset X(T) as shown in equation (1):

[0085] X(T)=[x(T) T ... x(T-(N-1)ΔT) T ] T (1)

[0086] In the formula, x(T) represents the m-dimensional state data at time T. The state data may include all state variables involved in the coordinated control, such as the main steam temperature, turbine inlet pressure, and feedwater pump flow rate at time T.

[0087] The real state data of the second preset number of sampling times before each sampling time are obtained respectively, and the historical state dataset corresponding to the current state dataset is constructed. Each sampling time can be understood as any one of the N sampling times (T, T-ΔT, ..., T-(N-1)ΔT) with a sampling interval of ΔT. The state data of the second preset number of sampling times before any time are collected. The second preset number can also be selected according to the application requirements. In this embodiment, the order q of the coordinated control dynamic model is preferred. Based on the mechanism analysis, the coordinated control is usually a two-level hierarchical control system, and q is generally taken as 2. That is, the historical state dataset as shown in equation (2) is obtained.

[0088]

[0089] In the formula, The state data consists of the q sampling times preceding time T;

[0090] Calculate the current state difference dataset and the historical state difference dataset corresponding to the current state dataset and the historical state dataset, respectively; wherein the current state difference dataset and the historical state difference dataset are shown in equations (3) and (4), respectively:

[0091] ΔX(T)=X(T)-X(T-DΔT) (3)

[0092]

[0093] Where D represents the lag characteristic of the coordinated control process, and can also be selected according to the actual application requirements, preferably set to 60;

[0094] Based on the current state difference dataset and the historical state difference dataset, a preset black-box model is trained to obtain the control model; the preset black-box model is represented as follows:

[0095]

[0096] Where, ΔX(T) and Let A represent the current state difference dataset and the historical state dataset, as shown in equations (3) and (4), respectively; A represents the coefficient matrix of the control model.

[0097] The process of training the pre-set black-box model can be understood as obtaining the coefficient matrix A of the control model based on the current state difference dataset and the historical state difference dataset. After obtaining the coefficient matrix A, a stable control model is obtained, which is used for subsequent real-time coordinated control state estimation. Specific training methods can include least squares, regression, or neural networks, etc., without specific limitations here. It should be noted that the coordinated control of thermal power generation is a process control system with a large time delay. Compared with other existing black-box methods, this embodiment preferably adopts a difference-based black-box method, which can effectively overcome the lag problem of the coordinated control system and improve the accuracy and reliability of coordinated control.

[0098] The aforementioned slow characteristic model of state deviation can be understood as a slow characteristic model established based on the deviation between the actual state and the estimated state; specifically, the steps for constructing the slow characteristic model of state deviation for a thermal power generation coordinated control system include:

[0099] Based on the historical state dataset and the control model, a current state estimation dataset is obtained; wherein, the current state estimation dataset is represented as:

[0100]

[0101] In the formula, and These represent the current state estimation dataset and the historical state dataset, respectively; A represents the coefficient matrix of the control model;

[0102] Based on the current state dataset and the current state estimation dataset, a state estimation bias dataset is obtained; wherein, the state estimation bias dataset is represented as:

[0103]

[0104] In the formula, U(T) represents the state estimation bias dataset; X(T) represents the current state dataset, corresponding to... As shown in equation (6);

[0105] Based on the state estimation bias dataset, a preset bias slow feature model is trained to obtain the state bias slow feature model; the preset bias slow feature model is expressed as:

[0106] S(T)=U(T)W (8)

[0107] Where S(T) represents the slow bias feature; U(T) represents the state estimation bias dataset; and W represents the coefficient matrix of the state bias slow feature model.

[0108] The above process of training the preset bias slow feature model can be understood as the process of obtaining the coefficient matrix W of the state bias slow feature model in equation (8) offline by training the state estimation bias dataset obtained in equation (7) based on the relevant theory of slow feature analysis. Specifically, the steps of training the preset bias slow feature model based on the state estimation bias dataset include:

[0109] Based on the state estimation bias dataset, the estimation bias covariance matrix is ​​obtained; wherein, the estimation bias covariance matrix is ​​expressed as:

[0110] C = Cov(U(T)) (9)

[0111] In the formula, C represents the estimation bias covariance matrix; U(T) represents the state estimation bias dataset; Cov(·) represents the covariance calculation;

[0112] The estimated bias covariance matrix is ​​subjected to singular value decomposition to obtain a singular decomposition matrix and an eigenvalue diagonal matrix; wherein, the singular value decomposition process adopts the existing decomposition algorithm, and the decomposition result shown in equation (10) can be obtained, without any specific restrictions here;

[0113] C = VΛV T (10)

[0114] Where V and Λ represent the singular decomposition matrix and the eigenvalue diagonal matrix, respectively; C is the deviation covariance matrix obtained by equation (9);

[0115] Based on the singular decomposition matrix and the eigenvalue diagonal matrix, the first slow eigenvalue coefficient decomposition matrix is ​​obtained; the first slow eigenvalue coefficient decomposition matrix is ​​expressed as:

[0116] Q = Λ -0.5 V T (11)

[0117] Where Q represents the first slow eigenvalue decomposition matrix; V and Λ represent the singular decomposition matrix and eigenvalue diagonal matrix shown in equation (10), respectively;

[0118] Based on the first slow feature coefficient decomposition matrix and the state estimation bias dataset, a whitening matrix is ​​obtained, and based on the whitening matrix, the corresponding whitening derivative matrix is ​​obtained; the whitening matrix is ​​expressed as:

[0119] Z(T)=U(T)Q (12)

[0120] Where Z(T) and U(T) represent the whitening matrix and the state estimation bias dataset, respectively; Q represents the first slow eigenvalue coefficient decomposition matrix;

[0121] Based on the whitening derivative matrix, the corresponding whitening covariance matrix is ​​obtained, and singular value decomposition is performed on the whitening covariance matrix to obtain the second slow eigenvalue decomposition matrix; wherein, the whitening covariance matrix is ​​expressed as:

[0122]

[0123] In the formula, D z Represent the whitening covariance matrix; Let denote the whitening derivative matrix, which is the derivative of the whitening matrix shown in equation (12);

[0124] Singular value decomposition of the whitening covariance matrix shown in equation (13) yields the second slow eigenvalue decomposition matrix P shown in equation (14):

[0125] D z =PΩP T (14)

[0126] Where Ω is the covariance matrix with whitening D z The corresponding eigenvalue diagonal matrix;

[0127] Based on the first slow feature coefficient decomposition matrix and the second slow feature coefficient decomposition matrix, the coefficient matrix of the state deviation slow feature model is obtained; the coefficient matrix of the state deviation slow feature model is expressed as follows:

[0128] W = PQ (15)

[0129] Where W represents the coefficient matrix of the slow feature model of state deviation; Q and P represent the coefficient decomposition matrices of the first and second slow features, respectively.

[0130] The coefficient matrix W of the slow characteristic model of state deviation is obtained by the above equation (15), and the construction of the slow characteristic model of state deviation is completed, which can be used for subsequent effective slow characteristic analysis of the state variables of coordinated control.

[0131] Furthermore, considering the numerous inconveniences of existing multi-threshold monitoring methods, this embodiment preferably sets two monitoring thresholds that can effectively distinguish between normal operating condition changes and truly abnormal situations. Specifically, the control monitoring thresholds are expressed as follows:

[0132]

[0133]

[0134] in, and These represent the true anomaly monitoring threshold and the normal operating condition change monitoring threshold, respectively. This represents the χ² values ​​with degrees of freedom m and nm at a significance level of α. 2 Distribution; F m,n-m-1,α Let f(x) represent the F-distribution with (m, nm-1) degrees of freedom at a significance level of α, where n is the number of samples in the state estimation bias dataset and m is the dimension of the variables in the state estimation bias dataset.

[0135] S12. Obtain historical state data of the unit and input the historical state data into the control model to obtain current state estimation data; wherein, the historical state data can be understood as a historical state vector composed of state variables from q sampling times prior to monitoring time t read from the DCS database through the SIS system of the unit during the monitoring process. It can be represented as:

[0136]

[0137] Where q represents the order of the coordinated control dynamic model, which is generally taken as 2 as mentioned above; Δt represents the sampling interval of the SIS system from the DCS database, which is usually 1s; x(t-Δt) represents the m-dimensional state data of the SIS system with sampling time t-Δt.

[0138] The current state estimation data can be understood as the estimated value of the coordinated control state at the monitoring time, obtained based on historical state data and the aforementioned established control model. It can be represented as:

[0139]

[0140] in, Equation (18) gives the historical state data, and A represents the coefficient matrix of the regulation model trained based on equation (5);

[0141] S13. Based on the obtained current state real data and the current state estimated data, the current state estimated deviation is obtained; wherein, the current state real data can be understood as the real state x(t) at monitoring time t read from the DCS database through the SIS system of the unit during the monitoring process; the difference between x(t) and the current state estimated data obtained by equation (19) is used to obtain the deviation u(t) shown in equation (20):

[0142]

[0143] S14. Input the current state estimation deviation into the state deviation slow feature model to obtain the current state deviation slow feature, and based on the current state deviation slow feature, obtain the normal operating condition change statistical index and the true anomaly statistical index; wherein, the current state deviation slow feature is represented as s(t):

[0144] s(t)=u(t)W (21)

[0145] Where u(t) is the current state estimation deviation obtained based on equation (20), and W represents the coefficient matrix of the slow feature model of the state deviation obtained based on equations (7)-(15);

[0146] Based on the current state deviation slow characteristic obtained from equation (21), two statistical indicators for monitoring and analysis can be obtained; specifically, the steps of obtaining the normal operating condition change statistical indicator and the true abnormality statistical indicator based on the current state deviation slow characteristic include:

[0147] Based on the slow deviation characteristic of the current state, the statistical index of normal operating condition change is obtained; the statistical index of normal operating condition change is expressed as:

[0148] T 2 =s(t)s(t) T

[0149] Among them, T 2 This represents the statistical indicators of changes under normal operating conditions; s(t) represents the slow deviation characteristic of the current state; (·) T Represents the transpose of a matrix;

[0150] Obtain the diagonal matrix of the slow deviation feature covariance eigenvalues ​​corresponding to the coefficient matrix of the slow deviation feature model, and obtain the true anomaly statistical index based on the derivative of the current slow deviation feature and the diagonal matrix of the slow deviation feature covariance eigenvalues; the true anomaly statistical index is expressed as:

[0151]

[0152] Among them, S 2 Indicates truly abnormal statistical indicators; Λ represents the derivative of the current state's slow deviation feature; Λ represents the diagonal matrix of the slow deviation feature covariance eigenvalues; (·) -1 This represents the inverse of a matrix.

[0153] S15. Based on the statistical indicators of changes in normal operating conditions, the statistical indicators of true anomalies, and the control monitoring threshold, obtain the coordinated control monitoring results.

[0154] Specifically, the step of obtaining the coordinated control monitoring results based on the normal operating condition change statistical indicators, the true anomaly statistical indicators, and the regulation monitoring threshold includes:

[0155] Determine whether the statistical index of true anomaly is greater than the monitoring threshold of true anomaly. If so, determine that the coordinated control monitoring result is true anomaly. Otherwise, further determine whether the statistical index of normal operating condition change is greater than the monitoring threshold of normal operating condition change.

[0156] If the statistical index of normal operating condition change is greater than the monitoring threshold of normal operating condition change, the coordinated control monitoring result is determined to be a normal operating condition change; otherwise, the coordinated control monitoring result is determined to be a normal operation.

[0157] The analytical logic for monitoring and coordinating the control process based on statistical indicators and their thresholds is as follows:

[0158] 1) As long as The monitoring program will then indicate that the monitoring result is a genuine anomaly and issue an alarm "An anomaly has occurred in the coordinated control process." Power plant operators should immediately contact experts to investigate the cause of the anomaly in order to restore normal and safe operation.

[0159] 2) When and If the monitoring program indicates that the monitoring result is a normal change in operating conditions, it will issue a prompt "Change in unit operating conditions". The power plant operator should check whether it is a common operating condition of the unit. If not, the possible cause should be investigated and recorded at an appropriate time.

[0160] 3) In other cases, it indicates that the unit is operating normally, and the monitoring program will not issue alarms or provide any prompts.

[0161] This application embodiment constructs a control model and a state deviation slow feature model for a thermal power generation coordinated control system. After determining the control monitoring threshold, it inputs the acquired historical state data of the unit into the control model to obtain the current state estimation data. The current state estimation deviation obtained based on the acquired current state real data and current state estimation data is input into the state deviation slow feature model to obtain the current state deviation slow feature. Based on the current state deviation slow feature, it obtains statistical indicators of normal operating condition changes and statistical indicators of true anomalies. Based on the statistical indicators of normal operating condition changes, statistical indicators of true anomalies, and control monitoring threshold, it obtains the coordinated control monitoring results. This method can use state deviation slow feature analysis on all state variables of thermal power generation coordinated control, effectively realize global state monitoring of coordinated control, automatically and accurately judge normal operating condition changes and true process anomalies, and provide strong guarantee for the reliable and stable operation of thermal power generation systems.

[0162] In one embodiment, such as Figure 3 As shown, a coordinated control and monitoring system for thermal power generation based on the slow characteristics of state deviation is provided. The system includes:

[0163] Preprocessing module 1 is used to construct the control model and the slow characteristic model of state deviation of the thermal power generation coordinated control system, and determine the corresponding control monitoring thresholds; the control monitoring thresholds include normal operating condition change monitoring thresholds and true anomaly monitoring thresholds.

[0164] State estimation module 2 is used to acquire historical state data of the unit and input the historical state data into the control model to obtain current state estimation data;

[0165] Deviation calculation module 3 is used to obtain the current state estimation deviation based on the acquired current state real data and the current state estimation data;

[0166] The index calculation module 4 is used to input the current state estimation deviation into the state deviation slow feature model to obtain the current state deviation slow feature, and to obtain the normal working condition change statistical index and the true abnormal statistical index based on the current state deviation slow feature.

[0167] Result acquisition module 5 is used to obtain coordinated control monitoring results based on the normal operating condition change statistical indicators, the true abnormality statistical indicators, and the regulation monitoring threshold.

[0168] Specific limitations regarding the coordinated control and monitoring system for thermal power generation based on the slow characteristic of state deviation can be found in the limitations of the coordinated control and monitoring method for thermal power generation based on the slow characteristic of state deviation mentioned above, and will not be repeated here. Each module in the aforementioned coordinated control and monitoring system for thermal power generation based on the slow characteristic of state deviation can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0169] Figure 4 An internal structural diagram of a computer device is shown in one embodiment. This computer device may specifically be a terminal or a server. Figure 4 As shown, the computer device includes a processor, memory, network interface, display, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a coordinated control and monitoring method for thermal power generation based on the slow characteristics of state deviation. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device casing, or an external keyboard, touchpad, or mouse.

[0170] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computing devices may include more or fewer components than those shown in the figure, or combine certain components, or have the same component arrangement.

[0171] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0172] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0173] In summary, the present invention provides a method and system for monitoring coordinated control of thermal power generation based on the slow characteristic of state deviation. The method constructs a control model and a slow characteristic model for the coordinated control system of thermal power generation. After determining the control monitoring threshold, it inputs the historical state data of the acquired unit units into the control model to obtain the current state estimation data. The current state estimation deviation, obtained from the actual current state data and the current state estimation data, is input into the slow characteristic model to obtain the current state deviation slow characteristic. Based on the current state deviation slow characteristic, it obtains statistical indicators of normal operating condition changes and statistical indicators of true anomalies. Finally, based on the statistical indicators of normal operating condition changes, statistical indicators of true anomalies, and the control monitoring threshold, it obtains the coordinated control monitoring results. This method, by using the slow characteristic of state deviation analysis on all state variables of coordinated control of thermal power generation, solves the application defects of existing multi-level threshold monitoring methods, effectively realizes global state monitoring of coordinated control, and automatically and accurately judges normal operating condition changes and true process anomalies, providing a strong guarantee for the reliable and stable operation of thermal power generation systems.

[0174] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0175] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for coordinated control and monitoring of thermal power generation based on the slow characteristic of state deviation, characterized in that, The method includes the following steps: A control model and a slow characteristic model of state deviation for a coordinated control system of thermal power generation are constructed respectively, and the corresponding control monitoring thresholds are determined; the control monitoring thresholds include normal operating condition change monitoring thresholds and true anomaly monitoring thresholds. Historical status data of the unit is acquired, and the historical status data is input into the control model to obtain current status estimation data; Based on the obtained current state real data and the current state estimated data, the current state estimation deviation is obtained; The current state estimation deviation is input into the state deviation slow feature model to obtain the current state deviation slow feature, and based on the current state deviation slow feature, the normal operating condition change statistical index and the true abnormality statistical index are obtained. Based on the statistical indicators of changes in normal operating conditions, the statistical indicators of true anomalies, and the control monitoring threshold, the coordinated control monitoring results are obtained. The steps for constructing the regulation model include: Obtain the real state data of the thermal power generation coordination and control system at a first preset number of sampling times, and construct the current state dataset. Obtain the real state data of the second preset number of sampling times before each sampling time, and construct the historical state dataset corresponding to the current state dataset. Calculate the current state difference dataset and the historical state difference dataset corresponding to the current state dataset and the historical state dataset, respectively; Based on the current state difference dataset and the historical state difference dataset, a preset black box model is trained to obtain the control model; The steps for constructing a slow characteristic model of state deviation in a coordinated control system for thermal power generation include: Based on the historical state dataset and the control model, the current state estimation dataset is obtained; Based on the current state dataset and the current state estimation dataset, a state estimation bias dataset is obtained; Based on the state estimation bias dataset, a preset bias slow feature model is trained to obtain the state bias slow feature model; the preset bias slow feature model is expressed as: In the formula, in, This indicates a slow deviation characteristic; represents the state estimation bias dataset; W represents the coefficient matrix of the state bias slow feature model; , and Let A represent the current state dataset, the current state estimation dataset, and the historical state dataset, respectively; A represents the coefficient matrix of the control model.

2. The method for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation as described in claim 1, characterized in that, The preset black box model is represented as follows: In the formula, in, and These represent the current state difference dataset and the historical state difference dataset, respectively; A represents the coefficient matrix of the control model; N represents the number of data points in the current state dataset and the historical state dataset collected at time T. and Let represent the m-dimensional state data at time T and the state data composed of the q sampled times preceding time T, respectively; q represents the order of the coordinated control dynamic model. represents the sampling interval; D represents the hysteresis characteristic of the coordinated control process; This represents the transpose of a matrix.

3. The method for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation as described in claim 1, characterized in that, The step of training a preset bias slow feature model based on the state estimation bias dataset includes: Based on the state estimation bias dataset, the estimation bias covariance matrix is ​​obtained; The estimated bias covariance matrix is ​​subjected to singular value decomposition to obtain the singular decomposition matrix and the eigenvalue diagonal matrix; Based on the singular decomposition matrix and the eigenvalue diagonal matrix, the first slow eigenvalue coefficient decomposition matrix is ​​obtained; the first slow eigenvalue coefficient decomposition matrix is ​​expressed as: In the formula, Where Q represents the first slow eigenvalue coefficient decomposition matrix; C represents the estimation bias covariance matrix; This represents the state estimation bias dataset; and Let represent the singular decomposition matrix and the eigenvalue diagonal matrix, respectively; Indicates the calculation of covariance; Represents the transpose of a matrix; Based on the first slow feature coefficient decomposition matrix and the state estimation bias dataset, a whitening matrix is ​​obtained, and based on the whitening matrix, the corresponding whitening derivative matrix is ​​obtained; the whitening matrix is ​​expressed as: in, and Let represent the whitening matrix and the state estimation bias dataset, respectively; Q represents the first slow eigenvalue coefficient decomposition matrix. Based on the whitening derivative matrix, the corresponding whitening covariance matrix is ​​obtained, and singular value decomposition is performed on the whitening covariance matrix to obtain the second slow eigenvalue decomposition matrix. Based on the first slow feature coefficient decomposition matrix and the second slow feature coefficient decomposition matrix, the coefficient matrix of the state deviation slow feature model is obtained; the coefficient matrix of the state deviation slow feature model is expressed as follows: in, represents the coefficient matrix of the slow characteristic model of state deviation; Q and P represent the coefficient decomposition matrices of the first and second slow characteristics, respectively.

4. The method for coordinated control and monitoring of thermal power generation based on the slow characteristic of state deviation as described in claim 1, characterized in that, The regulation and monitoring threshold is expressed as: in, and These represent the true anomaly monitoring threshold and the normal operating condition change monitoring threshold, respectively. Indicates the significance level as With lower degrees of freedom m and nm distributed; Indicates the significance level as With lower degrees of freedom (m, nm-1) The distribution is given by n, where n is the number of samples in the state estimation bias dataset, and m is the dimension of the variables in the state estimation bias dataset.

5. The method for coordinated control and monitoring of thermal power generation based on the slow characteristic of state deviation as described in claim 1, characterized in that, The steps for obtaining the statistical indicators of normal operating condition changes and the statistical indicators of true anomalies based on the slow deviation characteristics of the current state include: Based on the slow deviation characteristic of the current state, the statistical index of normal operating condition change is obtained; the statistical index of normal operating condition change is expressed as: in, Indicates statistical indicators of changes in normal operating conditions; This indicates a slow deviation from the current state. Represents the transpose of a matrix; Obtain the diagonal matrix of the slow deviation feature covariance eigenvalues ​​corresponding to the coefficient matrix of the slow deviation feature model, and obtain the true anomaly statistical index based on the derivative of the current slow deviation feature and the diagonal matrix of the slow deviation feature covariance eigenvalues; the true anomaly statistical index is expressed as: in, Indicates truly abnormal statistical indicators; The derivative representing the slow characteristic of the current state deviation; The diagonal matrix represents the covariance eigenvalues ​​of the bias slow feature; This represents the inverse of a matrix.

6. The method for coordinated control and monitoring of thermal power generation based on the slow characteristics of state deviation as described in claim 1, characterized in that, The step of obtaining the coordinated control monitoring results based on the normal operating condition change statistical indicators, the true anomaly statistical indicators, and the regulation monitoring threshold includes: Determine whether the statistical index of true anomaly is greater than the monitoring threshold of true anomaly. If so, determine that the coordinated control monitoring result is true anomaly. Otherwise, further determine whether the statistical index of normal operating condition change is greater than the monitoring threshold of normal operating condition change. If the statistical index of normal operating condition change is greater than the monitoring threshold of normal operating condition change, the coordinated control monitoring result is determined to be a normal operating condition change; otherwise, the coordinated control monitoring result is determined to be a normal operation.

7. A coordinated control and monitoring system for thermal power generation based on the slow characteristic of state deviation, characterized in that, The system employs the thermal power generation coordinated control and monitoring method based on the slow characteristics of state deviation as described in claim 1, wherein the system comprises: The preprocessing module is used to construct the control model and the slow characteristic model of state deviation of the thermal power generation coordinated control system, and to determine the corresponding control monitoring thresholds; the control monitoring thresholds include normal operating condition change monitoring thresholds and true anomaly monitoring thresholds. The state estimation module is used to acquire historical state data of the unit and input the historical state data into the control model to obtain current state estimation data; The deviation calculation module is used to obtain the current state estimation deviation based on the acquired current state real data and the current state estimation data; The indicator calculation module is used to input the current state estimation deviation into the state deviation slow feature model to obtain the current state deviation slow feature, and based on the current state deviation slow feature, to obtain the normal operating condition change statistical indicator and the true abnormal statistical indicator. The result acquisition module is used to obtain the coordinated control monitoring results based on the normal operating condition change statistical indicators, the true abnormality statistical indicators, and the regulation monitoring threshold.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images