A flue gas volume error soft correction method and system based on multi-source data fusion

Abstract

The application discloses a flue gas volume error soft correction method and system based on multi-element data fusion, and relates to the technical field of soft correction of flue gas volume error. The application comprises the following steps: screening of correlation coefficients of flue gas volume and establishment of a soft measurement model of flue gas volume based on multi-element data fusion; obtaining of flue gas volume by using a mapping model and fusing of the flue gas volume into soft measurement to predict flue gas volume; continuous difference between measured flue gas volume and soft measurement predicted flue gas volume; determination of the drift of the flue gas volume sensor according to the change trend of the difference value to realize error soft correction of the flue gas volume sensor. The application can realize soft calibration of flue gas volume and improve the accuracy of sensor measurement of flue gas volume by determining the drift of the measured flue gas volume sensor and applying the drift to error compensation of the sensor.

Description

Technical Field

[0001] This invention relates to the field of thermal testing technology, specifically to a method and system for soft correction of flue gas volume error through multi-source data fusion. Background Technology

[0002] Flue gas flow rate in coal-fired power units is not only a crucial parameter in environmental monitoring data but also a critical parameter affecting the safe and economical operation of power plants. The measurement of flue gas flow rate in coal-fired power units is primarily accomplished using flue gas flow sensors installed in the ducts. However, due to their inherent characteristics or the influence of harsh environmental conditions (temperature, humidity, etc.), flow sensors are prone to drift, and this drift increases over time, leading to significant deviations in flue gas measurement accuracy. When the sensor is not calibrated, the deviation is substantial and worsens over time. Current sensor calibration methods involve disassembling and recalibrating the sensor, but this requires unit shutdown, wasting considerable manpower and time. Therefore, achieving accurate soft correction of flue gas flow rate errors has become a critical technical problem urgently needing to be solved in the flue gas flow rate detection technology of coal-fired power units. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a soft calibration method and system for flue gas volume error by multi-data fusion, which addresses the above-mentioned problems in the prior art. The present invention determines the drift of the measured flue gas volume sensor and applies it to the sensor error compensation, thereby realizing soft calibration of flue gas volume and improving the accuracy of the sensor in measuring the exhaust gas volume.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] A soft correction method for flue gas flow error through multi-source data fusion includes:

[0006] S101, calculate the correlation coefficient between each parameter variable and the flue gas volume in the historical data sample, and finally retain the parameter variables with the highest correlation coefficient as the associated variables of the flue gas volume.

[0007] S102, establish a multivariate data fusion soft measurement model of air volume based on the correlation variables of flue air volume, wherein the multivariate data fusion soft measurement model of air volume includes a mapping model between each correlation variable and flue air volume;

[0008] S103, using the values ​​of the correlated variables measured under specified working conditions, the smoke exhaust volume is obtained by using a mapping model, and all the obtained smoke exhaust volumes are merged into a soft measurement prediction of the flue gas volume;

[0009] S104: The measured flue gas volume of the flue gas volume sensor at the same time under specified operating conditions is continuously differentially analyzed with the flue gas volume predicted by the soft measurement method to obtain the difference value between the measured flue gas volume and the flue gas volume predicted by the soft measurement method. The drift of the flue gas volume sensor is determined according to the trend of the difference value. The drift of the flue gas volume sensor is applied to the error compensation of the flue gas volume sensor, thereby realizing the soft correction of the error of the flue gas volume sensor.

[0010] Optionally, the function expression for calculating the correlation coefficient in step S101 is:

[0011]

[0012] In the above formula, r is the correlation coefficient, and X i Let X be the i-th sample value of parameter variable X. Let X be the mean of the parameter variable, and Y be the mean of the parameter variable. i Let Y be the i-th sample value of the flue gas volume. Let Y be the mean value of the flue gas volume, and n be the number of historical data samples.

[0013] Optionally, the mapping model between the correlated variable and the flue gas volume in step S102 is a univariate linear regression model between the correlated variable and the flue gas volume, and the functional expression of the univariate linear regression model is:

[0014] y=λ0+λ1x+ε ε~N(0,σ 2 ),

[0015] In the above formula, y represents the flue gas volume, x represents the related variables, and λ0, λ1, and σ are the variables. 2 Let ε be a parameter, and let ε be a distribution following N(0, σ). 2 The random variables λ0, λ1 and σ are given by σ. 2 estimator as well as The result is obtained through inverse regression analysis using the following formula:

[0016]

[0017]

[0018]

[0019] In the above formula, Let y be the average value of the flue gas volume. Let x be the mean of the related variable x. i Let y be the i-th sample value of parameter variable x. i Let y be the i-th sample value of the flue gas volume y, and n be the number of historical data samples.

[0020] Optionally, the mapping model between the correlated variable and the flue gas volume in step S102 is an exponential regression model between the correlated variable and the flue gas volume, and the functional expression of the exponential regression model is:

[0021] ln(y) = nln(x) + ln(k),

[0022] In the above formula, y is the flue gas volume, x is the related variable, n and k are parameters, and ln is the natural logarithm function;

[0023]

[0024]

[0025] In the above formula, Let y be the average value of the flue gas volume. Let x be the mean of the related variable x. i Let y be the i-th sample value of parameter variable x. i Let y be the i-th sample value of the flue gas volume y, and n be the number of historical data samples.

[0026] Optionally, in step S103, all the obtained exhaust air volumes are merged into a function expression for the soft measurement prediction of the flue gas volume, which is:

[0027]

[0028] In the above formula, Y z To predict flue gas volume using soft measurement, Let Y1, Y2, ..., Yn be the variances of n related variables. n The exhaust air volume is obtained by using a mapping model with the values ​​of n related variables measured under specified operating conditions.

[0029] Optionally, the calculation function expression for continuously differencing the measured flue gas volume of the flue gas volume sensor and the predicted flue gas volume of the soft sensor at the same time under the specified operating conditions in step S104 is as follows:

[0030] f t =y t -s t ,

[0031] In the above formula, f t Let y be the first difference value at time t. t For the soft measurement prediction of flue gas volume at time t, s t Let t be the measured flue air volume of the flue air volume sensor at time t.

[0032] Optionally, determining the drift of the flue gas flow sensor based on the changing trend of the difference value in step S104 includes: firstly, calculating the second-order difference value of the difference value according to the following formula:

[0033] E t =f t -2f t-1 +f t-2 ,

[0034] In the above formula, E t Let f be the second difference value at time t. t-1 f is the first-order difference value at time t-1. t-2 The first-order difference value at time t-2 is given; then the second-order difference value E at time t is given. t The drift of the flue gas flow sensor at time t.

[0035] Optionally, before step S101, the method further includes preprocessing the historical data samples obtained from the DCS system of the thermal power unit to remove gross values ​​from the historical data samples. The preprocessing to remove gross values ​​from the data samples of the DCS system of the thermal power unit includes: calculating the mean and standard deviation for each parameter variable. If the absolute value of the difference between the data of a certain parameter variable and the mean exceeds three times the standard deviation, then the data is removed as a gross value in the historical data sample.

[0036] Furthermore, the present invention also provides a multi-data fusion flue gas volume error soft correction system, including a microprocessor and a memory interconnected thereto, wherein the microprocessor is programmed or configured to execute the multi-data fusion flue gas volume error soft correction method.

[0037] Furthermore, the present invention also provides a computer-readable storage medium storing a computer program that is programmed or configured by a microprocessor to perform the multi-data fusion flue gas volume error soft correction method.

[0038] Compared with the prior art, the present invention has the following main advantages:

[0039] 1. The method of this invention is based on the principles of soft measurement and continuous difference to perform soft calibration on the flue gas volume sensor, thereby avoiding excessive uncertainty in the flue gas volume measured by the sensor and the measurement results not meeting the actual needs of engineering.

[0040] 2. The method of this invention enables online calibration, which saves labor and time costs compared to offline calibration.

[0041] 3. The method of the present invention determines the drift of the sensor and applies the drift to the error compensation of the sensor, thereby achieving accurate measurement of the flue air volume.

[0042] 4. The method of the present invention is simple in principle, clear in steps, and easy to implement. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the basic process of the method in Embodiment 1 of the present invention.

[0044] Figure 2 This is a schematic diagram of the complete process of the method in Embodiment 1 of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The purpose of the present invention is to provide a method for soft correction of flue gas volume error in power units. That is, a flue gas volume prediction model for thermal power units is established using a soft measurement method to predict the flue gas volume, and a continuous difference method is used to continuously differentiate the measured gas volume and the soft measurement predicted gas volume under specific operating conditions. By judging the trend of the difference value, the drift of the measured gas volume sensor is determined. Finally, the drift is applied to the error compensation of the sensor, thereby achieving soft calibration of the flue gas volume sensor. To make the above-mentioned objectives, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to specific embodiments.

[0046] Example 1:

[0047] like Figure 1 and Figure 2 As shown, the soft correction method for flue gas volume error based on multi-source data fusion in this embodiment includes:

[0048] S101, calculate the correlation coefficient between each parameter variable and the flue gas volume in the historical data sample, and finally retain the parameter variables with the highest correlation coefficient as the associated variables of the flue gas volume.

[0049] S102, establish a multivariate data fusion soft measurement model of air volume based on the correlation variables of flue air volume, wherein the multivariate data fusion soft measurement model of air volume includes a mapping model between each correlation variable and flue air volume;

[0050] S103, using the values ​​of the correlated variables measured under specified working conditions, the smoke exhaust volume is obtained by using a mapping model, and all the obtained smoke exhaust volumes are merged into a soft measurement prediction of the flue gas volume;

[0051] S104: The measured flue gas volume of the flue gas volume sensor at the same time under specified operating conditions is continuously differentially analyzed with the flue gas volume predicted by the soft measurement method to obtain the difference value between the measured flue gas volume and the flue gas volume predicted by the soft measurement method. The drift of the flue gas volume sensor is determined according to the trend of the difference value. The drift of the flue gas volume sensor is applied to the error compensation of the flue gas volume sensor, thereby realizing the soft correction of the error of the flue gas volume sensor.

[0052] This embodiment's method includes calculating the correlation coefficients between various parameter variables and flue gas volume in historical data samples. Finally, the parameter variables with the highest correlation coefficients are retained as the associated variables for flue gas volume. A multivariate data fusion soft measurement model for gas volume is established based on these associated variables. This model includes a mapping model between each associated variable and the flue gas volume. The measured values ​​of the associated variables under specified operating conditions are used to obtain the exhaust gas volume using the mapping model. All obtained exhaust gas volumes are then fused into a soft measurement predicted flue gas volume. This invention achieves soft calibration of the flue gas volume by determining the drift of the measured flue gas volume sensor and applying it to sensor error compensation, thereby improving the accuracy of sensor measurements of exhaust gas volume. This embodiment's method includes calculating the correlation coefficients between various parameter variables and flue gas volume in historical data samples. Finally, the parameter variables with the highest correlation coefficients are retained as the associated variables for flue gas volume. The correlation coefficient method is used to quickly process auxiliary parameter variables, satisfying the need for variable selection and offering simplicity and speed.

[0053] In this embodiment, based on the mechanism analysis of flue gas emissions, the following 18 parameters were selected as candidates from historical data samples obtained from the DCS system of the thermal power unit: unit load, air leakage coefficient, coal feed rate, main steam flow rate, main steam temperature, main steam pressure, feedwater flow rate, flue gas temperature, air supply volume, forced draft fan current, induced draft volume, induced draft fan current, motor current, forced draft fan outlet air pressure, induced draft fan outlet air pressure, differential pressure between flue gas inlet and outlet of the dust collector, and furnace negative pressure. Due to the difficulty in collecting some data, 10 related variables were retained: flue gas pressure at the outlets of induced draft fans A and B, unit load, main steam flow rate, motor current, air pressure at the outlets of forced draft fans A and B, and differential pressure between flue gas inlet and outlet of dust collectors A and B. In this embodiment, before step S101, preprocessing is performed on the historical data samples obtained from the DCS system of the thermal power unit to remove gross values ​​from the historical data samples. Through data preprocessing, the actual production data is restored as much as possible, so that the performance of the air volume soft measurement model fused by multi-data is not affected by noise.

[0054] Removing outliers (gross values) from the DCS system data samples of thermal power units can be implemented in any way required. For example, as a preferred implementation, the preprocessing in this embodiment to remove outliers from the DCS system data samples of thermal power units includes: calculating the mean and standard deviation for each parameter variable; if the absolute value of the difference between a parameter variable's data and the mean exceeds three times the standard deviation, then that data is removed as an outlier from the historical data sample. The expression for calculating the standard deviation is as follows:

[0055]

[0056] In the above formula, σ is the standard deviation, and X i Let X be the i-th sample value of parameter variable X. Let X be the mean of the parameter variable, and m be the number of historical data samples before preprocessing. In this embodiment, the preprocessing method described above uses the 3σ criterion (Laida criterion) to remove outliers in the dataset. However, this criterion is not absolute and must meet its assumptions, namely, that the dataset contains only random errors and the data volume is large enough to determine an interval according to a certain probability.

[0057] In this embodiment, the function expression for calculating the correlation coefficient in step S101 is:

[0058]

[0059] In the above formula, r is the correlation coefficient, and X i Let X be the i-th sample value of parameter variable X. Let X be the mean of the parameter variable, and Y be the mean of the parameter variable. i Let Y be the i-th sample value of the flue gas volume. Let Y be the mean of the flue gas volume, and n be the number of historical data samples, where the mean is... The expression for the computation function is:

[0060]

[0061] mean The calculation method is similar, so it will not be elaborated here. The correlation coefficient method is a classic data analysis method that is very simple and clear, quickly and directly showing the relationship between two sets of variables, so that the parameter variable with the highest correlation coefficient can be easily selected as the auxiliary correlation variable. In this embodiment, the correlation coefficient value of [0, 0.3] is defined as extremely weak correlation, which can be regarded as no correlation; the correlation coefficient value of (0.3, 0.5] is defined as low correlation; the correlation coefficient value of (0.5, 0.7] is defined as moderate correlation; and the correlation coefficient value of (0.7, 1] is defined as high correlation. Therefore, the parameter variables with a correlation coefficient value of (0.7, 1) are finally used as the correlation variables of flue gas volume.

[0062] It should be noted that the mapping model between the associated variables and the flue gas volume in step S102 can adopt the required function mapping model or machine learning model as needed.

[0063] The sum of the calculated value of the measured quantity y as a real function of the detected value x and the uncertainty ε (error) that occurs during detection and follows a standard normal distribution:

[0064]

[0065] In the above formula, For the prediction result, y represents the flue gas volume, x represents the related variable, and φ(·) represents the real function; λ i ,i=0,1,2,...,m and σ 2 All parameters are undetermined and can be determined by regression analysis. As an optional implementation, in step S102 of this embodiment, the mapping model between the correlated variables and the flue gas volume is a univariate linear regression model. The functional expression of the univariate linear regression model is:

[0066] y=λ0+λ1x+ε ε~N(0,σ 2 ),

[0067] In the above formula, y represents the flue gas volume, x represents the related variables, and λ0, λ1, and σ are the variables. 2 Let ε be a parameter, and let ε be a distribution following N(0, σ). 2 The random variables (uncertainties) of λ0, λ1, and σ are given by λ0, λ1, and σ. 2 estimator as well as The result is obtained through inverse regression analysis using the following formula:

[0068]

[0069]

[0070]

[0071] In the above formula, Let y be the average value of the flue gas volume. Let x be the mean of the related variable x. i Let y be the i-th sample value of parameter variable x. i Let be the i-th sample value of the flue gas volume y, and n be the number of historical data samples, and we have:

[0072]

[0073] In step S103 of this embodiment, when all the obtained exhaust air volumes are fused into a soft-measurement predicted flue gas volume, this embodiment uses a weighted data fusion method to fuse the regression models between each associated variable and the flue gas volume. That is, the weighted data fusion method is used to fuse the multiple established regression models between the flue gas volume and the associated variables into a single calculation formula for the flue gas volume and the associated variables, thus obtaining the soft-measurement predicted flue gas volume. Specifically, the function expression for fusing all the obtained exhaust air volumes into the soft-measurement predicted flue gas volume in step S103 is as follows:

[0074]

[0075] In the above formula, Y z To predict flue gas volume using soft measurement, Let Y1, Y2, ..., Yn be the variances of n related variables. n The exhaust air volume is obtained by using a mapping model with the values ​​of n related variables measured under specified operating conditions. As can be seen from the above formula, the measurement data with large variance is assigned a smaller weight, while the data with small variance is assigned a larger weight. Therefore, this data fusion method can obtain more reliable results than the arithmetic mean. Thus, the above weighted fusion method has the characteristics of optimality, unbiasedness, and minimum standard deviation.

[0076] In step S104 of this embodiment, the calculation function expression for continuously differentiating the measured flue gas volume of the flue gas volume sensor at the same time under specified operating conditions with the flue gas volume predicted by the soft sensor is as follows:

[0077] f t =y t -s t ,

[0078] In the above formula, f t Let y be the first difference value at time t. t For the soft measurement prediction of flue gas volume at time t, s t Let t be the measured flue gas volume from the flue gas volume sensor. Continuous difference is performed between the measured flue gas volume and the predicted flue gas volume under specific operating conditions. This is achieved by continuously differencing the measured flue gas volume from the sensor at the same time point with the obtained predicted flue gas volume from the soft measurement system.

[0079] The measured flue air volume is:

[0080] S t = [s1, s2, s3, ... s i , …s t-1 s t ],

[0081] The predicted flue gas volume using soft measurement is:

[0082] Y z = [y1, y2, y3, ... y i , ...y t-1 y t ],

[0083] By using the first-order difference method, the set of first-order difference values ​​between the measured flue gas volume and the soft-measurement predicted flue gas volume is obtained as follows:

[0084] F t = [f1, f2, f3, ... f i , ...f t-1 f t ].

[0085] The trend of the difference value is positively correlated with the drift of the flue gas flow sensor. Therefore, the drift of the flue gas flow sensor can be determined based on the trend of the difference value according to the positive correlation function. For example, as a simple implementation, in step S104 of this embodiment, determining the drift of the flue gas flow sensor based on the trend of the difference value includes: firstly, calculating the second-order difference value of the difference value according to the following formula:

[0086] E t =f t -2f t-1 +f t-2 ,

[0087] In the above formula, E t Let f be the second difference value at time t. t-1 f is the first-order difference value at time t-1. t-2 The first-order difference value at time t-2 is given; then the second-order difference value E at time t is given. t The drift of the flue gas flow sensor at time t. The second-order difference value E at time t. t The magnitude of the value reflects the changing trend of the difference between the predicted flue gas volume and the measured flue gas volume. When E t When E > 0, the difference between the predicted air volume and the measured air volume increases; when E t When E < 0, the difference between the predicted and measured air volume by the soft sensor decreases; when E t When the value is 0, the difference between the predicted air volume and the actual air volume remains unchanged. In this embodiment, a continuous difference method is used to continuously differentiate the actual air volume and the predicted air volume under specific operating conditions. By determining the trend of the difference value, the drift of the measured air volume sensor is determined. The drift is then applied to the sensor error compensation, thereby achieving soft calibration of the flue air volume and effectively solving the problem of soft correction of flue air volume error.

[0088] In summary, this embodiment's method, based on the measured flue gas volume of a thermal power unit, simultaneously employs a soft measurement method to establish a flue gas volume prediction model and predicts the flue gas volume. It also combines a continuous difference method to continuously differentiate between the measured flue gas volume and the predicted flue gas volume under specific operating conditions. By determining the trend of the difference value, the drift of the measured flue gas volume sensor is determined. Finally, this drift is applied to sensor error compensation, thus achieving soft calibration of the flue gas volume. This embodiment effectively solves the problem of soft correction of flue gas volume errors. Moreover, this embodiment's method uses soft measurement and continuous difference methods to perform soft calibration of the flue gas volume sensor, avoiding excessive uncertainty in the flue gas volume measured by the sensor and measurement results that do not meet actual engineering needs. This embodiment's method determines the drift of the flue gas volume sensor and applies the drift to sensor error compensation, thereby achieving flue gas volume error correction. This embodiment's method achieves online calibration of flue gas volume, saving labor and time costs compared to offline calibration.

[0089] Furthermore, this embodiment also provides a multi-data fusion-based flue gas flow error soft correction system, including a microprocessor and a memory interconnected, wherein the microprocessor is programmed or configured to execute the multi-data fusion-based flue gas flow error soft correction method. Additionally, this embodiment also provides a computer-readable storage medium storing a computer program for being programmed or configured by the microprocessor to execute the multi-data fusion-based flue gas flow error soft correction method.

[0090] Example 2:

[0091] This embodiment is basically the same as Embodiment 1, with the main difference being the mapping model between the correlated variables and the flue gas volume in step S102. In this embodiment, the mapping model between the correlated variables and the flue gas volume in step S102 is an exponential regression model, and the function expression of the exponential regression model is:

[0092] ln(y) = nln(x) + ln(k),

[0093] In the above formula, y is the flue gas volume, x is the related variable, n and k are parameters, and ln is the natural logarithm function;

[0094]

[0095]

[0096] In the above formula, Let y be the average value of the flue gas volume. Let x be the mean of the related variable x. i Let y be the i-th sample value of parameter variable x.i Let be the i-th sample value of the flue gas volume y, and n be the number of historical data samples. Based on the mapping model between the correlated variables and the flue gas volume in step S102 of Example 1, which is a univariate linear regression model between the correlated variables and the flue gas volume, a simple calculation expression for the exponential model can be derived as y = k·x. n Using the above regression model, a formula for calculating the flue gas volume and related variables can be established; Q = m·X h Where X is the associated variable, Q is the flue gas volume, and m and h are undetermined parameters, multiple formulas for calculating the relationship between flue gas volume and associated variables can be listed. Furthermore, by taking the natural logarithm of both sides of the equation to convert it into a linear equation, the functional expression of the above-mentioned exponential regression model can be obtained. Based on this functional expression of the exponential regression model, a mapping model between associated variables and flue gas volume can also be realized, achieving a similar technical effect to Example 1.

[0097] Furthermore, this embodiment also provides a multi-data fusion-based flue gas flow error soft correction system, including a microprocessor and a memory interconnected, wherein the microprocessor is programmed or configured to execute the multi-data fusion-based flue gas flow error soft correction method. Additionally, this embodiment also provides a computer-readable storage medium storing a computer program for being programmed or configured by the microprocessor to execute the multi-data fusion-based flue gas flow error soft correction method.

[0098] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A soft correction method for flue gas volume error fusion based on multi-source data, characterized in that, include: S101, calculate the correlation coefficient between each parameter variable and the flue gas volume in the historical data sample, and finally retain the parameter variables with the highest correlation coefficient as the associated variables of the flue gas volume. S102, establish a multivariate data fusion soft measurement model of air volume based on the correlation variables of flue air volume, wherein the multivariate data fusion soft measurement model of air volume includes a mapping model between each correlation variable and flue air volume; S103, using the values ​​of the correlated variables measured under specified working conditions, the smoke exhaust volume is obtained by using a mapping model, and all the obtained smoke exhaust volumes are merged into a soft measurement prediction of the flue gas volume; S104, continuously differencing the measured flue gas flow rate of the flue gas flow sensor at the same time under specified operating conditions with the predicted flue gas flow rate using soft measurement, to obtain the difference value between the measured flue gas flow rate and the predicted flue gas flow rate using soft measurement. Based on the trend of the difference value, the drift of the flue gas flow sensor is determined, and this drift is applied to the error compensation of the flue gas flow sensor, thereby achieving soft correction of the flue gas flow sensor error. The calculation function expression for continuously differencing the measured flue gas flow rate of the flue gas flow sensor at the same time under specified operating conditions is as follows: ， In the above formula, Let be the first difference value at time t. The soft measurement is used to predict the flue gas volume at time t. Let t be the measured flue airflow from the flue airflow sensor. The drift of the flue airflow sensor is determined based on the trend of the difference value, including: First, the second-order difference value is calculated using the following formula: ， In the above formula, Let be the second difference value at time t. The first difference value at time t-1 The first-order difference value at time t-2; then the second-order difference value at time t. The drift of the flue gas flow sensor at time t.

2. The method for soft correction of flue gas volume error by multi-source data fusion according to claim 1, characterized in that, The function expression for calculating the correlation coefficient in step S101 is as follows: ， In the above formula, r is the correlation coefficient. Let X be the i-th sample value of parameter variable X. Let X be the mean of the parameter variable. Let Y be the i-th sample value of the flue gas volume. Let Y be the mean value of the flue gas volume, and n be the number of historical data samples.

3. The method for soft correction of flue gas volume error by multi-source data fusion according to claim 1, characterized in that, In step S102, the mapping model between the associated variable and the flue gas volume is a univariate linear regression model, and the functional expression of the univariate linear regression model is: ， In the above formula, For flue air volume, For related variables, , and For parameters, To conform to the distribution The random number, and , and estimator , as well as The result is obtained through inverse regression analysis using the following formula: ， ， ， In the above formula, For flue air volume The mean, For related variables The mean, For parameter variables The i-th sample value, For flue air volume The i-th sample value, where n is the number of historical data samples.

4. The method for soft correction of flue gas volume error by multi-source data fusion according to claim 1, characterized in that, In step S102, the mapping model between the correlated variables and the flue gas volume is an exponential regression model, and the functional expression of the exponential regression model is: ， In the above formula, For flue air volume, For related variables, Harmony parameter, It is the natural logarithm function; ， ， In the above formula, For flue air volume The mean, For related variables The mean, For parameter variables The i-th sample value, For flue air volume The i-th sample value, where n is the number of historical data samples.

5. The method for soft correction of flue gas volume error by multi-source data fusion according to claim 1, characterized in that, In step S103, the combined exhaust air volume of all the obtained data is expressed as a function for predicting the flue gas volume using soft measurement: ， In the above formula, To predict flue gas volume using soft measurement, Let be the variances of the n related variables. The exhaust air volume is obtained by using a mapping model with the values ​​of n related variables measured under specified operating conditions.

6. The method for soft correction of flue gas volume error by multi-source data fusion according to claim 1, characterized in that, Before step S101, the method further includes preprocessing the historical data samples obtained from the DCS system of the thermal power unit to remove gross values ​​from the historical data samples. The preprocessing to remove gross values ​​from the DCS system data samples of the thermal power unit includes: calculating the mean and standard deviation for each parameter variable. If the absolute value of the difference between the data of a certain parameter variable and the mean exceeds three times the standard deviation, then the data is removed as a gross value from the historical data samples.

7. A multi-source data fusion-based flue gas volume error soft correction system, comprising a microprocessor and a memory interconnected, characterized in that, The microprocessor is programmed or configured to execute the multi-data fusion flue gas volume error soft correction method according to any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program, characterized in that, The computer program is used to be programmed or configured by a microprocessor to execute the multi-data fusion flue gas volume error soft correction method according to any one of claims 1 to 6.

Images