Online dynamic detection and evaluation method for iron loading in torpedo tank cars

Abstract

The present invention relates to an online dynamic detection and evaluation method for the iron loading amount of a torpedo tank car, which comprises the following steps in sequence: obtaining the liquid level height through a radar rangefinder, obtaining the weight of molten iron through a weighing meter; establishing an iron loading record each time iron is loaded; performing data processing on each iron loading record to obtain the characteristic value of the iron loading, wherein the data processing includes data statistics, equation fitting and characteristic value calculation, and the characteristic value is the liquid level height value when the liquid level height change speed is the smallest; establishing a database of all iron loading records within a certain time period, first screening and filtering all iron loading record data to obtain a characteristic value set, and calculating the mean value, median and standard deviation, randomly extracting sample data from the characteristic value set for normality test, and performing single factor variance analysis on the characteristic value set according to the iron mouth line number, and finally determining the reference value. The present invention can effectively evaluate the additional error and dynamic accuracy existing in the dynamic measurement of the molten iron level height of a torpedo tank car.

Description

Technical Field

[0001] The invention relates to a blast furnace molten iron production and transportation technology, and in particular to an online dynamic detection and evaluation method for the iron loading amount of a torpedo tank car suitable for blast furnace molten iron transportation. Background Art

[0002] Blast furnace tapping refers to the discharge of molten iron and liquid slag from the furnace through the tap hole at regular intervals. The tapping is a process from the tap hole opening to the tap hole closing. The torpedo tank car is a large-scale molten iron transportation equipment with the advantages of low heat loss, long heat preservation time, and energy saving. When the blast furnace is tapping, the torpedo tank car is generally transported to the corresponding tap hole to hold the molten iron. Each torpedo tank car is loaded with molten iron from the beginning to the end of the loading of molten iron (also called receiving iron). Usually, multiple torpedo tank cars are required to load iron for a blast furnace to discharge molten iron.

[0003] Blast furnace tapping control has strict safety management requirements and quality control standards. It is necessary to use a detection system to measure the molten iron level and weight in the torpedo tank car to safely control the molten iron level. Figure 1 The torpedo tank car 1 is arranged on the track, and a tank opening 2 for molten iron to enter is arranged on the top of the torpedo tank car 1. The measuring devices in the detection system mainly include a radar rangefinder 3 (such as a microwave radar type molten iron liquid level meter) and a weighing meter 4 (such as a track weight type molten iron liquid level meter). The radar rangefinder 3 is arranged above the tank opening 2, and the weighing meter 4 is arranged on the track below the torpedo tank car 1. However, due to the extremely harsh on-site operating environment of the measuring device, the measurement data of the molten iron liquid level and the molten iron weight of the torpedo tank car have large additional errors, and the dynamic accuracy cannot be directly estimated.

[0004] Chinese patent 200510028317.7 discloses a method and device for detecting the level of molten iron in a torpedo tank car, which measures the height from the sensor to the tank body reference plane plus the height from the tank body reference plane to the bottom of the molten iron tank by setting a sensor, and then subtracts the height from the sensor to the molten iron level in the tank measured by the sensor to obtain the actual height of the molten iron level. Chinese patent 201210068857.8 discloses a monitoring device for the level and flow rate of molten iron in a torpedo tank car, which selects the bogie of the wheel as the base point, utilizes the change in the gap size between the bogie foundation brake device and the bolster when the tank car is loaded, and converts the pressure signal into a voltage signal through its squeezing of the piezoelectric sensor, and after data processing and calibration, obtains the information of the liquid level and liquid flow rate in the torpedo tank car. Although the above patents all realize the dynamic measurement of the molten iron level in the torpedo tank car, eliminate the influence of various factors on the measurement results and improve the accuracy of the measurement results to a certain extent, they are unable to make an effective evaluation of the dynamic measurement of the molten iron level and molten iron weight in the torpedo tank car, and are unable to understand the additional errors and dynamic accuracy in the measurement results. Summary of the invention

[0005] The object of the present invention is to provide an online dynamic detection and evaluation method for the iron loading amount of a torpedo tank car. The detection and evaluation method establishes an iron loading record database based on the online dynamic detection system of the torpedo tank car, calculates the characteristic value after equation fitting for each iron loading record, and obtains a reference value that can be used to determine the stability and reliability of the detection system through statistical process analysis on the basis of the characteristic value set obtained through screening and filtering, and can effectively evaluate the additional error and dynamic accuracy existing in the dynamic measurement of the molten iron level height of the torpedo tank car.

[0006] The present invention is achieved in that:

[0007] An online dynamic detection and evaluation method for the iron loading amount of a torpedo tank car, wherein a tank opening is arranged on the top of the torpedo tank car, a radar rangefinder is arranged above the tank opening, and a weighing meter is arranged on the track at the lower side of the torpedo tank car; the detection and evaluation method comprises the following steps:

[0008] Step 1: obtain the liquid level height by using a radar rangefinder, the liquid level height is the direct line of sight distance between the radar rangefinder and the molten iron surface in the torpedo tank car; obtain the weight of the molten iron in the torpedo tank car by using a weighing meter; establish a loading record each time the torpedo tank car is loaded with iron, the loading record includes the iron order number, loading time, iron mouth number, iron mouth line number, torpedo tank car number, the liquid level height and molten iron weight at the start of iron loading, the liquid level height and molten iron weight at the end of iron loading, and collect the liquid level height and molten iron weight values ​​at fixed intervals;

[0009] Step 2: Perform data processing on each iron loading record to obtain a characteristic value of the iron loading, wherein the data processing includes data statistics, equation fitting and characteristic value calculation in sequence, and the characteristic value is the liquid level height value when the liquid level height change speed is the smallest;

[0010] Step three, repeat steps one and two and establish a database of all iron loading records within a certain time period, first screen and filter each iron loading record data to obtain the characteristic value that can be used to calculate the benchmark value, then form the characteristic value that can be used to calculate the benchmark value into a characteristic value set, and then calculate the mean, median and standard deviation of the characteristic value set; set the sample size and randomly extract sample data from the characteristic value set for normality test, and perform one-way analysis of variance on the characteristic value set according to the iron mouth line number, and finally determine the benchmark value.

[0011] In the step 1, the fixed time is 10 seconds.

[0012] In the step 2, the equation fitting is obtained by the following cubic polynomial:

[0013] h=aw 3 +bw 2 +cw+d

[0014] In the formula, h is the liquid level, w is the weight of molten iron, and a, b, c, and d are constants;

[0015] The eigenvalue is calculated by taking the second derivative of the cubic polynomial. The calculation formula is as follows:

[0016] h′(w)=3aw 2 +2bw+c

[0017] h″(w)=6aw+2b

[0018] When h″(w)=0, the weight of molten iron at the inflection point w0 is obtained, and the calculation formula is as follows:

[0019]

[0020] Substituting w0 into the cubic polynomial, the liquid level height h0 at the inflection point is obtained, and the calculation formula is as follows:

[0021] h0=aw0 3 +bw0 2 +cw0+d

[0022] The liquid level height h0 is set as the characteristic value of this iron loading.

[0023] In the step 2, the interference data is filtered during the data processing.

[0024] In the step three, the screening and filtering conditions are: the iron loading weight is greater than 200 tons, the liquid level height difference is greater than 0.9 meters, the number of data collection points is greater than 100 points, the determination coefficient of the data fitting regression is greater than or equal to 0.99, and the liquid level height of the characteristic point is between 1.5 and 2.7 meters.

[0025] In step 3, the sample size is set according to the minimum sample size. The calculation formula of the minimum sample size is as follows:

[0026]

[0027] Where sigma is the standard deviation of the eigenvalue set, and Δ is 8 / 1000 of the mean value of the eigenvalue set.

[0028] In the step three, the minimum sample size required for each factor variance analysis is calculated before performing the one-way variance analysis, and the sample size of each iron mouth line is set according to the minimum sample size and equal variance tests are performed respectively.

[0029] In the step 3, if the P value of the normality test and the P value of the one-way ANOVA are both greater than or equal to 0.05, the average value of the eigenvalue set is used as the benchmark value.

[0030] The method for online dynamic detection and evaluation of the iron loading of a torpedo tank car of the present invention, firstly, aims at the problem of large additional errors in the dynamic measurement process of the molten iron level height and the weight of the molten iron of the torpedo tank car, and obtains the most stable dynamic measurement point of each iron loading-"characteristic value" through equation fitting based on the measurement data of the molten iron level height and the molten iron weight, and obtains the reference value by combining the statistical process analysis of the iron loading record of the torpedo tank car, which can timely reflect the stability of the iron loading detection system, and help to achieve the stable detection and dynamic accuracy of the iron loading of the torpedo tank car for blast furnace molten iron transportation. Secondly, based on the difference analysis of each iron mouth line of the blast furnace and the big data analysis of the iron loading record, the slagging and refractory brick peeling inside the torpedo tank car can be judged, providing a basis for equipment maintenance and replacement. Furthermore, the existing torpedo tank car detection system device is utilized, without increasing equipment investment or modifying the existing system, combined with the recognition of the on-site operating environment and the measurement stability, on the basis of establishing the stable data of the iron loading of the torpedo tank car, a method for effectively evaluating the dynamic accuracy of the molten iron level height measurement data is provided.

[0031] Compared with the prior art, the present invention has the following beneficial effects: characteristic values ​​are obtained based on the data collection of the molten iron level height and the molten iron weight of the torpedo tank car, and a reference value that can be used to evaluate the dynamic measurement stability and accuracy of the detection system is obtained in combination with statistical process analysis, which can meet the safety management and quality control requirements of blast furnace iron tapping control, help to improve the precision control of the iron loading of the torpedo tank car, and provide data support for equipment maintenance and replacement. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the structure of the molten iron level and molten iron weight detection system of the torpedo tank car;

[0033] Figure 2 A scatter plot of the liquid level and the weight of molten iron when a cylindrical torpedo tank car is loaded with iron;

[0034] Figure 3 A scatter plot of the liquid level and the weight of molten iron when loading an olive-type torpedo tank car with iron;

[0035] Figure 4 It is a schematic flow chart of the on-line dynamic detection and evaluation method of the iron loading amount of a torpedo tank car of the present invention;

[0036] Figure 5 It is a summary diagram of the characteristic values ​​after screening and filtering of the present invention;

[0037] Figure 6 Anderson-Darling normality test diagram of the eigenvalue sample of the present invention;

[0038] Figure 7 It is the Ryan-Joiner normality test diagram of the eigenvalue sample of the present invention;

[0039] Figure 8 Kolmogorov-Smirnov normality test diagram of the eigenvalue sample of the present invention;

[0040] Fig. 9 It is the efficacy curve diagram of the one-way variance analysis of the present invention;

[0041] Fig.10 It is an equal variance test diagram of the characteristic values ​​of each iron mouth line of the present invention;

[0042] Fig.11 This is a screenshot of the single factor variance analysis results of the characteristic values ​​of each iron mouth line of the present invention;

[0043] Fig.12 It is a box plot of the characteristic values ​​of each iron mouth line of the present invention.

[0044] In the picture, 1 is a torpedo tank car, 2 is the tank mouth, 3 is a radar rangefinder, 4 is a weighing scale, and 5 is the molten iron level. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with specific embodiments.

[0046] See also Figure 1 , an online dynamic detection and evaluation method for the iron loading amount of a torpedo tank car, wherein a tank mouth 2 is arranged on the top of the torpedo tank car 1, a radar rangefinder 3 is arranged above the tank mouth 2, and a weighing meter 4 is arranged on the lower track of the torpedo tank car 1. The tank body of the torpedo tank car 1 is in an "olive" shape and is symmetrical about the horizontal center line. When the blast furnace taps iron, the molten iron enters the torpedo tank car through the tank mouth 2. The liquid level height h of the molten iron in the torpedo tank car is the direct line of sight distance between the radar rangefinder 3 and the molten iron liquid surface 5. According to the iron loading production process, when the torpedo tank car is empty, its relative weight (minus the weight of the torpedo tank car tank body) is set to about 1 ton, and the liquid level height is set to about 2.6 meters, and when the torpedo tank car is full, its relative weight is generally about 240 to 280 tons, and the liquid level height is about 1.05 meters.

[0047] First, the relationship between the liquid level height of the torpedo tank car and the weight of the molten iron under ideal conditions is analyzed. Assuming that the shape of the torpedo tank car is cylindrical, the weight of the molten iron increases with the increase of the molten iron liquid level. Correspondingly, the weight of the molten iron increases with the decrease of the liquid level height h, that is, the weight of the molten iron and the liquid level height are approximately negatively correlated linearly, see Figure 2 In fact, the torpedo tank car is an olive-shaped tank, and the weight of molten iron and the liquid level height are not in a simple linear relationship. However, through the analysis of the equipment and on-site working conditions as well as the internal structure of the torpedo tank car, it can be considered that there is still a certain correlation between the weight of molten iron (measured by a weighing meter) and the liquid level height (measured by a radar rangefinder).

[0048] From the analysis of the internal structure of the torpedo tank car, when the empty torpedo tank car begins to be loaded with iron, the liquid level will drop rapidly at first with the increase of the weight of the molten iron, because the lower part of the tank body is approximately bowl-shaped. However, as the weight of the molten iron increases to a certain value range, the rate of drop of the liquid level will gradually decrease until the molten iron level reaches the middle of the tank body; then, as the weight of the molten iron continues to increase, the rate of drop of the liquid level will gradually increase until the molten iron level reaches the tank mouth. Therefore, for a new torpedo tank car without slagging and refractory brick peeling, the change law of the liquid level with the weight of the molten iron is exactly opposite in the upper and lower parts of the tank body, and is symmetrically distributed with the horizontal center line of the tank body. Figure 3 , randomly select the measurement data of a certain iron loading, delete the abnormal points, and the scatter distribution of the liquid level height and the molten iron weight is as follows: the upper curve (corresponding to the iron loading process in the lower half of the tank body) is a concave curve, and the lower curve (corresponding to the iron loading process in the upper half of the tank body) is a convex curve, and the convexity transition is achieved at the horizontal center line of the tank body. At the position where the convexity transition is achieved, the speed of change of the liquid level height with the change of the molten iron weight is the smallest, that is, the "change rate" is the smallest. Therefore, the liquid level height corresponding to this position is defined as the "characteristic value" of this iron loading, which represents the position when the liquid level height changes the least during each iron loading measurement process, and is also the most stable dynamic measurement point.

[0049] See also Figure 4 According to the above analysis of the internal structure of the torpedo tank car and the correlation between the liquid level height and the weight of molten iron, the detection and evaluation method includes the following steps:

[0050] Step 1: obtain the liquid level height through the radar rangefinder, and the liquid level height is the direct line of sight distance between the radar rangefinder and the molten iron surface in the torpedo tank car. The weight of the molten iron in the torpedo tank car is obtained by a weighing scale. A loading record is established each time the torpedo tank car is loaded with iron. The loading record includes the iron order number, loading time, iron mouth number, iron mouth line number, torpedo tank car number, the liquid level height and molten iron weight at the start of iron loading, and the liquid level height and molten iron weight at the end of iron loading. The liquid level height and molten iron weight values ​​are collected at fixed intervals, wherein each loading of iron corresponds to an iron order number.

[0051] In this step, the fixing time is preferably 10 seconds.

[0052] Step 2: Fit the equation for each iron loading record and obtain the characteristic value of the iron loading, that is, the liquid level height when the liquid level height change rate (changes with the weight of molten iron) is the smallest. Since the change of the molten iron level height changes nonlinearly and continuously with the change of the molten iron weight, the upper curve is concave, the lower curve is convex, and the convexity changes at the horizontal center line. Therefore, the position of the characteristic value is equivalent to the inflection point of the continuous differentiable function in calculus. The relationship between the liquid level height and the molten iron weight is obtained by fitting the polynomial equation, and then the inflection point position is obtained to obtain the characteristic value.

[0053] In this step, the equation fitting is obtained by the following cubic polynomial:

[0054] h=aw 3 +bw 2 +cw+d

[0055] In the formula, h is the liquid level, w is the weight of molten iron, and a, b, c, and d are constants;

[0056] The eigenvalue is calculated by taking the second derivative of the cubic polynomial. The specific calculation formula is as follows:

[0057] h′(w)=3aw 2 +2bw+c

[0058] h″(w)=6aw+2b

[0059] When h″(w)=0, the weight of molten iron at the inflection point w0 is obtained, and the calculation formula is as follows:

[0060]

[0061] Substituting w0 into the cubic polynomial, the liquid level height h0 at the inflection point is obtained, and the calculation formula is as follows:

[0062] h0=aw0 3 +bw0 2 +cw0+d

[0063] Therefore, the liquid level height h0 is the characteristic value of the iron loading, that is, the dynamic measurement point where the liquid level height is most stable.

[0064] According to the measurement of the liquid level height and molten iron weight of the torpedo tank car every 10 seconds during the blast furnace iron-tapping process, it normally takes about 30 to 45 minutes for each torpedo tank car to be filled, thereby generating about 180 to 270 sets of liquid level height and molten iron weight data, and a blast furnace generally has 4 tapholes, each of which consists of two lines A and B, so the blast furnace can accommodate up to 8 torpedo tank cars to load iron at the same time, so according to the monthly iron production of a blast furnace, about 200,000 sets of data will be generated every month. In this embodiment, by collecting the measured data for nearly 7 months, a total of 8,217 iron times, a total of about 1.3743 million sets of data are generated.

[0065] Preferably, SAS software is used to process the data of each iron loading record, and unreal interference data is filtered during the processing, so as to obtain the cubic polynomial equation of liquid level height and molten iron weight. In the present embodiment, 8217 iron times, i.e. 8217 pots of molten iron, are separated from the measured data. Specifically, through a macro process of SAS software, data statistics, equation fitting and eigenvalue calculation for each iron number are realized, and after completing the data processing of an iron number, the data processing of the next iron number is automatically performed by the loop macro call of the main program, until all data processing is completed and the calculation result is sent to an excel file. SAS (Statistics Analysis System) software is a statistical analysis software, with powerful data management and statistical analysis functions, and different from many customized software (such as minitab, etc.) for realizing statistical analysis by interface operation in the past, SAS is a statistical software based on programming language, which comprehensively covers various statistical analysis functions, with simple operation, powerful functions, professional results, and has powerful data processing functions.

[0066] In addition, all the measurement data of the torpedo tank cars during the blast furnace iron-making process are collected by the on-site L2 process computer and transmitted to the server for storage in real time, which can realize convenient search and query of the actual values ​​of the torpedo tank car liquid level height and molten iron weight within a certain period of time.

[0067] Step three, repeat steps one and two and establish a database of all iron loading records within a certain period of time, first screen and filter each iron loading record data to obtain the eigenvalue that can be used to calculate the benchmark value, then form these eigenvalues ​​that can be used to calculate the benchmark value into a eigenvalue set, and then calculate the mean, median and standard deviation of the eigenvalue set; set the sample size and randomly extract sample data from the eigenvalue set for normality test, and perform one-way analysis of variance on the eigenvalue set according to the iron mouth line number, and finally determine the benchmark value that can be used to evaluate the online dynamic detection system of the torpedo tanker iron loading.

[0068] This step is to obtain a reference static benchmark value through statistical process analysis based on the data of a large number of eigenvalues ​​obtained in step 1 and step 2. This is because it is difficult to directly judge and evaluate the stability and dynamic accuracy of the detection system based on these unscreened and unfiltered eigenvalues. In particular, after the detection system has been in normal use for a period of time, the benchmark value is the standard for judging whether the detection system is stable and whether it has offset during subsequent use, as well as evaluating the dynamic accuracy. Statistical process analysis includes screening and filtering the iron loading records, establishing eigenvalue sets, normality tests of eigenvalue sets, and one-way analysis of variance.

[0069] In this embodiment, 8217 characteristic values ​​of iron times are obtained, among which a considerable part of the data is unstable and cannot be used as the data basis for calculating the reference value. The characteristics of these unstable data are mainly manifested in that the iron loading is only half a tank, the parking position of the torpedo tank car deviates from the probe position of the radar rangefinder, and the large fluctuation of the molten iron level causes the data to mutate and cannot be fitted, etc. Therefore, it is necessary to screen and filter the iron loading record data of these 8217 times, and the specific conditions are as follows:

[0070] 1) Iron loading weight>200 tons;

[0071] 2) Liquid level height difference > 0.9m, i.e. the difference between the liquid level height at the start of iron loading and the liquid level height at the end of iron loading;

[0072] 3) The number of data collection points is greater than 100;

[0073] 4) The coefficient of determination (adjusted) of data fitting regression is Rsq_adj ≥ 0.99;

[0074] 5) The eigenvalue is in a reasonable numerical range, that is, 1.5≤h0≤2.7 meters.

[0075] Table 1 lists the statistical results of 8217 iron loading records after screening and filtering.

[0076] Table 1

[0077]

[0078]

[0079] According to the experience of judging metrological stability, the iron loading records used to calculate the benchmark value must meet the above five conditions at the same time. Therefore, there are a total of 2,817 iron loading records that meet these five conditions, accounting for 34.28% of all iron loading records. The data of these 2,817 iron loading records are relatively stable and can truly reflect the characteristics of the entire detection system as the data basis for calculating the benchmark value.

[0080] Considering that these 2817 iron loading records are an extremely small part of all measured data, the Six Sigma statistical analysis tool Minitab16 is used to analyze these data, making the analysis process more convenient and quick. Figure 5 , in the stable state of torpedo tanker loading iron, after Anderson-Darling normality test, the mean value of all eigenvalues ​​after screening and filtering is 2.2126 meters, the median is 2.2166 meters, the standard deviation is 0.1253 meters, the 95% confidence interval of the mean is (2.2080 meters, 2.2173 meters), the 95% confidence interval of the median is (2.2103 meters, 2.2225 meters), and the 95% confidence interval of the standard deviation is (0.1221 meters, 0.1286 meters). However, although the distribution of these eigenvalues ​​is basically symmetrical, it is not normally distributed. The reason for this non-normal distribution is probably due to the large sample size involved in the calculation, because when the sample size is very large, any small variation will affect the distribution law of the entire sample. To this end, the eigenvalues ​​of these 2817 iron records are taken as the "eigenvalue set", and the minimum sample size is calculated according to the standard deviation of the eigenvalue set and the acceptable deviation. Then the sample size is set according to the minimum sample size and the eigenvalue set is randomly sampled. According to the standard deviation of the eigenvalue set of 0.1253, the acceptable deviation Δ is set to 1 / 100 to 1 / 1000 of the average value of the eigenvalue set, and the calculation formula for the minimum sample size is as follows:

[0081]

[0082] In the formula, sigma is the standard deviation of the eigenvalue set, and Δ is 8 / 1000 of the mean value of the eigenvalue set, thus we get:

[0083]

[0084] The minimum sample size calculated is 193. Based on the minimum sample size, more samples are appropriately taken, and the sample size N in this embodiment is set to 250. 250 eigenvalues ​​are randomly selected from the eigenvalue set and a normality test is performed. Figure 6 , after the Anderson-Darling normality test, the mean was 2.211, the standard deviation was 0.1168, the AD was 0.350, and the P value was 0.471. Figure 7 , after Ryan-Joiner normality test, the mean is 2.211, the standard deviation is 0.1168, the RJ is 0.996, and the P value is > 0.100. Figure 8, after the Kolmogorov-Smirnov normality test, the mean is 2.211, the standard deviation is 0.1168, the KS is 0.036, and the P value is greater than 0.150. It can be seen that the P values ​​of the three normality tests are all ≥ 0.05, indicating that the data are normally distributed. Therefore, the following conclusion can be drawn: Under the stable state of blast furnace iron tapping, the characteristic value set of the torpedo tank car is normally distributed, with a mean of 2.2126 meters, a median of 2.2166 meters, and a standard deviation of 0.1253 meters. If the normality test result is not ideal, that is, the P value is less than 0.05, it means that the characteristic value set does not meet all sampling results, and the data needs to be re-screened and filtered, and the unreasonable data needs to be removed and recalculated.

[0085] According to the blast furnace, there are generally 4 tapholes, each of which consists of two lines, A and B, which is equivalent to 8 iron loading positions. Each iron loading position corresponds to a detection system that can measure the height and weight of the molten iron. The 8 detection systems are marked as 1A, 1B, 2A, 2B, 3A, 3B, 4A, and 4B according to the taphole line number. Whether there are differences between these 8 detection systems will also have an important impact on the evaluation of the entire blast furnace tapping detection system. Therefore, it is necessary to perform a single-factor variance analysis on the eigenvalues ​​measured and calculated by these 8 detection systems. The variance analysis must meet the requirement that the number of sample data collected by each factor (i.e., 8 detection systems) is the same and the variance is equal.

[0086] First, calculate the minimum sample size required for each factor, see Fig. 9 According to the settings of significance level Alpha of 0.05, standard deviation of 0.1253, number of levels of 8, and power of 0.9, the minimum sample size required for variance analysis of each factor is 59. Therefore, the sample size for variance analysis of each detection system in this embodiment is set to 60.

[0087] Next, 60 eigenvalues ​​are randomly selected from the eigenvalue set according to the iron wire number, for a total of 480 eigenvalues, and then an equal variance test is performed on them, see Fig.10 , the test statistic of the Bartlett test is 7.28, the P value is 0.401, the test statistic of the Levene test is 0.82, the P value is 0.568, and the P values ​​of the two tests are ≥0.05, indicating that the variances of the characteristic values ​​measured by the eight detection systems are not significantly different, and the prerequisites for variance analysis are met. If the equal variance test is not ideal, that is, the P value is <0.05, it means that some tapholes do not meet the requirements. This may be due to differences in the parameters of individual tapholes. In this case, variance analysis is required for each taphole to find the deviated taphole and perform data correction, and then recalculate.

[0088] See also Fig.11The P value of the one-way ANOVA is ≥ 0.05, indicating that there is no significant difference in the characteristic values ​​obtained by the eight detection systems. Therefore, the average value of the characteristic values ​​of the 2817 iron-loading records can be used as the benchmark value of the liquid level height of the torpedo tank car at all iron-loading positions during the first tapping of the blast furnace. Fig.12 Through the box plot of the characteristic values ​​of each taphole line, it can be concluded that there is no significant difference in the average values ​​of the characteristic values ​​of the torpedo tank cars of each taphole line of the blast furnace. If the variance analysis is not ideal, that is, the P value is less than 0.05, it means that some tapholes do not meet the requirements, which may be caused by differences in individual taphole parameters. In this case, it is necessary to calculate each taphole separately or add correction values.

[0089] Finally, the benchmark value of the blast furnace iron tapping detection system is obtained through the above steps one to three, that is, when the blast furnace iron tapping is stable, the characteristic value set of the torpedo tank car is normally distributed, with an average value of 2.2126 meters, a median of 2.2166 meters, and a standard deviation of 0.1253 meters. There is no significant difference in the characteristic values ​​of all iron loading positions of the blast furnace. The average value of the characteristic value set of 2.2126 meters can be used as the benchmark value of the liquid level height of the torpedo tank car at all iron loading positions, and as a standard for judging whether the detection system is stable, whether offset occurs, and evaluating dynamic accuracy during subsequent use.

[0090] The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car of the present invention can realize the evaluation of the operating state and dynamic accuracy of the detection system by confirming the characteristic value when the liquid level height change speed is the smallest during one iron loading process of the torpedo tank car, and using this as the basis for statistical process analysis and determining the reference value that can be used to judge the stability and reliability of the blast furnace iron tapping detection system. In the subsequent iron tapping process, the reference value can be used to measure whether the measurement data obtained by the detection system is within the normal fluctuation range, and the statistical process analysis of the characteristic value can be used to determine whether it has mutated. The differences in the iron loading process of each torpedo tank car can also be analyzed through other data mining methods such as cluster analysis and discriminant analysis, so as to further detect the slagging and refractory brick peeling inside each torpedo tank car, and provide data support for equipment maintenance and replacement.

[0091] The above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An online dynamic detection and evaluation method for the iron loading amount of a torpedo tank car, wherein a tank opening is arranged on the top of the torpedo tank car, a radar rangefinder is arranged above the tank opening, and a weighing meter is arranged on the lower track of the torpedo tank car; the method is characterized by: The detection and evaluation method comprises the following steps: Step 1: obtain the liquid level height by using a radar rangefinder, the liquid level height is the direct line of sight distance between the radar rangefinder and the molten iron surface in the torpedo tank car; obtain the weight of the molten iron in the torpedo tank car by using a weighing meter; establish a loading record each time the torpedo tank car is loaded with iron, the loading record includes the iron order number, loading time, iron mouth number, iron mouth line number, torpedo tank car number, the liquid level height and molten iron weight at the start of iron loading, the liquid level height and molten iron weight at the end of iron loading, and collect the liquid level height and molten iron weight values ​​at fixed intervals; Step 2: Perform data processing on each iron loading record to obtain a characteristic value of the iron loading, wherein the data processing includes data statistics, equation fitting and characteristic value calculation in sequence, and the characteristic value is the liquid level height value when the liquid level height change speed is the smallest; Step three, repeat steps one and two and establish a database of all iron loading records within a certain period of time, first obtain characteristic values ​​that can be used to calculate the benchmark value by screening and filtering each iron loading record data, then form a characteristic value set with the characteristic values ​​that can be used to calculate the benchmark value, and then calculate the mean, median and standard deviation of the characteristic value set; set the sample size and randomly extract sample data from the characteristic value set for normality test, and perform one-way analysis of variance on the characteristic value set according to the iron mouth line number, and finally determine the benchmark value; In the step three, the screening and filtering conditions are: the iron loading weight is greater than 200 tons, the liquid level height difference is greater than 0.9 meters, the number of data collection points is greater than 100 points, the determination coefficient of the data fitting regression is greater than or equal to 0.99, and the liquid level height of the characteristic point is between 1.5 and 2.7 meters.

2. The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car according to claim 1 is characterized in that: In the step 1, the fixed time is 10 seconds.

3. The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car according to claim 1 is characterized in that: In the step 2, the equation fitting is obtained by the following cubic polynomial: h=aw 3 + bw 2 + cw + d In the formula, h is the liquid level, w is the weight of molten iron, and a, b, c, and d are constants; The eigenvalue is calculated by taking the second derivative of the cubic polynomial, and the calculation formula is as follows: <h2 style=";text-align:left;direction:ltr">h′(w) = 3aw<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> +2bw+c h″(w)=6aw+2b When h″(w)=0, the weight of molten iron at the inflection point is obtained, and the calculation formula is as follows: Substituting w0 into the cubic polynomial, the liquid level height h0 at the inflection point is obtained, and the calculation formula is as follows: w0=aw0 3 +bw0 2 +cw0+d The liquid level height h0 is set as the characteristic value of this iron loading.

4. The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car according to claim 1 is characterized in that: In the step 2, the interference data is filtered during the data processing.

5. The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car according to claim 1 is characterized in that: In step 3, the sample size is set according to the minimum sample size. The calculation formula of the minimum sample size is as follows: Where sigma is the standard deviation of the eigenvalue set, and Δ is 8 / 1000 of the mean value of the eigenvalue set.

6. The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car according to claim 1 is characterized in that: In the step three, the minimum sample size required for each factor variance analysis is calculated before performing the one-way variance analysis, and the sample size of each iron mouth line is set according to the minimum sample size and equal variance tests are performed respectively.

7. The method for online dynamic detection and evaluation of the iron loading amount of a torpedo tank car according to claim 1 is characterized in that: In the step 3, if the P value of the normality test and the P value of the one-way ANOVA are both greater than or equal to 0.05, the average value of the eigenvalue set is used as the benchmark value.

Images