RU2583421C1 - Method for statistical acceptance control of large-sized cylindrical shells of rocket fuel tanks - Google Patents

Abstract

FIELD: engines.

SUBSTANCE: invention relates to quality control of products, in particular, large-size rocket fuel tanks. Method consists in selection of informative quality parameters (ICP) when making thin-wall tank shell. Herewith so called reference points, determining the weakest places of its structure, containing boundary ICP characteristics are specified. Initial imperfections of shape, measured over the entire surface of fuel tank shell in several interconnected sections, are selected. Obtained data is converted and processed for use in acceptance and selective control.

EFFECT: technical result consists in improvement of reliability of making decision on approving article for series production, in automation and higher control efficiency.

2 cl, 4 tbl, 5 dwg

Description

The invention relates to methods for statistical control of product quality in production and can be used for acceptance control when transferring a product from a pilot production enterprise to a mass production enterprise, in particular in quality control systems in rocket science.

In many branches of engineering (in the aerospace and shipbuilding industries, where thin-walled circular cylindrical shell structures (pipes, tanks, tanks, fuel tanks, etc.) are used, in developing the manufacturing technology, partial modeling is used, reflecting the controversial aspect, which requires confirmation.

 In rocket science, in the manufacture of a rocket hull, in particular a fuel tank hull, a full-size sample is used, which is subjected to comprehensive tests (ground and flight). Thus, a reference product instance (prototype) arises before the product goes into series. Serial production of missiles is organized in factories with significantly greater capacities than pilot production [1 p.109]. To debug the technological process at a serial plant, the release of an installation batch of products is appointed, the quality of which is carefully monitored. At the stage of production of the installation batch, the reliability of the product may change (sometimes for the worse) due to the refinement of the technological process, due to improvements carried out to improve the performance and the need to eliminate the established causes of failures, which are sometimes revealed during the control of the manufactured products. After successful ground and flight tests of products selected from the installation batch, the production technology and methods for controlling serial production are finalized [1 p.139].

The determination of the probabilistic characteristics of the mechanical reliability of the product casing in the conditions of ground tests at the stands before destruction is fraught with large material costs. Therefore, the number of such tests is extremely limited, and therefore, the spread of statistical estimates is large [1 p. 59]. The task of calculating reliability is simplified if we neglect the difference in the stress-strain state at different points in the structure. In this case, reliability should be calculated in relation to the “weakest” point at which the state of the structure is closest to the limit [1 p.65]. Since at other points in the structure the probability of destruction is less than at the “weakest” point, that is, they have a certain “safety margin”, the structure becomes unnecessarily heavier. In practice, the designers use the method of successive approximations to ensure the uniform reliability of the structure as a whole: they determine the probability of non-destruction not only at the “weakest” point, but also at other characteristic points of the structure, equalizing the safety margin by changing the design parameters of the structure, for example, changing the wall thickness, and then repeat the calculation of the probability of non-destruction. Numerous studies of the mechanical reliability of thin-walled rocket shells indicate a large discrepancy between theoretical and experimental data and prove that the reason for the inconsistency of the results of theory and practice is the presence of initial shape irregularities in real objects caused by defects in manufacturing, transportation, installation and operation [2, 3]. It is the structural units of thin-walled shells that have such shape imperfections that are the “weakest places” in the design of the rocket body and can be used as “reference points”, and the characteristics of their parameters can be used as boundary values in assessing quality.

This allows us to propose an acceptance control method, when a database in the form of an electronic version of quality parameters, which can be used as the boundary values of the initial shape irregularities, is created in a pilot plant with debugged technology, experienced personnel and control methods, and using this database for acceptance control is carried out at a serial plant. This is the essence of the invention.

A known method of statistical control of finished products [4 p.262-265], in which to assess the state of technological processes and adjust the values of process parameters according to the results of selective control of controlled parameters of products carry out the determination with the required accuracy of the actual values of the controlled parameter for units of production from the sample. The actual values of the controlled parameter are necessary for the subsequent calculation of the statistical characteristics by which a decision is made on the state of the technological process. The disadvantages of this method are: the time spent on statistical processing; the use of Shekhart control charts, which is cumbersome and unacceptable in small-scale production of rocketry products.

The analysis of the prior art made it possible to establish that analogues that are characterized by a combination of features that are identical to all the features of the claimed technical solution are absent, which indicates the compliance of the proposed method with the condition of patentability "novelty". The search results for known solutions in this and related fields of technology in order to identify features that match the distinctive features of the proposed method, did not give a result, which indicates that the claimed method meets the patentability condition "inventive step".

The objective of the invention is to increase the reliability of acceptance control of fuel missile bodies in quality control systems when starting up the product in series by obtaining systematized complete information about the boundary values of deviations of the parameters of initial shape irregularities, which improves the quality of assessment of the technical condition of the bodies and the stability of the manufacturing process.

The problem is achieved by the fact that the method of statistical acceptance control of large-sized cylindrical shells of fuel tanks of rockets, including quality control of a prototype of an object at a pilot plant before transferring it for production to a serial enterprise, based on the choice of informative parameters characterizing the quality of manufacture of a rocket fuel tank shell, this in the design of the prototype distinguish "reference" points that define the weakest places of the structure containing boundary ha characteristics of informative quality parameters, for which initial shape irregularities of thin-walled cylindrical shells are selected with the creation of a reference base of initial shape irregularity parameters; for this, at the stage of creating the reference database of the prototype, before transferring to serial production, the initial irregularities of the shape of the shell of the compartment are measured over the entire surface of the compartment in several parallel sections of the shell at the same time and repeatedly, rotating the compartment relative to the displacement sensors of the measuring system, while the compartment is reinstalled after the previous the measurement cycle at some predetermined fixed angle and again carry out the measurement cycle received from the sensors moved of a measuring system, a multicomponent analog signal is digitized, forming a matrix of quantized samples, using the values of its elements, a coefficient matrix is calculated using two-dimensional discrete cosine transformation, the elements of which are used as parameters of the initial shape irregularities, then these values of the coefficients are stored and the mean and average are calculated from them quadratic deviation, so that after the last measurement cycle, the matrix of coefficients will be The average values of the parameters of the initial shape irregularities, which will be used at the stage of quality diagnostics, are written, the data obtained is written to the hard drive and attached to the design documentation transferred to the serial plant, a comprehensive assessment of the conformity of the prototype design documentation is made, on the basis of which the set of diagnostic parameters is established qualities and permissible limits of their change, which are included in the reference database; in the process of acceptance control of a product sample from the installation batch in serial production, measurements are taken of the initial irregularities in the shape of the shell of the compartment over the entire surface of the compartment in several parallel sections of the shell, rotating the compartment relative to the displacement sensors of the measuring system, similar to the above measurement technology of the prototype, for this the compartment is installed on the measuring stand in accordance with the documentation for the prototype of the product, conduct a cycle of initial measurements incorrect form components obtained from the displacement sensors of the measuring system, the multicomponent analog signal is digitized, forming a matrix of quantized samples, using the values of its elements, a coefficient matrix is computed using two-dimensional discrete cosine transformation, the elements of which are compared with the corresponding elements available in the reference base of the prototype, according to the difference the values of the elements of the reference record and the control record of measurements make a decision on the admission of the product to mass production; in serial production for selective quality control of product bodies made in the form of thin-walled cylindrical shells, the matrix of initial shape irregularities coefficients of the prototype is used to control the technological process, comparing the parameters from the reference database with the initial shape irregularities of the control product’s shape, and also reveal the nature changes in the parameters of "reference" points with possible deviations of the technological process.

The specified set of essential features allows you to create a computerized system of acceptance control that allows you to track deviations in the manufacturing technology of product bodies at a serial plant using boundary values of quality parameters - the initial irregularities in the shape of the prototype.

The implementation of the method is illustrated by drawings, where

 figure 1 presents the stages of the method of acceptance control,

figure 2 - diagram of the measurements of the initial irregularities in the shape of the fuel tank on the measuring stand: the rotary part of the stand; displacement sensors of the measuring system; measuring system; PC;

figure 3 is a schematic diagram of a preliminary processing of a multi-channel analog signal,

 figure 4 - the formation of the vector of quantized samples of analog signals,

figure 5 - formation of a matrix of coefficients of a two-dimensional discrete cosine transform - quality parameters.

_N из M $$\times$$

N элементов f_ij, где f_ij - амплитуда квантованного сигнала i=1,2,..,M, j=1,2,…N. С целью выделения параметров отклонений от круглости отсека проводят двумерное дискретное косинусное преобразование матрицы квантованных отсчетов, описанное, например, в книге [9, с.159 - 162].The proposed method consists of three stages: the stage of creating a database of quality parameters - the initial irregularities in the shape of the prototype of the product, the stage of diagnostics of the quality of the test sample of the product - control recording and processing of the initial irregularities in the form of the serial sample from the installation batch during acceptance control and the decision-making phase - stage comparing the parameters of the reference and control records of the parameters of the initial irregularities of the form, on which by changing the parameters of the initial irregularities forms judge the change in the evenness of the shell of the product and, therefore, deviations, violations in the manufacturing process and make a decision on the admission of the product in mass production. These steps are shown in figure 1. At the first stage, it is supposed to create a reference base of parameters for the initial irregularities of the prototype shape before admitting the product to the series, for this the following actions are carried out. Mounted on a measuring stand 1 (Fig. 2), made, for example, similarly to [5] (Test bench for shell structures, RU No. 2 195 642 G01N 3/00 of December 7, 2002), a fuel tank body or other thin-walled shell of a rocket compartment driven in rotation around the longitudinal axis. Measurement system 2,3, built on the known principles [6, 7] using a contact or non-contact method, using the difference method, measures deviations from the roundness of the contours of the fuel tank body (compartment) —the initial irregularities in shape. In FIG. Figure 2 shows the contact method for measuring deviations using displacement strain gauges 2. Here, as a result of the interaction between the elastic element of the meter and the product body, the pressure of the force receiving unit on the displacement strain gauges 2 is converted into a multichannel (by the number of measurement belts) electrical analog signal. For further processing, it is necessary to convert a multi-channel analog signal into a multi-channel discrete signal. This problem is solved by an analog-to-digital converter, which performs discretization and quantization of an analog signal on each channel. Figure 3 shows a schematic diagram of the pre-processing of a multi-channel analog signal, where the device for entering data into a computer is connected to an external electrical circuit, including sensors of the measurement system 5, multi-channel strain gauge amplifier 6 of the type "Topaz - 3 (4)", filter 7, analog digital converter (ADC) 8, microcontroller 9, interface converter (UART / RS - 485) 10 and PC 11. The analogue signal is sampled in accordance with the Kotelnikov theorem. Then perform the quantization of discrete samples (Figure 4). Quantization is based on the methods described, for example, in the book [8, p. 142-161]. Next, based on the set of quantized discrete samples of the analog signal, a matrix of quantized samples is formed (Fig. 5) B _M $$\times$$

_N of M $$\times$$

N elements f _ij , where f _ij is the amplitude of the quantized signal i = 1,2, .., M, j = 1,2, ... N. In order to highlight the parameters of deviations from the circularity of the compartment, a two-dimensional discrete cosine transformation of the matrix of quantized samples is carried out, described, for example, in the book [9, p.159 - 162].

Two-dimensional discrete cosine transform (DKKP) is performed according to the following expression:

_N=A_M $$\times$$

_M $$\times$$

B_M $$\times$$

_N $$\times$$

A^T _N $$\times$$

_N,X _M $$\times$$

_N = A _M $$\times$$

_M $$\times$$

B _m $$\times$$

_N $$\times$$

A ^T _N $$\times$$

_N

_N - матрица коэффициентов ДДКП,where X _M $$\times$$

_N is the matrix of coefficients

_N - матрица квантованных отсчетов,B _m $$\times$$

_N is a matrix of quantized samples,

_M - матрица прямого одномерного ДКП,A _m $$\times$$

_M is a matrix of direct one-dimensional DCT,

_N - матрица обратного одномерного ДКП.A ^T _N $$\times$$

_N is the matrix of the inverse one-dimensional DCT.

One of the properties of the DCT is the separability property, according to which the DDT can be performed through one-dimensional DCT in rows and columns.

With this approach, direct DDKP defined by the expression:

B $$\times$$

A^T ,рассматривается как два произведения:X = A $$\times$$

B $$\times$$

A ^T , is considered as two works:

b - вычисление одномерного ДКП для каждого столбца матрицы.X ₁ = A $$\times$$

b - calculation of one-dimensional DCT for each column of the matrix.

A^T - произведение промежуточной матрицы на транспонированную матрицу ДКП - одномерное ДКП для каждой строки промежуточной матрицы. ДКП оперирует с исходными блоками, размером N $$\times$$

N отсчетов, и формирует блок размером N $$\times$$

N некоторых коэффициентов. В частности при N=4 вычисления ДКП можно выполнить с помощью матрицы преобразования, элементы которой равны:X = X ₁ $$\times$$

A ^T is the product of the intermediate matrix by the transposed DCT matrix - the one-dimensional DCT for each row of the intermediate matrix. DCT operates with source blocks of size N $$\times$$

N samples, and forms a block of size N $$\times$$

N of some coefficients. In particular, for N = 4, DCT calculations can be performed using the transformation matrix, whose elements are equal to:

Having calculated the cosines, we obtain a matrix of numerical coefficients A:

For example, suppose that, as a result of digitization, a matrix B is obtained that has sample values:

b, для данного примера имеет значение:The result of computing a one-dimensional discrete cosine transform performed for each column of the original matrix, i.e. X_one= A $$\times$$

b, for this example it matters:

A^T, для данного примера имеет вид:The final result of the two-dimensional discrete cosine transform X = X ₁ $$\times$$

A ^T , for this example, has the form:

, вычисляя математическое ожидание и среднее квадратическое отклонение. Для экономии памяти при создании базы данных для вычисления этих статистик будем использовать следующие формулы:As a result of the transformation, a matrix of coefficients of a two-dimensional discrete cosine transform is obtained, which is then used as a matrix of reference parameters of initial shape irregularities for the comparison operation with a serial sample. When creating a database in the mode of creating a reference record of initial irregularities of the form, the reference compartment of the rocket is installed on the rotary stand in a sequentially specified number of times. Each time, it is carefully verticalized, moreover, it is installed uniformly along the planes of rocket stabilization (I, III, and II, IV). That is, when reinstalling the product, the first time it is installed on the stand along the I - III plane and deviations are measured, then the product is reinstalled relative to the first measurement at an angle of 30 ° relative to the I - III plane and the rotary part of the stand is rotated again to measure the initial shape irregularities. The reinstallation operation of the product continues until the product is completely rotated relative to the fixed part of the stand - return to its original state (plane I - III). This allows you to reduce the measurement error due to the influence of the eccentricity and verticality of the installation, as well as calculate the mathematical expectation and standard deviation. Each of the array of deviation parameters introduced into the measurement processing system is stored individually, while the deviation parameters are calculated with a statistical check of the belonging of the obtained parameters to one sample. In view of the fact that the deviation parameters for the reference product are recorded many times, this allows us to give an interval estimate for each element of the matrix

calculating the mean and standard deviation. To save memory when creating a database, we will use the following formulas to calculate these statistics:

вектор параметров отклонений;Here:

vector of deviation parameters;

строка, столбец ячейки матрицы;

row, column of the matrix cell;

номер реализации записи при создании базы данных.

record implementation number when creating the database.

Thus, the sample of initial shape irregularities from the database is a matrix of coefficients of a two-dimensional discrete cosine transform, each element of which is the mathematical expectation of the corresponding values of the coefficients of the matrices of a two-dimensional discrete cosine transform of a reference sample obtained during a multiple measurement cycle.

N элементов так, что каждому ее элементу присваивают квантованное значение дискретного отсчета аналогового сигнала. Матрицу квантованных отсчетов подвергают двумерному дискретному косинусному преобразованию (ДДКП) (как подробно было показано выше на примере) путем перемножения матрицы коэффициентов одномерного дискретного косинусного преобразования (ДКП) A на матрицу квантованных отсчетов аналогового сигнала B и на транспонированную матрицу коэффициентов одномерного дискретного косинусного преобразования A^T (ДДКП) X=A $$\times$$

B $$\times$$

A^T. Полученную матрицу используют как матрицу контрольной записи отклонений и сравнивают при диагностике качества с эталонной матрицей базы данных эталонного образца, отвечающего всем требованиям конструкторской документации.At the second stage - the stage of diagnosing the quality of the investigated serial product sample — the control recording and processing of the initial irregularities in the shape of the serial sample during acceptance testing, the serial product sample is installed on a stand for measuring initial shape irregularities, similar to the above [5], they are verticalized and the initial irregularities are measured forms. Moreover, to reduce the installation error, the compartment is installed by means of an intermediate ring, which is the equivalent of the docking frame of the reference product. The resulting multi-channel analog signal is converted to digital form. For this, a continuous analog signal is sampled and quantized by discrete samples for each channel of the device. Then form a matrix of quantized samples of size M $$\times$$

N elements so that each of its elements is assigned a quantized value of the discrete reading of the analog signal. The matrix of quantized samples is subjected to a two-dimensional discrete cosine transform (DCT) (as was shown in the example above) by multiplying the matrix of coefficients of the one-dimensional discrete cosine transform (DCT) A by the matrix of quantized samples of the analog signal B and the transposed matrix of coefficients of the one-dimensional discrete cosine transform A ^T (DDKP) X = A $$\times$$

B $$\times$$

A ^T. The resulting matrix is used as the matrix of the control record of deviations and compared in the quality diagnostics with the reference matrix of the database of the reference sample that meets all the requirements of the design documentation.

At the decision-making stage, conclusions about the quality status of the serial sample from the installation batch, about the sufficiency of acceptance control methods and about admitting the product to the series are made by comparing the parameters of the reference and control records of the parameters of the initial shape irregularities, which are used to judge the change in the evenness of the product shell and, therefore , about deviations, violations in the technological process of manufacturing a serial sample. The following actions are performed. The quality parameters of the reference product taken from a pre-created database in the form of parameters of initial shape irregularities, presented as a matrix of coefficients of a two-dimensional discrete cosine transform, are compared element-wise with the control record matrix of a serial sample from the installation lot. If we assume that the statistics belonging to the whole set of measurements of the reference product are close to the normal distribution law, and the DCT coefficients for each element of the matrix of the serial sample fit into the interval of spread of the coefficients of the reference sample, then the serial sample from the installation lot is considered to satisfy the manufacturing quality, and a decision is made about admission of the product to the series.

The main advantage of the proposed method compared with the known methods of quality control in mechanical engineering is as follows:

- the method provides an increase in the informativeness of quality control and a decrease in subjectivity when deciding on the admission of a product to mass production;

- the above approach allows you to diagnose the stability of the production process in a serial enterprise;

- also the method is a means of diagnosing changes in individual technological operations of production;

- the optimally configured operations of the process will correspond to the matrix of parameters of the initial irregularities of the form, which we will call the reference matrix. This matrix can be used as a diagnostic for assessing the state of the process.

- the introduction of the method complements the existing control methods in the industry, increases the degree of automation of control methods, accelerates and increases the reliability of control.

In the end, it is possible to assess the stability of the process, the reliability of acceptance control of products and assess the real state of technological equipment.

As a result, the task of the invention was solved - improving the reliability of the data obtained during acceptance control by obtaining reliable, complete information, which ultimately reduces the cost of operating technological equipment and control measures.

 LITERATURE

1. Volkov L.I., Shishkevich A.M. Reliability of aircraft. - M .: Higher school. 1975, p. 109, p. 139.

2. Balabukh L.I., Alfutov N.A., Usyukin V.I. - Building mechanics of rockets. - M.: Higher School, 1984, p. 229 - 231.

3. Leizerovich G.S. Investigation of the dynamic characteristics of circular cylindrical shells with initial irregularities in shape. The dissertation on the specialty VAK 01.02.04 - mechanics of a deformable solid. - Komsomolsk-on-Amur: 2011, p.333. [electronic resource]: http://www.dissercat.com/content/.

4. Storm Regina. Probability theory. Math statistics.

 Statistical quality control. - M.: Mir, 1970 .-- p.262 - 265.

5. Test bench for shell structures, patent RU No. 2 195 642 G01N 3/00, publ. 12/27/2002.

6. Device for non-contact measurements, patent RU No. 108 599 from 09/20/2011, G 01 B 11/00.

7. Device for measuring and displaying technological parameters, patent RU No. 110 479 U1 dated 11/20/2011, G 01 B 11/00).

8. M.V. Nazarov, Yu.N. Petrov. - Methods of digital processing and transmission of digital signals. - M .: Radio and communications, 1985, p. 142-161.

9. Dyakonov V.P., Abramenkova I.A. Matlab signal and image processing. - St. Petersburg: Peter, 2002. S.162.

Claims (2)

1. The method of statistical acceptance control of large cylindrical shells of fuel tanks of rockets, including quality control of the prototype of the object at the pilot plant before transferring it for production to a serial enterprise, based on the choice of informative parameters characterizing the quality of manufacture of the rocket fuel tank shell, while reference points are identified in the prototype design to determine the weakest points of the structure containing boundary characteristics informative quality parameters, which are selected as the initial irregularities in the shape of thin-walled cylindrical shells with the creation of the reference base of the parameters of the initial irregularities of the shape, for this, at the stage of creating the reference database of the prototype, before transferring to serial production, measurements of the initial irregularities of the shape of the shell of the compartment over the entire surface of the compartment in several parallel sections of the shell are performed simultaneously and repeatedly, rotating the compartment relative to the measuring displacement sensors systems, while the compartment is reinstalled after a previous measurement cycle to a predetermined fixed angle and again carry out the measurement cycle obtained from the displacement sensors of the measuring system a multicomponent analog signal is digitized, forming a matrix of quantized samples, using the values of its elements, using a two-dimensional discrete cosine transform, a coefficient matrix, the elements of which are used as parameters of the initial shape irregularities, is then stored these values of the coefficients and calculate the mathematical expectation and the standard deviation from them, so that after the last measurement cycle, the coefficient matrix will contain the averaged values of the parameters of the initial shape irregularities, which will be used at the stage of quality diagnostics, the data are written to the hard disk and attached to design documentation, transmitted to the serial plant, produce a cumulative assessment of the conformity of the prototype design document On the basis of which the set of diagnostic quality parameters and the permissible limits of their change are established, which are included in the reference database, in the process of acceptance control of a product sample from the installation batch, serial measurements of initial irregularities in the shape of the compartment shell over the entire surface of the compartment in several parallel sections are carried out shell, rotating the compartment relative to the displacement sensors of the measuring system, similarly to the above measurement technology experienced about for this purpose, the compartment is installed on the measuring stand in accordance with the documentation for the prototype of the product, a cycle of measurements of initial shape irregularities is obtained, obtained from the displacement sensors of the measuring system the multicomponent analog signal is digitized, forming a matrix of quantized samples, using the values of its elements, using a two-dimensional discrete cosine transform, a coefficient matrix, the elements of which are compared with the corresponding elements available in the reference base of the prototype, are accepted for the value of the difference between the values of the elements of the reference record and decision on admission of the product to mass production. 2. The method according to claim 1, characterized in that in the conditions of mass production for selective quality control of product bodies made in the form of thin-walled cylindrical shells, the matrix of coefficients of initial irregularities in the shape of the prototype is used to control the process, comparing the parameters from the reference database with parameters of initial irregularities in the shape of the control sample of the product, and also reveal the nature of the change in the parameters of these reference points with possible deviations one process.

Images