JP2002007375A - Compromised balance solution for rotary machine and control method therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To surely and speedily obtain a really proper solution. SOLUTION: A radius ri at a position to attach a weight wi of a mass mi to a rotator, the mass mi and the number of rotations of the rotator are defined as constants, arbitrary rotating angle position coordinates θ on the surface part of the rotator are defined as a variable, and a first effect vector fm(θ) corresponding to (m) axial positions is calculated. A residual square sum S concerning a function with fm(θ) as a multivariable is calculated in a minimizing direction by a marquardt method. Angle position coordinates θ(k) around the rotational axis centerline of the weight Wi, with which S is settled within an allowable range, inside a rotation number area as a target of vibration evaluation are found. By avoiding the absence of a really effective solution, the execution of useless calculation or calculation requiring long time is avoided and the really proper solution can be speedily found by the calculating method of marquardt surely toward the solution in the manner of steepest diving.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for compromising balancing of a rotating machine and a method for adjusting the same, and more particularly to a large and small rotating machine such as a wind machine, a turbine for power generation, a supercharger, and an engine. The present invention relates to a compromise balancing solution for quickly and surely optimizing the eccentricity of the robot, and a method of adjusting the compromise.

[0002]

2. Description of the Related Art Rotary machines are positioned as one of the cores of industrial capital and are provided as generators, various power generation turbines, superchargers for engines, water turbines, and windmills. The rotor of the rotating machine is supported by bearings arranged at a plurality of points, and reaches the maximum rotation speed (the rotation speed corresponding to the rotation speed per unit time is hereinafter referred to as the rotation speed) in an appropriate time zone after starting. Reach. The vibration that occurs according to the rotation speed has complicated vibration characteristics. The vibration of the rotor has complicated vibration characteristics due to the number of rotations, the static balance of the rotor itself (existence of eccentricity at rest), and the dynamic balance based on the rotation of the rotating shaft.

[0003] The rotor is machined from a variety of materials,
Various parts like wings are attached and formed,
The center of mass of the rotor does not coincide with the axis of rotation at all axially different points. Such inconsistencies cause complicated vibrations in the rotor and excessive dynamic loads on the bearings. The generation of such excessive vibration and dynamic load of the bearing is not preferable from the viewpoint of the operational strength and function of the machine and from the viewpoint of human sensation.
In order to quantitatively suppress the generation of vibration and bearing dynamic load, there are the following two types of suppression methods. (1) Extremely improving the machining accuracy of the rotor and the machining / fitting of parts. (2) Reduce unbalance by proper balancing. Basically, the method (1) is important, but the large rotor of the turbine and the generator is several μm only by the method (1) over the entire length.
It is difficult to keep unbalance (eccentricity) within the accuracy range of m. It is relatively easy to set the unbalance (eccentricity) in the range by the method (2), and it is economically advantageous.

[0004] Balancing is performed by measuring the amount of eccentricity in each bearing (step S1), creating a sum of squares to be minimized with respect to the amount of eccentricity, and solving a system of linearized equations based on the least squares method (step S2). And determining the position of the weight to be added to the rotor and its mass based on the determined variables (step S3), and these processes are automated.

[0005] Due to the increase in the number of revolutions from start to steady operation, the vibration has several modes. Mode order is 3
, The steady operation is performed between the secondary critical speed included in the secondary mode and the tertiary critical speed included in the tertiary mode. It is required to suppress the vibration at the critical speed before starting the steady operation and the vibration at the operating speed. A modal balancing method is known as a technique for applying the least-squares method to suppress unbalance in each of the modes, and an influence coefficient method and a mode circle balance method are known as other techniques (for example, Japanese Patent Application Laid-Open No. 264704).

[0006] The increase in precision required as the size of the rotating machine increases leads to an increase in the number of simultaneous equations, and a solution exceeding the mountable range of the rotor may be obtained or no solution may be obtained. It is necessary to perform not only analytical tests on the factory side but also on-site field balance.However, depending on the surrounding environment, it is difficult to attach the balance weight according to the solution obtained, or the balance The arrangement is limited, and the time required to solve the linearized equation based on the least squares method becomes longer. In addition, known solutions have the problem that it is often difficult in principle to derive an optimal solution.

In addition to such general problems, there is another problem that must be dealt with in the strict API STANDARD accompanying the recent increase in precision. API STAND
In the ARD, the resonance avoidance rate of the compressor shaft vibration is shown in FIG.
As shown in FIG. 9, regarding the error between the test result at the factory and the analysis value, the analysis accuracy in Table 1 shown in FIG. 9 is required. Furthermore, the latest version of Ver. In 6, the main contents are
The changes shown in Table 2 of FIG. 10 have been made, and in order to perform the balance adjustment within a limited time of the verification test, it is indispensable to advance the on-site balance technology.

As described above, it is often difficult for the user to perform field balancing. In addition, it is required to be able to preferentially cope with shaft bending due to the occurrence of temperature distribution during rated operation that reaches the maximum rotation speed. Furthermore,
In response to such a request, it is desired to quickly derive an effective solution under various constraint conditions. It is desirable to share data between the site and the design.

[0009]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a compromise balancing solution for a rotary machine and a method for adjusting the same, which can facilitate the field balance by the user. Another object of the present invention is to provide a compromise solution for a rotating machine that can preferentially optimize the balance at the time of rated operation, and an adjustment method thereof. SUMMARY OF THE INVENTION It is an object of the present invention to provide a compromise solution method for a rotating machine capable of quickly deriving an effective solution under various constraint conditions, and an adjustment method thereof.

[0010]

Means for solving the problem are described as follows. The technical items appearing in the expression are appended with numbers, symbols, and the like in parentheses (). The numbers, symbols, and the like are technical items that constitute at least one embodiment or a plurality of the embodiments of the present invention, in particular, the embodiments or the examples. Corresponds to the reference numerals, reference symbols, and the like assigned to the technical matters expressed in the drawings corresponding to the above. Such reference numbers and reference symbols clarify the correspondence and bridging between the technical matters described in the claims and the technical matters of the embodiments or examples. Such correspondence / bridge does not mean that the technical matters described in the claims are interpreted as being limited to the technical matters of the embodiments or the examples.

[0011] compromise balancing solution of a rotary machine according to the invention, the mass m _{i (i} is 1 to n, n is 1 or a positive integer greater than 1) attaching the weight W _i having a rotating body The radius r _{i of the} position, its mass _mi, and the number of rotations of the rotating body are constants, and arbitrary rotational angle position coordinates θ of the surface portion of the rotating body are variables, corresponding to m axial positions. The first effect vector (the amount related to the amount of dynamic eccentricity) f
calculating _m (θ), calculating the residual sum of squares S relating to a function having f _m (θ) as a multivariable in a direction to minimize by the Marcard method, in a rotational speed region which is a target of vibration evaluation. the S is composed of the obtaining about the angular position coordinate of the rotation axial line of the weight W _i that fall within the allowable range θ ^(k). By avoiding the execution of useless or time-consuming calculations by avoiding the absence of a realistically valid solution, the Marcard calculation method that ensures a steepest descent toward the solution ensures a realistic A solution that is appropriate for the user can be quickly obtained. It is possible to stop the calculation before the calculation direction goes to a solution that is not actually appropriate, and to reliably obtain the optimum solution among the solutions that are actually appropriate. Here, the solution that is practically appropriate is a solution when parameters such as the magnitude of the mass, the number of attachments, the number of multiple variables, and the radius of the attachment position are set to appropriate values. By quickly calculating a solution within a permissible range for a parameter that is changed within an appropriate range, a model of a rotating body design can be changed quickly. Regarding a plurality of weights arranged on the same plane, it is possible to unitize them in calculation.

Prior to making such a calculation, W
_i2nd effect vector for rotating body with no
The torr is measured. W_iRotating body without attached
Means that even if a weight has already been added,
The rotating body is initially defined including the weight. Big bias
The addition of a weight to reduce the amount of heart is empirical without calculation
The on-site worker can make a judgment. To some extent
The rotating body with a small center is the object to which the weight is added.
Initially set as a rotating body. Initially small residual
The Newton method, the Gauss-Newton method,
Card method is more effective, calculation speed is faster
No. Let the second effect vector be ε_mThe residual sum of squares S
More specifically, the following equation: S = Σ (ε_m−f_m)²  Characteristically represented by Here, the characteristic is each term in this equation.
Means that a weight function can be applied, etc.
are doing. Σ means to add about m
You.

It is actually appropriate that the axial position is a set and fixed constant. In the rotating body to be tested, a plurality of weight mounting holes are provided in advance on one circumference on a rotation plane perpendicular to the axis set at a plurality of positions different in the axial direction. If there are four such planes, the maximum number of multivariables is 80.

It is also realistic that the mass _mi is a set and fixed constant. In this case, all of the maximum 80 weights have the same value. It is realistic both axial position and the mass m _i is a fixed constant is set.

[0015] The method for adjusting a rotary machine according to the present invention includes the following steps.
By arranging the computer (11) corresponding to the design department, arranging the second computer 12 corresponding to the field department, and connecting the first computer 11 and the second computer 12, the data can be transferred to the first computer ( 11) and be shared by the second computer (12), the mass _m i (i is 1 to n, n is attaching the weight _{W i} having 1 or a positive integer greater than 1) to the rotating member position radius and r _i and the mass m _i between the rotational speed and the constant of the rotary member, corresponding to an arbitrary rotational angular position coordinate θ as a variable m pieces of the axial position of the surface portion of the rotary body defined by the arrangement of the weight W _i of Calculating the first dynamic effect vector (eccentricity) f _m (θ) to be performed, in the direction of minimizing the (residual) sum of squares S by a Marcard method for a function having f _m (θ) as a multivariable. Calculating, vibration rating Obtaining angular position coordinates θ ^{(k )} about the rotation axis of the weight W _i in which the S falls within the allowable range in the rotation speed region which is the target of the price, and actually measuring the second effect vector on the rotating body at the site And the data is a second
Includes dynamic eccentricity. The variables are weight mass m and phase θ
And the number is usually the number of balance planes × 2.

By sharing the design parameters and actual measurement values on the net between the design department and the site, the designer and the on-site inspector have an appropriate allowable range based on a common viewpoint. Compromise balancing can be executed, and an autonomous test on site can be appropriately promoted, so that a quick model design based on measured values can be performed. A unified test can be conducted between both parties. Further, multiple tests can be performed simultaneously at multiple sites.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Corresponding to the drawings, an embodiment of the compromise balancing solution for a rotating machine according to the present invention comprises the steps of setting an allowable range for each rotation speed and obtaining an effective solution from a system of linear equations. It consists of a process that involves the steps of a calculation that is quickly sought. The rotor 1 manufactured and installed on a test bench in a factory, or the rotor 1 shipped and installed on-site at the site is shown in an abstract form as shown in FIG. The rotating shaft 2 of the rotor 1 is supported on both sides by two # 1 bearings 3 and # 2 bearings 4.

Nyquist diagrams for the # 1 bearing 3 and the # 2 bearing 4 are shown in FIGS. 1 (b) and 1 (c), respectively. The Nyquist diagram is expressed in a polar coordinate system, in which the length of a vector ε _j extending in the radial direction from the origin is the amplitude, and the phase is α shown in FIGS. 2 (a) and 2 (b). This α
Is a phase angle between the position of the synchronous rotation signal 5 of the rotor 1 and the position of the maximum value of the vibration waveform 6, and is in the range of 0 ° to 360 °. The amplitude is the amplitude R of the vibration waveform 6.
As described above, the Nyquist diagram is a locus drawn by the tip point of the vector ε _j while the rotation speed changes from zero to the target rotation speed. As illustrated in FIGS. 1B and 1C, the origin O indicates the vector end point of ε when the rotation speed is zero.

[0019] In FIG. 1 (b), (c), the end point of the motion vector epsilon ₁ is the primary critical rotation speed region corresponding to the first mode mode order of the rotational state is 1 (solid line display), the dynamic The end point of the vector ε ₂ is 2 where the mode order in the rotating state is 2.
Secondary mode, secondary critical rotation speed area (indicated by dashed line)
It is in. FIG. 1B relates to the # 1 bearing 3 and FIG. 1C relates to the # 2 bearing 4.

FIG. 3 shows a balance way added to the rotor.
Point P_jkIs shown. Rotary axis of rotor
Four circles C on a plurality of planes (example: four planes) orthogonal to the line L
_j(J = 1 to 4) correspond. Yen C_jProper spacing above
(Exemplary: equiangular spacing) at arrangement point P _jkIs distributed
You. k can be given a different value for each circle. Multiple points P
_jkIs the center point of the multiple circular holes provided on the rotor
Matches. Multiple circular holes must have the same shape
Is convenient for calculation convenience. How many of the multiple circular holes
A balance weight is attached to the hole. All baluns
The weights are calculated to have their mass equal to each other
It is convenient for convenience.

Step 1: The diagrams in FIGS. 1B and 1C are drawn using actually measured values. The least squares method is
With respect to four points (normally several tens) on the locus of both Nyquist diagrams, the basal amount S is defined as follows. When the modal balancing method is used as a technique for applying the least-squares method and suppressing unbalance in each mode, the dynamic eccentricity value ε _i of each mode is defined by the following equation. ^{_{S = Σ | ε i | 2}} = ε 1 2 + ε 2 2 + ε 3 2 + ε 4 2 ··· (1) In general, i is 1 to m, m is sufficiently large. ^{_{S = Σ | ε i | 2}} = ε 1 2 + ε 2 2 + ··· + ε m 2 ··· (2)

Ε _i is a quantity defined by a theoretical equation.

(Equation 1) Here, U _k indicates the balance at the k point, φ _rk indicates the r-th mode at the k point, and M _r is the r-th order modal mass. n is theoretically 1 to ∞, but n is actually 2
A good approximation is obtained with ~ 3. The unbalanced response of a general rotor system can be represented by the superposition of each vibration mode, and the number of modes considered for superposition is one of the higher-order modes or the higher-order modes that have a large effect on the number of modes up to the maximum rotation speed. Modal balance is a method that balances the unbalance remaining in the actual rotor for each mode from the lower order to the highest mode of interest, which is sufficient for practical use. Is considered, the expression (1) is further modified so as to be expressed by the rotational angular velocity ω, the r-th critical angular velocity, and the like. Equation (1) is an example, and the equation representing the vibration amplitude is variously expressed.

Step 2: A balance weight is attached to a plurality of circular holes at appropriate positions, and a rotation test is performed. ε _j is described by a parameter group and a variable.
The parameter group includes a mass distribution relating to the static contact inherent to the rotor, the mass of the added balance weight, the radius R of the circle C _jk , the rotational angular velocity ω, and the axial coordinate positions of the above-described plurality of surfaces. including. The imbalance is the vibration data obtained from the rotation test when there is no additional mass,
A good approximation can be easily obtained from the vibration data obtained from the rotation test when an appropriate additional mass is added. The mass and radius R of the balance weight and the axial coordinate positions of the plurality of surfaces can be treated as constants.

The variable is the angle θ _jk of the center of gravity of the balance weight attached to the reference radial coordinate axis in the polar coordinate system of each circle C _jk . Since the balance weights are arranged at equal intervals in principle on the same circumference, the actual value of θ _jk is calculated on the same circumference if the number of balance weights attached on the same circumference is n. Is the angular coordinate of the _sth balance weight of θ _jk = 0 + 360 ·
(S / n). θ _jk is a multivariable consisting of (n ₁ + n ₂ +... + n _j ) variables if the number of balance weights attached to the j-th plurality of surfaces described above is n _j .

Step 3: Regardless of whether no balance weight is added and the number of balance weights is zero or whether the number of balance weights is already N, equation (2) is established in that state, and θ _{jk is obtained.}
Is a constant. Next, assuming that a new balance weight is added, the angular coordinates θ of the equivalent balance weight that can be regarded as having only one piece added can be determined according to the number and mass of the balance weight. . In this case, equation (2) is changed to the following equation.

S = Σ | ε_i|²= (Ε₁−f₁)²+ (Ε₂−f₂)²+ ... + (ε_m−f _m )²... (4) f_m= F_m(Θ) This equation is a basic equation to which the least squares method is applied. this
The formula is m variables f_mIs a non-linear quadratic equation with rose
The problem of minimizing S in the weight equation (4) is that
This is consistent with finding θ that minimizes the numerical function S.
The simplest minimization algorithm is the steepest descent method (meth
od of steepest descent, well known as the maximum slope method
Have been. In this method, the decrease in S is local at each iteration.
This is a method of searching along the direction that maximizes
Direction d from torque θ_jWhen it moves slightly, d_j= -∂S / ∂θ_j  , S decreases most greatly per angular distance. now
, The vector d_jIs always oriented in the circumferential direction
Where θ can be treated as a scalar. d = -∂S / ∂θ Balance weight is minimum point θ_k(Hereinafter, represented by β)
、 S / ∂θ |_{θ = β}= 0 ... (5)

This equation is a simultaneous nonlinear equation, and a minimum point is obtained by a partial differential matrix of the second or higher order. The Newton method has long been known as a method for solving a nonlinear equation. According to the Newton method, the k-th approximate value θ
By solving Σ (∂ ² S / ∂θ _j ∂θ _k ) Δθ _k = −∂S / ∂θ _j (6) with respect to ^(k) , the following approximate value: θ ^{(k + 1)} = Θ ^(k) + Δθ can be obtained. Here, Δθ is called a correction vector. The Gauss-Newton method is known as a method of approximately solving the Newton method. Residual (ε ₁ −f ₁ )
Is small, or when the second derivative of f is small and close to linear, the Gauss-Newton method is expected to be effective. However, there is no quadratic convergence unless the residual becomes 0 for the true solution. If the Jacobian matrix of the first derivative appearing in the system of linear equations is represented by A, the equation (5) ignoring the second derivative can be expressed by the following equation.

(A ^to WA) Δθ = A ^to W (ε ₋ f (θ)) (7) Hereinafter, in this specification, W ^1/2 A is A, and W ^1/2 ε is ε.
And W ^1/2 f (θ) is represented by f (θ). According to this expression, equation (7) is represented by the following equation. A ^to AΔθ = A ^to (ε ₋ f (θ)) = Av≡b (8)

Step 4: The Gauss-Newton method is stable when the initial value θ is far from the true value or when the nonlinearity of f (θ) is large (the nonlinearity of the rotating body is large). It is not guaranteed that the residual sum of squares S (θ) decreases at each step of the iteration. In order to stabilize it, a correction vector factor α (0 <α <1) is introduced: If θ ^{(k + 1)} = θ ^(k) + αΔθα is taken sufficiently small, the following equation must be obtained as long as the rounding error can be ignored. It is easily proved that the equation holds. ^{S (θ (k + 1)} ) ≦ S (θ k) ··· (9)

The problem of such a Gauss-Newton method lies in how to take α. If α is reduced, the convergence becomes slower, but the convergence becomes slower. Therefore, at the beginning of the process, α is set to be relatively large, and α is successively reduced to ま で until Expression (8) is satisfied. In the next step of the repetition, if the last α of the previous time is less than half, use twice as the starting value. Thereby, the value of α is adjusted so that α is small when the nonlinearity is large, and becomes large as the linearity approaches.

Step 4 (Marquardt method, Marcard method): An additional term is added to the diagonal element of the coefficient matrix of equation (8), and the correction vector Δθ is set as follows. (A ^to A + λI) Δθ = b (10) where I is a unit matrix. λ is zero or a positive number. Therefore, Δθ = (A ^to A + λI) ⁻¹ b (11) If λ is zero, the Markard method matches the Gauss-Newton method. If λ≫‖A ^to A‖, then Δθ ^to λ-1b

Here, ~ indicates that they are substantially equal. The solution in this case is consistent with the steepest descent method. The Marcard method is a compromise between the Gauss-Newton method and the steepest descent method. If the influence of the nonlinearity is large far from the solution, λ is increased and λ is reduced as the solution approaches, so that the solution can be obtained stably and quickly. Such a solution is called the Carmer Card method.

.DELTA..theta. Is treated as a function of .lamda..theta. (. Lambda.). Differentiate equation (11). Δθ (λ) + (A ^to A + λI) (dΔθ (λ) / dλ) = 0 (12) Therefore, dΔθ (λ) / dλ = − (A to ^A + λI) ⁻¹ Δθ (13) ) deforms this, d (‖Δθ (λ) || ^{2) / dλ = 2Δθ ∧ (} dΔθ / dλ) = -2Δθ ∧ (a ~ a + λI) -1 Δθ <0 ··· (14)

This equation shows that {Δθ (λ)} decreases monotonically as λ increases from 0. Therefore, λ acts as a kind of reduction factor. Further, as λ increases, Δθ (λ) becomes the steepest descent direction d
(D _j ) = b approaches. That is, if an angle formed by _dj and Δθ (λ) is represented by φ (λ), cos (φ (λ)) = d _j Δθ (λ) / ‖d _j ‖‖Δθ
(Λ) ‖ Hereinafter, the expression is changed to a scalar expression. cos (φ (λ)) = dΔθ (λ) / ‖d‖‖Δθ
(Λ) ‖ According to Schwarz's inequality, d (cos (φ (λ))) / dλ ≧ 0 (15)

These two remarkable properties (Equation (14))
According to Equation (15), if λ is made sufficiently large, it is guaranteed that a smaller sum of squares point can always be found. When λ is small, {Δθ (λ)} often moves almost linearly with λ, which is another advantage of the Marcard method. By eigenvalue decomposition of A ^to A,

[0036]

(Equation 2) Thinking,

(Equation 3) Since it is, if _{e 1 ≧ e 2 ≧ ··· ≧} e m-1 »e m, for the λ«e _m-1 satisfying lambda,

(Equation 4) Holds. That is, if the smallest eigenvalue is sufficiently far from the other eigenvalues, for small λ, Δθ (λ) is u
Move on a line parallel to _m . Such a movement is shown in FIG. This is the behavior of Δθ in the case of the following matrix.

(Equation 5)

Step 5: By adopting the Marcard method,
Factory testing of large rotating machines, shortening the installation period, and improving mass productivity of small rotating machines are possible, but pure mathematical solutions alone can limit the diameter R of rotating bodies, the number of balance weights, and the layout. There is a case where there is no solution. A solution in which the diameter R becomes infinite is actually meaningless. The inventor therefore proposes a compromised balancing method.

The compensated balance is not a method of finding an optimum value, but a method of setting a specified value that defines an allowable range and obtaining an allowable value falling within the allowable range defined by the specified value for each rotation speed by the Marcard method. It is. FIG.
As shown in (a) and (b), a center circle region is set as an allowable range as a Nyquist diagram. The size of the radius r of the center circle is set as a specified value (for example, 25 μm). As shown in FIG. 5A, the unbalanced Nyquist line Lb before obtaining the allowable value (or compromise value) by the Marcard method is in a range from 0 rpm to 3600 rpm, and most of the Nyquist line Lb is larger than the allowable range. It is protruding. The Nyquist line Lb before the balance is moved almost parallel to the left by the balancing by the Marcard method, and the Nyquist line La after the balance is obtained. 3600 rp of Nyquist line La after balance
The rotational speed region portion of m is within the allowable range circle C.

FIG. 5B shows another embodiment of the compensated balancing. Balancing target area portion (governor range) Lb ′ approaching 3600 rpm, which is the balancing target of pre-balance Nyquist line Lb.
Is completely out of the permissible range circle C. Nyquist line La after balance by balancing by Marcard method
In the graph, the entire range La ′ of the portion to be balanced approaching 3600 rpm is within the allowable range circle C.

Step 6: FIG. 6 shows the thermal unbalance of the rotor (including the rotating shaft) supported by the multipoint bearing. Even if the amount of eccentricity is within the allowable range in the entire range by the calculation of the balancing, a temperature distribution occurs in the rotor rotating at the rotation speed during the steady operation, and the high temperature region D exists. Due to the uneven distribution of the high-temperature region, the shaft bending Q occurs in the rotor. This thermal imbalance is kept within an allowable range by adding and deleting balance weights. In order to change from the thermal unbalance to the mechanical balance S (θ), the allowable range of the mechanical balance S (θ) is calculated by the calculation by the Marcard method. With such an allowable range, a compromise optimization during steady-state (rated) operation can be quickly performed. Such an implementation is not only a factory test,
It is also carried out during on-site installation and on-site periodic repairs.

FIG. 7 shows a compensated balancing system by sharing data between a factory / design department and a site. In the design department, the PC 11 is arranged as a terminal (the center of the design or the design department). An EWS 12 for vibration analysis is arranged at the center of the site. P
The C11 and the EWS 12 for vibration analysis are bidirectionally connected via the LAN 13.

The vibration analysis EWS 12 receives data ε _m (for example, a voltage signal) output from a displacement meter or a vibration amplitude meter disposed at a plurality of points on one circumference of a rotating rotor bearing. In the EWS 12 for vibration analysis, a solution program by the Marcard method is resident, and the EWS 12 for vibration analysis is used.
Executes the axial vibration analysis of the rotor due to the addition or deletion of the balance weight, it calculates the vibration amplitude f _m according to equation (1). The parameters specific to the rotor of the formula (1) are transmitted from the PC 11 of the design department via the LAN 13 to the vibration analysis EW.
Input to S12. The model correction according to the measurement result and the re-analysis using the correction model based on the correction are executed by the PC 11. The EWS 12 for vibration analysis can be incorporated in the rotor dynamics system CAE15. The rotor dynamics system CAE15
Output the vibration analysis result. According to the analysis result, addition / deletion of the balance weight is executed by the operator. The PC 11 can be incorporated in a design system 14 that performs a new design with reference to the obtained data. The design system 14 can output the actual measurement results before and after the compromise.

The rotating machines are distributed on a global scale. With the knowledge and experience of a small number of designers and assemblers in the design department, repair and maintenance of a myriad of rotating machines is becoming impossible. Under the current situation where it is difficult to respond in a professional manner, a network of companies and organizations specializing in repair and repair is required. By sharing data, specialists perform autonomous repair and maintenance independently of the design department, and design guidelines that are more optimized based on compromises obtained from a huge number of sites. Is obtained. Such optimization can approach a non-compromised solution by the true least squares method, and can construct an evolving network.

[0044]

The method for compromising the balancing of a rotating machine according to the present invention and the method for adjusting the same can practically utilize the advantages of the Marcard method, which can surely obtain a solution by applying the compensated balancing. .

[Brief description of the drawings]

FIGS. 1A, 1B, and 1C are conceptual views showing Nyquist lines in FIGS. 1B and 1C for the bearing of FIG. 1A.

FIGS. 2A and 2B are analysis diagrams showing variables of a Nyquist diagram. FIG.

FIG. 3 is a front view showing an arrangement of balance weights.

FIG. 4 is an analysis diagram showing a steepest descent solution method of the Marcard method.

5 (a) and 5 (b) are Nyquist diagrams showing an embodiment of a compromise solution method for a rotating machine according to the present invention.

FIG. 6 is an axial projection view showing thermal imbalance.

FIG. 7 is a system block diagram showing an embodiment of a method for adjusting a rotary machine according to the present invention.

FIG. 8 is a graph showing a resonance avoidance criterion.

FIG. 9 is a table showing allowable error criteria.

FIG. 10 is a table showing evaluation criteria.

[Explanation of symbols]

 11: first computer 12: second computer

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Katsuya Yamashita 2-1-1, Shinhama, Arai-machi, Takasago-shi, Hyogo F-term in Takasago Research Laboratory, Mitsubishi Heavy Industries, Ltd. (Reference) 5B046 AA05 CA03 JA01 5B056 BB71 BB91 HH00

Claims (10)

[Claims] 1. A mass _m i (i is 1 to n, n is 1 or a positive integer greater than 1) and a radius _{r i} and the mass _{m i} of the position of mounting the weight _{W i} having a rotating body Calculating a first effect vector f _m (θ) corresponding to m axial positions by using the rotation speed of the rotator as a constant and using arbitrary rotation angle position coordinates θ of the surface portion of the rotator as variables. Calculating the residual square sum S in a direction minimizing by the Marcard method with respect to the function having f _m (θ) as a multivariable, the S being within an allowable range in a rotational speed region to be evaluated for vibration. weight W _i around the angular position coordinate of the rotation axial line of θ entering
^(K) . Compromise balancing solution for rotating machines, including: 2. The method according to claim 1, wherein the step of calculating_iTake
A second effect vector for the unrotated rotator
And measuring the second effect vector as ε_mS can be expressed as
Equation: S = Σ (ε_m−f_m)²  The compromise of the rotating machine of claim 1 characterized by:
Solution method. 3. A compromise balancing solution for a rotating machine according to claim 1, wherein said axial position is a fixed and fixed constant. 4. The compromise balancing solution of a rotating machine according to claim 3, wherein said mass _mi is a set and fixed constant. 5. The solution according to claim 1, wherein the number of rotations is plural. 6. A compromise solution for a rotating machine according to claim 5, wherein said axial position and said mass _mi are set and fixed constants. 7. Arranging a first computer in correspondence with a design department, arranging a second computer in correspondence with a field department, connecting the first computer and the second computer to store data. The first computer and the second computer
Be shared computer, the mass m _{i (i} is 1 to n, n is 1 or a positive integer greater than 1) the radius r _i of the position of mounting the weight W _i having a rotating body mass m _i and the speed and the constant of the rotary body, the weight W _i first effect vector corresponding to the m-axis direction position and an arbitrary rotational angular position coordinates θ variable surface portion of the rotary body defined by the arrangement of calculating f _m (θ); calculating a minimum square sum S in a direction to minimize the least square sum S by a Marcard method with respect to a function having the f _m (θ) as a multivariable function; around the angular position coordinate of the rotation axial line of the weight W _i that the at area S falls within the allowable range θ
^(K) ; and measuring a second effect vector on site with respect to the rotating body, wherein the data includes the second effect vector. 8. The method for adjusting a rotating machine according to claim 7, wherein said data further includes said θ ^(k) . 9. The method for adjusting a rotating machine according to claim 8, wherein said on-site department is plural. 10. A weight W having the following steps: mass m _i (i is 1 to n, n is 1 or a positive integer greater than 1).
wherein a radius r _i of the position for attaching the _i on the rotating body mass m _i
And the number of rotations of the rotator as constants, and using the arbitrary rotation angle position coordinate θ of the surface portion of the rotator as a variable, calculate a first effect vector f _m (θ) corresponding to m axial positions. Calculating the residual sum of squares S in the direction of minimizing the function using the f _m (θ) as a multivariable by the Marcard method, where S is an allowable range in a rotational speed region to be evaluated for vibration. around the angular position coordinate of the rotation axial line of the weight W _i that fall within θ
^(K) a recording medium having a program input to a first computer for executing a process including: obtaining a solution of the constant and the variable of the program of the recording medium; An adjustment network for a rotary machine, comprising: a network formed from a communication medium that bidirectionally connects the first computer and the plurality of second computers to communicate the constant and the solution of the variable.