CN112668125A - Method, system, medium and device for improving evaluation precision of incomplete small arc - Google Patents

Abstract

The invention relates to a method, system, medium and equipment for improving the evaluation accuracy of non-complete small circular arcs, comprising: determining the first optimal smoothing coefficient of a pre-established triple exponential smoothing prediction model; ; Fit the change trend curve of the smoothing coefficient with the best smoothing coefficient, and obtain the smoothing coefficient of the predicted period outside the data point set according to the fitting formula; use the smoothing coefficient of the predicted period outside the data point set to calculate and predict the outside of the measured data point set to determine whether the preset number of forecast periods is reached, and reverse the order of the data point set; combine the original measured data point set and the predicted data point set to form a new data point set, and use the new data point set to use the evaluation method The evaluation of curvature parameters is carried out to improve the evaluation accuracy of non-complete small arcs. The invention can improve the accuracy of the evaluation of the curvature radius parameter, and effectively improve the stability and accuracy of the evaluation of the curvature radius parameter of the incomplete small arc.

Description

Method, system, medium and equipment for improving evaluation accuracy of non-holonomic small arc

technical field

The invention relates to the technical field of performance detection of mechanical parts, in particular to a method, system, medium and equipment for improving the evaluation accuracy of non-complete small circular arcs.

Background technique

The non-complete small arc refers to the arc or arc surface with the corresponding center angle of the arc contour less than 120° and the radius of curvature ranging from 0 to 25 mm. The non-complete small arc profile is widely used in the defense technology industry and precision manufacturing due to its special properties: for example, the front and rear edge profiles of aero-engine blades have a direct impact on the aerodynamic performance of the engine and the fatigue performance of the blades; The tool tip directly determines its machining accuracy; the edge chamfering of some precision parts is used to remove machining burrs and improve the stress concentration of parts. As the core parameter of the small arc profile, the curvature radius parameter is very important to the production performance of the workpiece, so there is an urgent need for accurate measurement and evaluation. However, due to its non-integrity and small arc radius, it is difficult to measure and evaluate, and there has been controversy over the accuracy of curvature radius parameter estimation.

The main arc parameters that affect the evaluation accuracy are the theoretical radius of the arc, the center angle of the arc contour point set and the contour error of the arc contour point set. In general, the smaller the theoretical radius of the arc, the greater the difficulty and error of evaluation; the smaller the center angle of the arc contour point set, the worse the accuracy of the evaluation of the curvature radius; the contour error of the contour point set The larger the value, the lower the evaluation accuracy.

Among the above three influencing factors, the arc radius is the required target value and cannot be adjusted. Therefore, scholars generally start from the center angle of the arc contour point set and the contour error of the arc contour point set, and improve the evaluation accuracy by reducing the influence of the contour error or increasing the center angle. For example, Guevara et al. proposed a robust geometric method for fitting a set of points based on the mean absolute error. This method minimizes the sum of the geometric distances to the data points, and uses a fast iterative algorithm based on gradients or second derivatives. The right-hand derivative determines the new direction of iteration, fits the data points to an arc, and then finds the radius. The algorithm can be used as an alternative to conventional algorithms in terms of computational efficiency. In principle, the influence of contour error on the evaluation process is reduced, the robustness of the algorithm is enhanced, and it is insensitive to outliers and data noise, thereby improving the evaluation accuracy. . Fei et al. started from the perspective of the center of the non-complete small arc, and adopted a bidirectional prediction method based on radial basis function neural network (RBFNN), which regarded the observation data as a time series, and extended it in both directions, and then interpolated it. , by increasing the length of the arc to increase the center angle of the arc profile, and after training and learning with a large amount of data, it is finally confirmed that the stability and accuracy of the method fitting are far better than those before prediction.

The above methods have good applications in some specific fields and working conditions, but they all have certain limitations. For example, the bidirectional prediction method based on Radial Basis Function Neural Network (RBFNN) requires a huge amount of arc profile data samples. not tall.

SUMMARY OF THE INVENTION

In view of the above problems, the purpose of the present invention is to provide a method, system, medium and equipment for improving the evaluation accuracy of non-complete small arcs. The evaluation of the processed data point set can improve the accuracy of the evaluation of the radius of curvature parameter, and effectively improve the stability and accuracy of the evaluation of the radius of curvature of the non-complete small arc.

In order to achieve the above object, the present invention adopts the following technical solutions: a method for improving the evaluation accuracy of non-complete small circular arcs, comprising:

Step 1), determine the first optimal smoothing coefficient α ₁ of the pre-established triple exponential smoothing prediction model;

Step 2), further determine the remaining n-k optimal smoothing coefficients; n is the number of measured sample points, k is the number of periods of the time series, k<n;

Step 3), fitting n-k best smoothing coefficients to the variation trend curve of smoothing coefficient α, and obtaining the smoothing coefficient of the predicted period number outside the data point set according to the fitting formula;

Step 4), utilize the smoothing coefficient of prediction period number outside the data point set to calculate and predict the data point outside the measured data point set;

Step 5), judging whether the preset number of forecast periods is reached, and if it is reached, reverse the order of the data point set, and repeat steps 1) to 4);

Step 6): Combine the original measured data point set and the predicted data point set to form a new data point set, and use the evaluation method to evaluate the curvature parameters of the new data point set, thereby improving the evaluation accuracy of the incomplete small arc.

Further, in the step 1), the method for determining the first optimal smoothing coefficient α ₁ is: using the existing measured small circular arc outline data abscissa point set and ordinate point set, the number of measured sample points is n, and The ordinal number is regarded as the number of periods of the time series; it is assumed that k periods of data are fixed for each forecast; starting from the first period of data, the first calculation performs three exponential smoothing forecasts from the first period of data to the kth period of data, and moves to the right Calculation; the smoothing coefficient α traverses and searches in the value space [0,1], sets the step size of each calculation, and obtains several groups of k+1 period predicted values, and compares the k+1 period predicted value with the k+1 period measured The data is compared, and the optimal smoothing coefficient α ₁ is determined according to the principle of minimizing the sum of squares of errors.

Further, in the step 2), the determination method is: remove the first phase data, use the second phase to k+1 phase measured data, repeat the first optimal smoothing coefficient determination method, and obtain the second optimal smoothing coefficient coefficient α ₂ , and finally the nkth best smoothing coefficient α _nk is obtained, and nk best smoothing coefficients are obtained in total.

Further, in the described step 4), the calculation method of predicting the data points outside the measured data point set is: when the calculation moves to the n+1 period prediction point calculated from the nkth period to the nth period measured arc data point set , the smoothing coefficient α _n-k+1 obtained after fitting is substituted into the triple exponential smoothing prediction model, the prediction points are merged into the data point set of the previous round of calculation after each prediction, and the data points of the earliest period are eliminated, The new set of data points obtained is used for the next round of calculation, and finally, the data of the n+mk-1 to n+m-1 periods are used, and α _n-k+m is used to calculate the predicted value points of the n+mth period.

Further, the triple exponential smoothing prediction model is:

和

为第m期预测点的横纵坐标值，t为实测数据的期数，m为预测的步数，a_t1，b_t1，c_t1和a_t2，b_t2，c_t2分别为圆弧数据点x方向和y方向预测模型参数。in,

and

is the abscissa and ordinate value of the prediction point of the mth period, t is the period of the measured data, m is the number of predicted steps, a _t1 , b _t1 , c _t1 and a _t2 , b _t2 , c _t2 are the arc data points respectively The model parameters are predicted in the x and y directions.

Further, the parameters a _t1 , b _t1 , and c _t1 are obtained in the same way as the parameters a _t2 , b _t2 , and c _t2 , and the formula is as follows:

A system for improving the evaluation accuracy of non-complete small arcs, comprising: a first determination module, a second determination module, a fitting module, a data point acquisition module outside the point set, a judgment module and an evaluation module;

The first determining module determines the first optimal smoothing coefficient α ₁ of the pre-established triple exponential smoothing prediction model;

The second determination module further determines the remaining n-k optimal smoothing coefficients; n is the number of measured sample points, k is the number of periods of the time series, and k<n;

The fitting module fits the n-k best smoothing coefficients to obtain the change trend curve of the smoothing coefficient α, and obtains the smoothing coefficients of the predicted periods outside the data point set according to the fitting formula;

The data point acquisition module outside the point set uses the smoothing coefficient of the forecast period outside the data point set to calculate and predict the data points outside the measured data point set;

The judging module judges whether the preset number of prediction periods is reached, and if it is reached, the data point set is reversed, and the first determining module, the second determining module, the fitting module and the data point acquiring module outside the point set are repeatedly executed;

The evaluation module combines the original measured data point set and the predicted data point set to form a new data point set, and uses the evaluation method to evaluate the curvature parameters of the new data point set, thereby improving the evaluation accuracy of the incomplete small arc.

Further, in the first determination module, the method for determining the first optimal smoothing coefficient α ₁ is: using the existing measured small arc outline data abscissa point set and ordinate point set, the number of measured sample points is n, The ordinal number is regarded as the number of periods of the time series; it is assumed that k periods of data are fixed for each forecast; starting from the first period of data, the first calculation will perform three exponential smoothing forecasts on the first period of data to the kth period of data, rightwards Mobile calculation; the smoothing coefficient α traverses and searches in the value space [0,1], sets the step size of each calculation, and obtains several groups of k+1 period predicted values, and compares the k+1 period predicted value with the k+1 period Comparing the measured data, the optimal smoothing coefficient α ₁ is determined according to the principle of minimizing the sum of squares of errors.

A computer-readable storage medium storing one or more programs comprising instructions that, when executed by a computing device, cause the computing device to perform any of the above methods.

A computing device comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the The one or more programs include instructions for performing any of the methods described above.

Because the present invention adopts the above technical scheme, it has the following advantages: 1, the present invention reduces the research object to a small non-complete arc with a central angle of less than 60°, which enhances the stability of the evaluation of the curvature radius parameter of this type of arc, The effect of improving the evaluation accuracy is more obvious. 2. Aiming at the problem that the selection of the smoothing coefficient α is too subjective and the sensitivity is too weak in the general triple exponential smoothing prediction process, the present invention adds an adaptive improvement algorithm on the basis of the triple exponential smoothing, and searches in real time through the trend change of its own data. The best smoothing coefficient for the prediction. The original cumulative iteration is changed to a fixed sample point moving iteration, which reduces the influence of the change in the number of predicted sample points in the prediction process, makes the change trend of the optimal smoothing coefficient easier to analyze, and obtains more accurate prediction points. 3. The present invention uses the comparative analysis of the evaluation results of different data under the same evaluation method to verify the prediction angle that this method has the best effect of improving the evaluation accuracy; the comparative analysis of the evaluation results under different evaluation methods of the same group of data is used to verify the universality of the law's application.

Description of drawings

FIG. 1 is a schematic overall flow diagram of a method in an embodiment of the present invention.

2 is a schematic diagram of a fixed sample moving calculation in an embodiment of the present invention;

Fig. 3 is the trend diagram of the influence of the center angle of the small circular arc on the evaluation in the embodiment of the present invention;

Fig. 4 is the arc fitting diagram after using three exponential smoothing alone;

5 is a fitting diagram of an adaptive cubic exponential smoothing coefficient and a later derivation value diagram in the embodiment of the present invention;

FIG. 6 is a fitting effect diagram of bidirectional prediction by the adaptive triple exponential smoothing method in the embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, embodiments of the present invention. Based on the described embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art fall within the protection scope of the present invention.

The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

The invention provides a method for improving the evaluation accuracy of a non-complete small circular arc, which is aimed at evaluating the curvature radius parameter of a small circular arc with a central angle opposite to a contour of 60° or less. The invention adopts an adaptive dynamic cubic exponential smoothing model, that is, an adaptive algorithm is added on the basis of the traditional cubic exponential smoothing prediction model for optimization. The adaptive triple exponential smoothing method does not substitute all the points into the direct prediction like the traditional triple exponential smoothing method, but according to the existing data point set, starting from the first period of data, only some of the points are intercepted for triple exponential smoothing ; After calculating the forecast data of this period, move forward one step to calculate, and each time you move, the data of the first period will be removed and added to the measured data of the next period. At the same time, it traverses the value space of the smoothing coefficient α, performs fixed-step calculation, and obtains a series of forecast points. All forecast points are compared with the current measured points. According to the principle of minimizing the sum of squares of errors, the forecast point with the smallest error is obtained. , and the smoothing coefficient value that predicts this point is the real-time best smoothing prediction value. After the training of the previous historical data, the change trend curve of the smoothing coefficient α can be fitted. When forecasting outside the point set, due to the lack of comparison of the measured values, it is necessary to calculate the later smoothing coefficient value according to the fitting curve equation of α. Directly use this smoothing coefficient to participate in the calculation and prediction of data points outside the measured data point set. The invention increases the center angle corresponding to the contour data point set based on the method of the adaptive cubic exponential smoothing prediction model, thereby improving the evaluation accuracy of the curvature radius parameter of the non-complete small arc.

As shown in Figure 1 and Figure 2, the method of the present invention specifically comprises the following steps:

Step 1), determine the first optimal smoothing coefficient α ₁ of the pre-established triple exponential smoothing prediction model;

Specifically: using the existing measured small arc outline data abscissa point set X and ordinate point set Y, X=[x ₁ , x ₂ , x ₃ ,...], Y=[y ₁ , y ₂ , y ₃ ,...], the number of measured sample points is n, and the ordinal number is regarded as the number of periods of the time series; it is assumed that k periods (k<n) data are fixed for each forecast, starting from the first period of data, the first The calculation performs three exponential smoothing predictions on the data from the first period to the kth period, and shifts the calculation to the right. The smoothing coefficient α is traversed and searched in the value space [0,1], the step size of each calculation is set, and several groups of k+1 period predicted values are obtained, and the k+1 period predicted value is compared with the k+1 period measured data , according to the principle of minimizing the sum of squares of errors, determine the optimal smoothing coefficient α ₁ ;

Step 2), further determine the remaining n-k optimal smoothing coefficients;

Specifically: remove the data of the first period, use the measured data from the second period to the k+1 period, repeat step 1), obtain the second best smoothing coefficient α ₂ , and so on, and finally obtain the nkth best smoothing coefficient coefficients α _nk , a total of nk optimal smoothing coefficients are obtained.

Step 3), fitting n-k best smoothing coefficients to the variation trend curve of smoothing coefficient α, and obtaining the smoothing coefficient of the predicted period number outside the data point set according to the fitting formula;

In the prediction outside the point set, due to the lack of comparison of the measured values, it is necessary to calculate the later smoothing coefficient value according to the change trend curve of the smoothing coefficient α and the fitting formula.

Specifically: fitting α ₁ to α _nk to obtain a fitting curve and a fitting formula, and calculating to obtain α _n-k+1 , α _n-k+2 , ..., α _n-k+m ; where m The number of periods forecasted for outside the set of data points.

Step 4), utilize the smoothing coefficient of prediction period number outside the data point set to calculate and predict the data point outside the measured data point set;

Specifically: when the calculation moves to the calculation of the forecast point of the n+1 period using the measured arc data point set from the nkth period to the nth period, the smoothing coefficient α _n-k+1 obtained after fitting is substituted into the three-time exponential smoothing prediction Model, after each prediction, the prediction points are merged into the data point set of the previous round of calculation, and the data points of the earliest period are eliminated, and the new data point set obtained is used for the next round of calculation, and is kept for each prediction. The number of points is always k, and so on, using the data from the n+mk-1 period to the n+m-1 period, and substituting α _n-k+m for triple exponential smoothing to calculate the predicted value of the n+mth period .

Step 5), determine whether the preset number of predicted periods is reached, and if so, reverse the order of the data point set, that is, the first period of data becomes the nth period, and the nth period becomes the first period, repeating steps 1) to steps 4). If the number of predicted periods is not reached, return to step 1) and repeat the execution.

Step 6), merge the original measured data point set and all predicted data points to form a new data point set, and use the evaluation method to evaluate the curvature parameter of the new data point set, which improves the evaluation accuracy of the incomplete small arc; wherein the evaluation method can be A mature evaluation method in the prior art is adopted, which will not be repeated here.

In the above steps, it is known that the abscissa point set X and the ordinate point set Y of the existing measured small arc outline data, then the triple exponential smoothing prediction model is:

和

为第m期预测点的横纵坐标值，t为原实测数据的期数，m为预测的步数，a_t1，b_t1，c_t1和a_t2，b_t2，c_t2分别为x方向和y方向预测模型参数。in,

and

is the abscissa and ordinate value of the prediction point of the mth period, t is the period of the original measured data, m is the number of prediction steps, a _t1 , b _t1 , c _t1 and a _t2 , b _t2 , c _t2 are the x-direction and y-direction prediction model parameters.

Among them, the prediction method using the triple exponential smoothing prediction model is:

对X进行三次平滑(Y同理)：Set smoothing coefficient α value and smoothing initial value

Smooth X three times (same for Y):

Then the formulas for obtaining the model parameters a _t1 , b _t1 , and c _t1 in the x-direction are as follows:

The y-direction prediction model parameters a _t2 , b _t2 , and c _t2 are the same;

Substitute a _t1 , b _t1 , c _t1 and a _t2 , b _t2 , c _t2 into formula (1) to obtain the forecast point of the first period, merge the obtained forecast point into the point set, and perform three times exponential smoothing again Prediction, set the number of predicted steps m to get the predicted value of m period. Figure 3 shows the effect of evaluating the new data point set by merging the predicted points predicted by triple exponential smoothing with the measured points.

By analyzing the results of three exponential smoothing predictions in Figure 3, it can be observed that although the accuracy of the data forecast in the first few periods is very high, there will be a more obvious shift after less than 3 periods of prediction, as shown in Figure 4 . It is analyzed that the reason for the deviation is that the selected smoothing coefficient α is too subjective and too weak in sensitivity. Aiming at this problem, the present invention adds an adaptive algorithm to the traditional cubic exponential smoothing method for optimization, such as steps 1) to 6), thereby solving the optimization problem of the smoothing coefficient.

The present invention also provides a system for improving the evaluation accuracy of an incomplete small arc, which includes: a first determination module, a second determination module, a fitting module, a data point acquisition module outside the point set, a judgment module and an evaluation module;

a first determination module, which determines the first optimal smoothing coefficient α ₁ of the pre-established triple exponential smoothing prediction model;

The second determination module further determines the remaining n-k optimal smoothing coefficients; n is the number of measured sample points, k is the number of periods of the time series, and k<n;

The fitting module fits the n-k best smoothing coefficients to the variation trend curve of the smoothing coefficient α, and obtains the smoothing coefficients of the predicted periods outside the data point set according to the fitting formula;

The data point acquisition module outside the point set uses the smoothing coefficient of the forecast period outside the data point set to calculate and predict the data points outside the measured data point set;

a judging module, for judging whether a preset number of prediction periods is reached, and if it is reached, the data point set is reversed, and the first determining module, the second determining module, the fitting module and the data point acquiring module outside the point set are repeatedly executed;

The evaluation module combines the measured data point set and the predicted data point set to form a new data point set, and uses the evaluation method to evaluate the curvature parameters of the new data point set, which improves the evaluation accuracy of the incomplete small arc.

In the above-mentioned embodiment, in the first determination module, the determination method of the _first optimal smoothing coefficient α1 is: using the existing measured small circular arc outline data abscissa point set and ordinate point set, the number of measured sample points is: n, and the ordinal number is regarded as the number of periods of the time series; it is assumed that k periods of data are fixed for each forecast; starting from the first period of data, the first calculation will perform three exponential smoothing forecasts from the first period of data to the kth period of data, Move to the right and calculate; the smoothing coefficient α traverses the search in the value space [0,1], sets the step size of each calculation, and obtains several groups of k+1 period predicted values. Comparing the measured data in Phase 1, according to the principle of minimizing the sum of squares of errors, determine the optimal smoothing coefficient α ₁ .

The present invention also provides a computer-readable storage medium storing one or more programs, the one or more programs comprising instructions that, when executed by a computing device, cause the computing device to perform any of the above methods.

The present invention also provides a computing device comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, a The or more programs include instructions for performing any of the above-described methods.

Example:

In this embodiment, the evaluation results before and after prediction of different arc data are compared.

Nine sets of data points with theoretical radii of 0.05mm, 1mm and 25mm, and the gradients of the corresponding central angles of the contours are respectively 15°, 30° and 45°. The number of point sets is 50. The adaptive cubic exponential smoothing method is used to perform bidirectional prediction of the data, and then the least squares method is used to evaluate the curvature radius. Compared with the direct use of the least squares method, the change of the evaluation accuracy of the curvature radius in the two cases is analyzed.

The principle of this method is to use the previous data to analyze the change trend of the smoothing coefficient, and then adjust the later smoothing coefficient through this trend, and use the obtained smoothing coefficient to predict the data points outside the measured contour data point set. In the process of searching for the best smooth coefficient in each iterative traversal in the early stage, the step size is set to 0.001, and the value space of α is 0 to 1. In order to improve the iteration speed and reduce unnecessary calculations, α can be reduced according to the actual situation The value range of ; the k value of the number of sample points used for prediction each time is fixed at 20, so 30 α values can be obtained from the 50 measured point set data in the entire moving iteration process, and the change trend of the α value can be observed. Fit it to obtain the change curve and the change formula, and use the change formula to obtain the smooth coefficient used for the prediction of the data outside the point set in the later stage, and use the obtained α value to continue to carry out the fixed number of samples to move the calculation to get the accurate contour point. Out-of-set prediction points. The changing trend and fitting of the smoothing coefficient α are shown in Figure 5.

In the process of smooth coefficient fitting, the polyfit function of matlab is used to fit it, and the variation trend of α in the later period of different fitting times has a certain difference. After comprehensive analysis of multiple sets of data, among quadratic fitting, cubic fitting and quartic fitting, choosing cubic fitting has higher goodness of fit.

As shown in Figure 6, using the same set of data for prediction, compared with the prediction results of using triple exponential smoothing alone in Figure 4, the prediction error after adding adaptive optimization is significantly reduced, which greatly slows down the speed of error accumulation and makes the later prediction points relatively The offset from the ideal circle is reduced a lot, the predicted angle is increased, and the evaluation accuracy of the radius of curvature is improved to a certain extent.

After the evaluation and analysis of the above data, the feasibility of applying the self-adaptive cubic exponential smoothing prediction method of the present invention to the evaluation of the curvature radius parameters of non-complete small arcs is verified. This method has the effect of improving the evaluation accuracy. The comparison and analysis of the evaluation results under different evaluation methods of the same group of data verifies the universality of the application of the method.

To sum up, the present invention aims at the problem of low evaluation accuracy of curvature radius parameters caused by poor integrity of small circular arcs in metrology, and uses simulation to analyze the influence trend of the center angle of the small circular arc profile on the evaluation process. According to the results of the simulation analysis, the present invention The method based on the adaptive cubic exponential smoothing prediction model increases the central angle of the contour data point set, thereby improving the evaluation accuracy of the curvature radius parameter. The invention adds an adaptive optimization algorithm on the basis of the triple exponential smoothing method, solves the problem that the optimal smoothing coefficient is difficult to determine in the prediction process, optimizes the traditional prediction model and the selection criteria of each parameter, and improves the accuracy of prediction points. The experimental results show that for the incomplete small arc contour data point set below 60°, using this method to expand the center angle of the contour by about 5° is the best for the later evaluation process; and it can be used as a preliminary data processing method. Different evaluation methods are used to improve the evaluation accuracy of the radius of curvature parameter to different degrees.

As will be appreciated by those skilled in the art, the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device produce Means for implementing the functions specified in a flow or flow of a flowchart and/or a block or blocks of a block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions The apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that The instructions provide steps for implementing the functions specified in the flow or blocks of the flowcharts and/or the block or blocks of the block diagrams.

Claims (10)

1. a method for improving the evaluation accuracy of non-complete small arc, is characterized in that, comprises: Step 1), determine the first optimal smoothing coefficient α ₁ of the pre-established triple exponential smoothing prediction model; Step 2), further determine the remaining n-k optimal smoothing coefficients; n is the number of measured sample points, k is the number of periods of the time series, and k<n; Step 3), fitting n-k best smoothing coefficients to the variation trend curve of smoothing coefficient α, and obtaining the smoothing coefficient of the predicted period number outside the data point set according to the fitting formula; Step 4), utilize the smoothing coefficient of prediction period number outside the data point set to calculate and predict the data point outside the measured data point set; Step 5), judging whether the preset number of forecast periods is reached, and if it is reached, reverse the order of the data point set, and repeat steps 1) to 4); Step 6): Combine the original measured data point set and the predicted data point set to form a new data point set, and use the evaluation method to evaluate the curvature parameters of the new data point set, thereby improving the evaluation accuracy of the incomplete small arc. 2. method as claimed in claim 1 is characterized in that, in described step 1), the first best smooth coefficient α ₁ determination method is: utilize existing actual measurement small circular arc outline data abscissa point set and ordinate Point set, the number of measured sample points is n, and the ordinal number is regarded as the number of periods of the time series; it is assumed that k periods of data are fixed for each forecast; starting from the first period of data, the first calculation will The k-period data is predicted by exponential smoothing three times, and the calculation is shifted to the right; the smoothing coefficient α is traversed and searched in the value space [0, 1], and the step size of each calculation is set to obtain several groups of k+1 period prediction values. The predicted value of the +1 period is compared with the measured data of the k+1 period, and the optimal smoothing coefficient α ₁ is determined according to the principle of minimizing the sum of squares of errors. 3. method as claimed in claim 2 is characterized in that, in described step 2), determining method is: remove the 1st phase data, adopt the 2nd phase to the k+1 phase measured data, repeat the first best In the smoothing coefficient determination method, the second optimal smoothing coefficient α ₂ is obtained, and finally the nkth optimal smoothing coefficient α _nk is obtained, and nk optimal smoothing coefficients are obtained in total. 4. method as claimed in claim 1, is characterized in that, in described step 4), the calculation method of the data point outside the predicted actual measurement data point set is: when the calculation moves to adopt the nth period to the nth period actual measurement arc When calculating the forecast point of the n+1th period from the data point set, the smoothing coefficient α _n-k+1 obtained after fitting is substituted into the three-time exponential smoothing forecast model, and the forecast point is merged into the data calculated in the previous round after each forecast. point set, and remove the data points of the earliest period, use the new data point set obtained for the next round of calculation, and finally use the data of the n+mk-1 to n+m-1 periods, and substitute it into α _{n-k+ m} Calculates the predicted value point of the n+mth period. 5. method as claimed in claim 1, is characterized in that, described three times exponential smoothing prediction model is:

in,

and

is the abscissa and ordinate value of the prediction point of the mth period, t is the period of the measured data, m is the number of predicted steps, a _t1 , b _t1 , c _t1 and a _t2 , b _t2 , c _t2 are the arc data points respectively The model parameters are predicted in the x and y directions. 6. The method of claim 5, wherein the parameters a _t1 , b _t1 , and c _t1 are obtained in the same manner as the parameters a _t2 , b _t2 , and c _t2 , and the formula is as follows:

7. A system for improving the evaluation accuracy of non-complete small arcs, characterized in that it comprises: a first determination module, a second determination module, a fitting module, a data point acquisition module outside the point set, a judgment module and an evaluation module; The first determining module determines the first optimal smoothing coefficient α ₁ of the pre-established triple exponential smoothing prediction model; The second determination module further determines the remaining n-k optimal smoothing coefficients; n is the number of measured sample points, k is the number of periods of the time series, and k<n; The fitting module fits the n-k best smoothing coefficients to obtain the change trend curve of the smoothing coefficient α, and obtains the smoothing coefficients of the predicted periods outside the data point set according to the fitting formula; The data point acquisition module outside the point set uses the smoothing coefficient of the forecast period outside the data point set to calculate and predict the data points outside the measured data point set; The judging module judges whether the preset number of prediction periods is reached, and if it is reached, the data point set is reversed, and the first determining module, the second determining module, the fitting module and the data point acquiring module outside the point set are repeatedly executed; The evaluation module combines the original measured data point set and the predicted data point set to form a new data point set, and uses the evaluation method to evaluate the curvature parameters of the new data point set, thereby improving the evaluation accuracy of the incomplete small arc. 8. system as claimed in claim 7, is characterized in that, in described first determination module, the first best smooth coefficient α ₁ determination method is: utilize existing actual measurement small circular arc outline data abscissa point set and vertical Coordinate point set, the number of measured sample points is n, and the ordinal number is regarded as the number of periods of the time series; it is assumed that k periods of data are fixed for each forecast; starting from the first period of data, the first calculation will The kth period data is predicted by exponential smoothing three times, and the calculation is shifted to the right; the smoothing coefficient α is traversed and searched in the value space [0, 1], the step size of each calculation is set, and several groups of k+1 period prediction values are obtained. The predicted value of the k+1 period is compared with the measured data of the k+1 period, and the optimal smoothing coefficient α ₁ is determined according to the principle of minimizing the sum of squares of errors. 9. A computer-readable storage medium storing one or more programs, characterized in that the one or more programs comprise instructions that, when executed by a computing device, cause the computing device to perform as claimed in the claims Any of the methods described in 1 to 6. 10. A computing device, comprising: one or more processors, a memory, and one or more programs, wherein one or more programs are stored in the memory and configured as the one or more programs A processor executes the one or more programs comprising instructions for performing any of the methods of claims 1-6.

Images