Numerical control machine tool performance evaluation method based on multilayer evidence reasoning

Abstract

The invention is suitable for the technical field of numerical control machine tool performance evaluation, and provides a numerical control machine tool performance evaluation method based on multilayer evidence reasoning, comprising the following steps: acquiring numerical control machine tool data; establishing a comprehensive performance evaluation system of the whole numerical control machine tool, wherein the comprehensive performance evaluation system comprises first-level indexes and second-level indexes; constructing a multi-layer evidence reasoning model, and setting initial parameters of the model; second-level index evaluation is achieved through the built multi-layer evidence reasoning model, the evaluation result serves as input of the previous layer evidence reasoning model, and finally the performance state of the whole numerical control machine tool is evaluated; and outputting performance state evaluation values of the whole numerical control machine tool, and connecting into a curve according to a time sequence. According to the method, the numerical control machine tool performance evaluation model based on multilayer evidence reasoning is constructed, the collected multi-dimensional data can be comprehensively analyzed, the performance condition of the numerical control machine tool can be considered from multiple angles and levels, the overall performance of the numerical control machine tool can be comprehensively evaluated, and a more scientific and reasonable evaluation scheme is provided.

Description

Technical Field

[0001] The invention belongs to the technical field of CNC machine tool performance evaluation, and in particular relates to a CNC machine tool performance evaluation method based on multi-layer evidence reasoning. Background Art

[0002] As the core equipment of the manufacturing industry, CNC machine tools play an indispensable role in aerospace, automobile manufacturing, power generation equipment, large ships, weapons and equipment, rail transportation and other fields. However, due to the changing working conditions and complex operating environment, the performance of CNC machine tools gradually deteriorates during service, affecting the processing accuracy and production efficiency. Therefore, it is particularly important to establish a scientific CNC machine tool service performance evaluation system.

[0003] CNC machine tools are complex systems that integrate mechanical, electrical, and hydraulic functions. Their internal structure consists of multiple subsystems and components. To ensure efficient and stable operation, the safety and reliability of each component are crucial. When evaluating performance, the operation of each component must be fully considered to obtain an accurate and complete performance status. To this end, a multi-layer performance status evaluation indicator system needs to be constructed, which can achieve the whole machine evaluation of CNC machine tools "from bottom to top, from small to large, and from point to surface". Through the fusion of multi-source heterogeneous information, a qualitative analysis of the whole CNC machine tool is performed. However, multi-source heterogeneous information is multi-dimensional and multi-faceted, independent of each other and complex in style, resulting in confusing data and lack of support between each other. Therefore, how to effectively integrate this information is a key issue that needs to be solved. In addition, as a typical CNC basic manufacturing equipment, the performance evaluation of CNC machine tools needs to be interpretable from process to result, which means that the evaluation not only relies on the real-time monitoring data of CNC machine tools, but also needs to integrate the experience and knowledge accumulated by industry experts over the years to ensure that the evaluation is both scientific and close to the actual operation of the equipment.

[0004] At present, performance evaluation methods are mainly divided into model-based, data-driven and qualitative knowledge-based methods. Among them, although the model-based method is suitable for systems with clear mathematical expressions of working principles and failure mechanisms, CNC machine tools face many challenges in establishing accurate mathematical models due to their complex structure, strong correlation and high nonlinearity. The data-driven method focuses on using a large amount of historical and real-time observation data of engineering systems to build a nonlinear dynamic evaluation framework. Compared with analytical models, its advantage is that it can avoid the problem of over-reliance on the precise mathematical model of the system. However, during the actual service life of CNC machine tools, it is difficult to collect data that meets the model training requirements, and the model lacks interpretability, resulting in an inability to accurately evaluate system performance. The qualitative knowledge-based method mainly relies on personal experience, domain knowledge and other non-quantitative information. Although it can reflect the health status of the system and use the knowledge or experience of experts to establish a model, it cannot effectively use quantitative data, resulting in a strong subjectivity in the evaluation results. Moreover, these three methods lack the ability to handle various uncertain information, and it is difficult to effectively solve the problem of CNC machine tool status evaluation. Therefore, a performance evaluation method that is interpretable and can integrate multi-source heterogeneous information is needed.

[0005] The evidence reasoning (ER) method is a semi-quantitative information method that can effectively utilize expert knowledge and quantitative data, combining the advantages of both. It can achieve performance evaluation through the calculated confidence level and the defined evaluation level, and the model has a certain degree of interpretability, so that the evaluation results can accurately reflect the performance of the system.

[0006] Based on the above background, the present invention proposes a CNC machine tool performance evaluation method based on multi-layer evidence reasoning. Summary of the invention

[0007] The purpose of the present invention is to provide a CNC machine tool performance evaluation method based on multi-layer evidence reasoning, aiming to solve the problems raised in the above background technology.

[0008] The purpose of the present invention is achieved through the following technical solutions:

[0009] A method for evaluating the performance of a CNC machine tool based on multi-layer evidence reasoning includes the following steps:

[0010] Step 1: Obtain CNC machine tool data;

[0011] Step 2: Establish a comprehensive performance evaluation system for CNC machine tools, including primary and secondary indicators;

[0012] Step 3: Construct a multi-layer evidence reasoning model and set the initial parameters of the model;

[0013] Step 4: Use the constructed multi-layer evidence reasoning model to evaluate the secondary indicators, use the evaluation results as the input of the previous layer of evidence reasoning model, and finally evaluate the performance status of the CNC machine tool;

[0014] Step 5: Output the performance status evaluation value of the entire CNC machine tool and connect it into a curve according to the time series.

[0015] Furthermore, in step 2, the first-level indicators include reliability based on fault data, accuracy retention, reliability based on performance degradation, and high efficiency; each first-level indicator includes multiple second-level indicators: reliability based on fault data includes 3 second-level indicators: mean repair interval, mean repair time, and mean first failure time; accuracy retention includes 4 second-level indicators: quantitative evaluation value of accuracy retention of X, Y, and Z axes, and quantitative evaluation value of spatial accuracy retention; reliability based on performance degradation includes 2 second-level indicators: comprehensive observation value of processing performance and reliability of processing performance; high efficiency includes 4 second-level indicators: maximum speed of X, Y, and Z axes, and maximum speed of spindle.

[0016] Furthermore, in step 3, the CMA-ES algorithm is used to optimize the initial parameters of the multi-layer evidence reasoning model. During the parameter optimization process, the monotonicity index and the correlation index are used as the objective function of the multi-layer evidence reasoning model. The parameters corresponding to the maximum value of the objective function are the optimal parameters of the model.

[0017] Furthermore, the objective function is constructed as follows:

[0018] The formula for the monotonicity index is:

[0019] Formula 1:

[0020] Among them, dF i Indicates the difference in characteristic sequence values; Num of dF i >0 and Num of dF i <0 indicates the number of positive and negative values ​​in the feature sequence, respectively. i represents the total number of features;

[0021] The formula for the correlation index is:

[0022] Formula 2:

[0023] Among them, y t represents the performance characteristic value in the tth cycle; Indicates the average value of the performance characteristic value; l t Represents the number of the tth sampling point; is the average value of the sampling point number; T tIndicates the total number of cycles; corresponds to the length of the performance characteristic curve;

[0024] Using Equation 1 and Equation 2, the objective function for parameter optimization of the multi-layer evidence reasoning model is constructed:

[0025] Formula 3:

[0026] Among them, Ω represents the parameters of the evaluation model; It represents the weight corresponding to the input indicator (i.e., evidence); T represents the number of indicators.

[0027] Furthermore, the specific steps of optimizing the objective function using the CMA-ES algorithm are as follows:

[0028] a. Set initial value: Set the initial parameter vector Ω 0 , initial covariance matrix C g , initial step length η, central expectation mean 0 , evolutionary generations g, population size λ;

[0029] b. Generate the initial population: The process of generating the population follows the following strategy:

[0030] Formula 4:

[0031] in, represents the i-th solution in the population, Represents multivariate normal distribution, mean g represents the central expectation of the g-th generation population, η g represents the update step size of the distribution;

[0032] c. Selection operation: This operation is used to select the first ε optimal solutions in the population, and select the optimal solution according to the value of the fitness function f(Ω);

[0033] d. Recombination operation: The recombinant operation is achieved by the following formula:

[0034] Formula 5:

[0035] Among them, mean g+1 represents the central expectation of the g+1 generation population, γ i represents the weight of the ith solution, represents the i-th solution among the λ solutions;

[0036] e. Update the covariance matrix: The covariance matrix C represents the equal probability density ellipse of the population distribution and is updated according to the following formula:

[0037] Formula 6:

[0038] Among them, a1 and aε represents the learning factor, q represents the evolution path, the initial value is 0, and q is updated according to the following formula:

[0039] Formula 7:

[0040] Among them, a q ≤1 is the evolution path parameter, and the step length η is updated according to the following formula:

[0041] Formula 8:

[0042] Among them, d η represents the damping coefficient, Represents Euclidean Normal Form The expectation, initially C 0 Set to the identity matrix I,q η and a η denote the conjugate evolution path and path parameter, respectively, q η Update according to the following formula: Formula 9:

[0043] f. Repeat equation 4 to equation 9 until the accuracy requirement is met and output the optimal parameters of the model.

[0044] Furthermore, in the multi-layer evidence reasoning model, when evaluating the first-level indicator, the CNC machine tool y is represented by the following confidence distribution:

[0045] Formula 10:

[0046] The following formula is then used to calculate the degree of influence of the input data on the performance status of the CNC machine tool:

[0047] Formula 11:

[0048] Among them, H n m Indicates the reference level of the mth indicator of the first-level indicator, It represents the confidence that the mth indicator attribute of the first-level indicator is evaluated as the reference level, n=1,...,N represents the number of reference levels, m=1,…,M represents the number of first-level indicators, and X(y) represents the utility value of CNC machine tool y after iterative fusion of attributes.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention constructs a CNC machine tool performance evaluation model based on multi-layer evidence reasoning, which can comprehensively analyze the collected multi-dimensional data and consider the performance of the CNC machine tool from multiple angles and levels, thereby realizing a comprehensive evaluation of the overall performance of the CNC machine tool. It solves the problems of the existing CNC machine tool performance evaluation method that it is difficult to effectively handle the multi-source heterogeneous information of the multi-layer indicator system and the weak interpretability of the evaluation results, and provides a more scientific and reasonable evaluation scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Flow chart of the method of the present invention.

[0052] Figure 2 It is a comprehensive performance evaluation system for CNC machine tools.

[0053] Figure 3 It is the basic structure of the multi-layer evidence reasoning model.

[0054] Figure 4 Optimization steps for CMA-ES.

[0055] Figure 5 This is a framework flow chart for the comprehensive performance evaluation of CNC machine tools.

[0056] Figure 6 This is the schematic diagram for the comprehensive performance evaluation of CNC machine tools. DETAILED DESCRIPTION

[0057] In order to have a clearer understanding of the technical features, purposes and beneficial effects of the present invention, the technical solution of the present invention is now described in detail below, but it should not be construed as limiting the applicable scope of the present invention.

[0058] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0059] An embodiment of the present invention provides a method for evaluating the performance of a CNC machine tool based on multi-layer evidence reasoning, the flow chart of which is as follows: Figure 1 As shown, the method comprises the following steps:

[0060] Step 1: Acquire CNC machine tool data to provide data support for the performance evaluation model of CNC machine tools;

[0061] In order to comprehensively evaluate the performance of CNC machine tools, it is necessary to collect multi-source data by installing multiple sensors during the service life of the machine tools, and combine quantitative data and qualitative data to judge the performance of the machine tools from multiple angles and dimensions. This is an important means to judge the changes in the performance status of the machine tools, which is of great significance to the optimization of CNC machine tool design schemes. It is also an important reference for users to select CNC machine tools with suitable working conditions, so as to accurately judge the state changes of machine tool performance. First, use the work log to record the time and number of machine tool failures, calculate the average repair interval time, average repair time and average first failure time; secondly, install a laser interferometer to collect the accuracy data of each axis, and obtain the quantitative evaluation value of accuracy retention through the accuracy retention calculation method; at the same time, use vibration, pressure, temperature, position, current and voltage sensors to collect machine tool operation status data, calculate the comprehensive observation value of processing performance and the reliability of processing performance; the speed data of each axis can be obtained from the CNC system through the protocol.

[0062] Step 2: According to the actual engineering background, select the comprehensive performance evaluation indicators of the CNC machine tool during its service life, and establish a comprehensive performance evaluation system for the CNC machine tool, including primary and secondary indicators;

[0063] There are many types of comprehensive performance indicators for CNC machine tools during their service life, including accuracy indicators, processing performance indicators, economic indicators, high efficiency indicators, and reliability indicators. However, it is impossible to fully consider all indicators when selecting performance indicators. The basic principles for constructing a general comprehensive evaluation indicator system should be followed: comprehensiveness, measurability, independence, appropriateness, and scientificity, as shown in Table 1.

[0064] Table 1 Principles for selecting performance indicators of CNC machine tools

[0065]

[0066] It is obviously unrealistic and impractical to consider all indicators at the same time in the indicator system. Based on the opinions of experts, several key performance indicators were selected. Figure 2As shown in the figure, the comprehensive performance evaluation system of CNC machine tools is divided into two levels of indicator sets. The first level is the first-level indicator, which includes 4 main indicators: reliability based on fault data, precision retention, reliability based on performance degradation, and high efficiency. The second level is refined into the second-level evaluation indicators of each first-level indicator. Among them, the reliability based on fault data includes 3 second-level indicators: mean time between repair failures, mean repair time, and mean first failure time; precision retention includes 4 second-level indicators: quantitative evaluation value of X, Y, and Z axis precision retention and quantitative evaluation value of spatial precision retention; reliability based on performance degradation includes 2 second-level indicators: comprehensive observation value of processing performance and reliability of processing performance; high efficiency includes 4 second-level indicators: maximum speed of X, Y, and Z axes and maximum speed of spindle. Different factors have different effects on comprehensive performance. The types of indicators can be divided and the subordinate relationship can be analyzed to establish a multi-level indicator set. Finally, a two-level performance indicator system that can more comprehensively reflect the performance status of CNC machine tools is established.

[0067] Step 3: Build a multi-layer evidence reasoning model and set the initial parameters of the model. Figure 3 It is the basic structure of the multi-layer evidence reasoning model. Each type of secondary indicator needs to build an evidence reasoning model. The obtained secondary indicator data is input into the first-layer evidence reasoning model to obtain the performance evaluation value of the first-layer indicator; the obtained result is input into the constructed second-layer evidence reasoning model, and finally integrated into the health status evaluation value X(y) of the whole machine.

[0068] In order to evaluate the performance status of the entire CNC machine tool that is in line with reality, it is necessary to establish an accurate performance evaluation model. The evidence reasoning method can effectively integrate the multi-source information of the CNC machine tool, and can realize performance evaluation through the calculated confidence and the defined evaluation level. In addition, expert knowledge can be integrated into the evaluation process to make the model interpretable. The evidence reasoning model effectively combines quantitative data with expert knowledge to achieve accurate evaluation of the performance of the entire CNC machine tool, so that the obtained evaluation results can accurately reflect the performance of the system. Since the indicator system for the performance evaluation of CNC machine tools is multi-level, it is necessary to construct a multi-level evidence reasoning model to evaluate the entire CNC machine tool step by step, from bottom to top, and finally evaluate the performance status of the entire machine. The advantage of the multi-level evidence reasoning model is that it can realize multi-level evaluation, and can comprehensively consider the performance of the entire CNC machine tool from multiple dimensions and aspects, avoiding the hidden danger of "representing the whole by a point", so that the obtained evaluation value can be closest to the state of the CNC machine tool at this time.

[0069] The multi-layer evidence reasoning model needs to set the initial parameters of the model according to the actual engineering background. These initial parameters are given by experts based on years of accumulated experience and may be inaccurate. Therefore, the initial parameters of the multi-layer evidence reasoning model are optimized using the Covariance Matrix Adaptation Evolutionary Strategies (CMA-ES) optimization algorithm. It performs well on the optimization problem of medium number of parameter variables, especially on high-dimensional nonlinear optimization problems. It has no gradient optimization and does not use gradient information. It can use fewer individuals to converge to the global optimal point in a relatively short time. The core idea of ​​the CMA-ES algorithm is to continuously optimize the covariance matrix, and its major axis is always in the direction of the optimal solution to guide the search strategy towards the best solution. Its direction determines the evolutionary trend of the population, and the length is used to adjust the search range of the population. In the process of model parameter optimization, it is necessary to construct an optimization objective function. The present invention uses the monotonicity (Mon) index and the correlation (Correlation, Corr) index as the objective function. The corresponding parameters when the objective function is the maximum value are the optimal parameters of the model.

[0070] (1) The construction process of the objective function is as follows:

[0071] 1) Monotonicity index: The degradation of CNC machine tool components is an irreversible process, showing a continuous increasing or decreasing trend. This characteristic is called monotonicity. The formula of monotonicity index is:

[0072] Formula 1:

[0073] Among them, dF i Indicates the difference in characteristic sequence values; Num of dF i >0 and Num of dF i <0 indicates the number of positive and negative values ​​in the feature sequence, respectively. i Represents the total number of features.

[0074] 2) Correlation index: As the service life increases, the performance of CNC machine tools gradually deteriorates. This characteristic is called correlation. The formula for the correlation index is:

[0075] Formula 2:

[0076] Among them, y t represents the performance characteristic value in the tth cycle; Indicates the average value of the performance characteristic value; l t Represents the number of the tth sampling point; is the average value of the sampling point number; Tt Indicates the total number of cycles, corresponding to the length of the performance characteristic curve.

[0077] The value range of Mon and Corr is 0-1. The larger the Mon value, the better the model monotonicity and the higher the evaluation efficiency. The larger the Corr value, the higher the correlation between the model and the actual data and the more accurate the evaluation result.

[0078] Using Equation 1 and Equation 2, the objective function for parameter optimization of the multi-layer evidence reasoning model is constructed:

[0079] Formula 3:

[0080] Where Ω represents the parameters of the evaluation model; It represents the weight corresponding to the input indicator (i.e., evidence); T represents the number of indicators.

[0081] (2) Use the CMA-ES algorithm to optimize the objective function. The specific optimization process is as follows: Figure 4 The specific steps are as follows:

[0082] a. Set initial value: Set the initial parameter vector Ω 0 , initial covariance matrix C g , initial step length η, central expectation mean 0 , initial parameters such as evolutionary generations g, population size λ, etc.

[0083] b. Generate the initial population: The process of generating the population follows the following strategy:

[0084] Formula 4:

[0085] in, represents the i-th solution in the population, Represents multivariate normal distribution, mean g represents the central expectation of the g-th generation population, η g represents the update step size of the distribution;

[0086] c. Selection operation: This operation is used to select the first ε optimal solutions in the population. Some optimal solutions are selected according to the value of the fitness function f(Ω). The higher the fitness value, the better the solution.

[0087] d. Recombination operation: The recombinant operation is achieved by the following formula:

[0088] Formula 5:

[0089] Among them, mean g+1 represents the central expectation of the g+1 generation population, γ i represents the weight of the ith solution, represents the i-th solution among the λ solutions.

[0090] e. Update the covariance matrix: The covariance matrix C represents the equal probability density ellipse of the population distribution and is updated according to the following formula:

[0091] Formula 6:

[0092] Among them, a1 and a ε represents the learning factor, q represents the evolution path, the initial value is 0, and q is updated according to the following formula:

[0093] Formula 7:

[0094] Among them, a q ≤1 is the evolution path parameter, and the step length η is updated according to the following formula:

[0095] Formula 8:

[0096] Among them, d η represents the damping coefficient, Represents Euclidean Normal Form The expectation, initially C 0 Set to the identity matrix I,q η and a η denote the conjugate evolution path and path parameter, respectively, q η Update according to the following formula: Formula 9:

[0097] f. Repeat equation 4 to equation 9 until the accuracy requirement is met and output the optimal parameters of the model.

[0098] Step 4: Use the constructed multi-layer evidence reasoning model to evaluate the secondary indicators, use the evaluation results as the input of the previous layer of evidence reasoning model, and finally evaluate the performance status of the CNC machine tool.

[0099] Figure 5 This is the framework flow chart of the comprehensive performance evaluation of CNC machine tools. First, the initial performance evaluation model is constructed, the initial weight W0 is set, the performance state of the CNC machine tool is quantified, the monotonicity index and the correlation index are calculated, and the CMA-ES optimization algorithm is used for optimization. If the optimal weight parameter W is obtained, b Then output the final evaluation result, otherwise re-execute this process.

[0100] In the multi-layer evidence reasoning model, the performance state is regarded as a dynamic process and described by a series of characteristics. Assume that the evidence (i.e., attribute) x i The nth reference value of n,i (n=1,2,...,N), and h n+1,i ≥hn,i ,h N,i and h 1,i Respectively represent the attributes x i The maximum and minimum reference values ​​are used to convert the quantitative data into confidence distribution based on the initial reference values, that is:

[0101] Formula 10: S(x i )={(h n,i ,β n,i ),n=1,2,...,N}

[0102] in, β n+1,i =1-β n,i ,h n,i ≤x i ≤h n+1,i , β m,i =0, m=1,2,...,N, m≠n,n+1.

[0103] Use the following distribution to represent quantitative properties:

[0104] Formula 11: S(x i )={(H n,i ,β n,i ),n=1,2,...,N}

[0105] Among them, β n,i ≥0, β n,i Represents attribute x i Rated as reference grade H n,i The confidence level, S(x i ) represents the confidence distribution.

[0106] When the attribute x i When m is single, assuming that n,i and m H,i Respectively represent the attributes x i Rated as reference grade H n The basic confidence value of and the unassigned basic confidence value, that is:

[0107] Formula 12: m n,i =ω i β n,i ,n=1,2,...,N

[0108] Among them, ω i Represents attribute x i The weight of , and 0≤ω≤1, Where L represents the number of weights.

[0109] Formula 13:

[0110] In the evidential reasoning method, m H,i Can be decomposed into and Two parts, and:

[0111] Formula 14:

[0112] Formula 15:

[0113] in, Represents the attribute x i The degree of incompleteness of the evaluation, when the unassigned confidence is 0, Represents the remaining basic confidence value.

[0114] Using the evidential reasoning method to aggregate the first i attributes and the (i+1)th attribute, then at the reference level H n The credit value on is:

[0115] Formula 16: {H n}:m n,I(i+1) =K I(i+1) [m n,I(i) m n,i+1 +m H,I(i) m n,i+1 +m H,i+1 m n,I(i) ]

[0116] Formula 17:

[0117] Formula 18: {H}:

[0118] Formula 19: {H}:

[0119] Formula 20:

[0120] After all attributes are synthesized according to the iterative synthesis algorithm, the confidence is distributed according to Equation 21 and Equation 22, that is:

[0121] Formula 21: {H}:

[0122] Formula 22: {H}:

[0123] in, and Respectively represent H n The confidence level and the confidence level assigned to the global level are set to the range of [β n ,(β n +βH )].

[0124] The comprehensive performance evaluation system of the CNC machine tool of the present invention is divided into two levels of indicator sets, and the first-level indicator evaluation is taken as an example. The CNC machine tool y can be expressed as the following confidence distribution:

[0125] Formula 23:

[0126] The following formula is then used to calculate the degree of influence of the input data on the performance status of the CNC machine tool:

[0127] Formula 24:

[0128] Among them, H n m Indicates the reference level of the mth indicator of the first-level indicator, It represents the confidence that the mth indicator attribute of the first-level indicator is evaluated as the reference level, n=1,…,N represents the number of reference levels, m=1,…,M represents the number of first-level indicators, and X(y) represents the utility value of CNC machine tool y after iterative fusion of attributes.

[0129] Divide all input attributes of the first-level indicators into m+1 subsystems, Represents the attribute of the i-th input data in the m-th indicator of the first-level indicator, y m It represents the output of the mth subsystem of the first-level indicator, that is, the evaluation value of the mth subsystem. The output of the subsystem is used as the input of the next-level system until the overall performance evaluation value of the CNC machine tool is obtained. The data of the first-level indicator is obtained by inputting the data of the second-level indicator into the multi-layer evidence reasoning model. The data of the second-level indicator is obtained through sensor collection and calculation and log collection, and finally the overall performance evaluation of the CNC machine tool is realized. Figure 6 The schematic diagram for the comprehensive performance evaluation of the entire CNC machine tool is used. The secondary indicator data of the CNC machine tool is input into the second-layer model to obtain the quantitative value of the performance evaluation. Based on this, it is input into the first-layer model to obtain the performance evaluation value of the entire CNC machine tool, so that enterprise personnel can accurately grasp the performance status of the CNC machine tool at this time and formulate appropriate maintenance plans in time.

[0130] Step 5: Output the performance status evaluation value of the entire CNC machine tool, connect it into a curve according to the time series, and thus obtain a performance evaluation curve.

[0131] The above are only preferred embodiments of the present invention. It should be pointed out that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention. These should also be regarded as the protection scope of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent.

Claims

1. A CNC machine tool performance evaluation method based on multi-layer evidence reasoning, characterized in that: The following steps are involved: Step 1: Obtain CNC machine tool data; Step 2: Establish a comprehensive performance evaluation system for CNC machine tools, including primary and secondary indicators; Step 3: Construct a multi-layer evidence reasoning model and set the initial parameters of the model; Step 4: Use the constructed multi-layer evidence reasoning model to evaluate the secondary indicators, use the evaluation results as the input of the previous layer of evidence reasoning model, and finally evaluate the performance status of the CNC machine tool; Step 5: Output the performance status evaluation value of the entire CNC machine tool and connect it into a curve according to the time series.

2. The CNC machine tool performance evaluation method based on multi-layer evidence reasoning according to claim 1 is characterized in that: In step 2, the first-level indicators include reliability based on fault data, accuracy retention, reliability based on performance degradation, and efficiency; each first-level indicator includes multiple second-level indicators: reliability based on fault data includes 3 second-level indicators: mean time between repairs, mean time to repair, and mean time to first failure; Precision retention includes four secondary indicators: quantitative evaluation values ​​of X, Y, and Z axis precision retention and quantitative evaluation values ​​of spatial precision retention; Reliability based on performance degradation includes two secondary indicators: comprehensive observation value of machining performance and machining performance reliability; high efficiency includes four secondary indicators: maximum speed of X, Y, and Z axes and maximum speed of spindle.

3. The CNC machine tool performance evaluation method based on multi-layer evidence reasoning according to claim 1 is characterized in that: In step 3, the CMA-ES algorithm is used to optimize the initial parameters of the multi-layer evidence reasoning model. During the parameter optimization process, the monotonicity index and the correlation index are used as the objective function of the multi-layer evidence reasoning model. The parameters corresponding to the maximum value of the objective function are the optimal parameters of the model.

4. The CNC machine tool performance evaluation method based on multi-layer evidence reasoning according to claim 3 is characterized in that: The construction process of the objective function is as follows: The formula for the monotonicity index is: Formula 1: Among them, dF i Indicates the difference in characteristic sequence values; Num of dF i >0 and Num of dF i <0 indicates the number of positive and negative values ​​in the feature sequence, respectively. i represents the total number of features; The formula for the correlation index is: Formula 2: Among them, y t represents the performance characteristic value in the tth cycle; Indicates the average value of performance characteristics; l t Represents the number of the tth sampling point; is the average value of the sampling point number; T t Indicates the total number of cycles; corresponds to the length of the performance characteristic curve; Using Equation 1 and Equation 2, the objective function for parameter optimization of the multi-layer evidence reasoning model is constructed: Formula 3: Among them, Ω represents the parameters of the evaluation model; Represents the weight corresponding to the input indicator; T represents the number of indicators.

5. The CNC machine tool performance evaluation method based on multi-layer evidence reasoning according to claim 4 is characterized in that: The specific steps of optimizing the objective function using the CMA-ES algorithm are as follows: a. Set initial value: Set the initial parameter vector Ω 0 , initial covariance matrix C g , initial step length η, central expectation mean 0 , evolutionary generations g, population size λ; b. Generate the initial population: The process of generating the population follows the following strategy: Formula 4: in, represents the i-th solution in the population, Represents multivariate normal distribution, mean g represents the central expectation of the g-th generation population, η g represents the update step size of the distribution; c. Selection operation: This operation is used to select the first ε optimal solutions in the population, and select the optimal solution according to the value of the fitness function f(Ω); d. Recombination operation: The recombinant operation is achieved by the following formula: Formula 5: Among them, mean g+1 represents the central expectation of the g+1 generation population, γ i represents the weight of the ith solution, represents the i-th solution among the λ solutions; e. Update the covariance matrix: The covariance matrix C represents the equal probability density ellipse of the population distribution and is updated according to the following formula: Formula 6: Among them, a1 and a ε represents the learning factor, q represents the evolution path, the initial value is 0, and q is updated according to the following formula: Formula 7: Among them, a q ≤1 is the evolution path parameter, and the step length η is updated according to the following formula: Formula 8: Among them, d η represents the damping coefficient, Represents Euclidean Normal Form The expectation, initially C 0 Set to the identity matrix I,q η and a η denote the conjugate evolution path and path parameter, respectively, q η Update as follows: Formula 9: f. Repeat equation 4 to equation 9 until the accuracy requirement is met and output the optimal parameters of the model.

6. The CNC machine tool performance evaluation method based on multi-layer evidence reasoning according to claim 1 is characterized in that: In the multi-layer evidence reasoning model, when evaluating the first-level indicator, the CNC machine tool y is represented by the following confidence distribution: Formula 10: The following formula is then used to calculate the degree of influence of the input data on the performance status of the CNC machine tool: Formula 11: Among them, H n m Indicates the reference level of the mth indicator of the first-level indicator, It represents the confidence that the mth indicator attribute of the first-level indicator is evaluated as the reference level, n=1,…,N represents the number of reference levels, m=1,…,M represents the number of first-level indicators, and X(y) represents the utility value of CNC machine tool y after iterative fusion of attributes.

Images