A method and device for optimizing part quality prediction and evaluation

Abstract

The present invention provides a method and device for predicting and evaluating the quality of a part, and relates to the technical field of intelligent prediction of the quality of a part processing. The present invention collects part production data and part quality data in historical data, performs a comprehensive analysis on the part quality data to generate a part qualification index that can evaluate the quality of the part, judges the qualification of the part by comparing the part qualification index of each part with a preset threshold, and generates a part quality prediction coefficient by the qualification of the part qualification index of this batch of parts. The historical production data is a data set, and the part quality prediction coefficient is used as a label. A part quality prediction model is constructed using a regression method, and the data set is used to train and optimize the part quality prediction model; the production data to be predicted is input into the optimized part quality prediction model to obtain the part quality prediction coefficient, and the evaluation and optimization of the part are completed through the part quality prediction coefficient.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent prediction of parts processing quality, and in particular to a method and device for optimizing part quality prediction and evaluation. Background Art

[0002] In modern manufacturing, the prediction of part quality is a key link to ensure product performance and reliability. However, with the increasing complexity of production processes, especially when multiple variables work together, traditional quality control methods have become difficult to adapt to the needs of efficient production. Traditional methods usually rely on empirical rules or simple statistical analysis, which makes it difficult to predict the quality of parts under different production conditions in real time. This deficiency easily leads to quality problems in the production process, which are often discovered later, which not only leads to high scrap rates, but also increases the cost of rework and repair, affecting overall production efficiency. In addition, with the promotion of intelligent manufacturing, manufacturing companies need to gradually realize automated and data-driven decision-making. Therefore, the traditional model that relies on manual inspection and experience judgment appears to be increasingly backward. Modern production requires a system that can use real-time data for dynamic prediction and quality control, so as to obtain feedback in time and make necessary adjustments.

[0003] In the prior art, the publication number CN111931307B discloses a part quality prediction and evaluation optimization method and device, a part quality prediction and evaluation optimization method and system. This method is based on the theoretical guidance of the cutting mechanism of typical materials, combined with quality inspection data, and uses the Pearson correlation coefficient formula to calculate the correlation coefficient of key parameters in the part quality parameters, and fuses the correlation coefficient with the Gaussian kernel function in the SVR model to obtain the weighted Gaussian kernel function. The SVR model with the weighted Gaussian kernel function and the particle swarm algorithm are used to train the part quality prediction model, and it is judged whether the part quality characteristics output by the model meet the quality requirements to complete the quality evaluation, and the particle swarm algorithm is used to optimize the key parameters of the parts that do not meet the quality requirements. The method and system of the prior art can realize intelligent decision-making on the quality of part processing and solve the problem that decision-making relies on manual labor and the decision-making method has serious lag in the process of part processing. However, the prior art still has shortcomings. First, there are many factors that affect the quality of parts in the production process. The prior art only extracts features of important parameters of cutting parameters, and uses the features extracted by this method as the basis for evaluation and decision-making. Obviously, other potential influencing factors are ignored, which will lead to insufficient accuracy of part quality evaluation and lack of persuasiveness. Secondly, the prior art requires part data for model training and prediction. This prediction decision-making process is still lagging, which will lead to increased scrap rate, increased rework and maintenance costs, and reduced overall production efficiency.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for optimizing part quality prediction and evaluation to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for optimizing part quality prediction and evaluation, comprising the following steps:

[0008] Step 1: Collect part production data and part quality data from historical data, wherein the part production data includes production temperature, machine tool spindle speed, processing time, cutting depth, feed rate and cutting speed, and the part quality data includes part size, surface data and part property test results;

[0009] Step 2: Extract the features of the part size to obtain the part size deviation, extract the features of the surface data to obtain the surface flatness, and generate the spatial standard coefficient according to the part size deviation and surface flatness;

[0010] Step 3: Perform a de-quantification and tempering process on the part property test results to obtain the part function coefficient, and generate the part qualification index based on the space standard coefficient and the part function coefficient;

[0011] Step 4: Set the part qualification threshold, compare the part qualification index of each part with the preset threshold, and judge the qualification of the part according to the comparison result; perform statistical analysis on the part qualification to obtain the average part qualification index and part qualification rate, and obtain the part quality prediction coefficient according to the average part qualification index and part qualification rate;

[0012] Step 5: Using historical production data as a data set and part quality prediction coefficients as labels, the LASSO regression method is used to build a part quality prediction model, and the data set is used to train and optimize the part quality prediction model;

[0013] Step 6: Input the production data to be predicted into the optimized part quality prediction model to obtain the part quality prediction coefficient, set the part quality prediction threshold, if the part quality prediction coefficient is greater than the part quality prediction threshold, then a batch of parts produced under this production condition is defined as qualified, otherwise it is defined as unqualified, and the main contributing production data of this batch of unqualified parts is determined according to the part quality prediction model, and the main contributing production data is optimized, and the optimized production data is input into the part quality prediction model and the results are compared with the threshold until they are qualified, and an optimization report is generated.

[0014] Furthermore, the specific logic for obtaining the surface data is: dividing the surface into n blocks of equal area, taking the lowest point of all the block surfaces as a reference, and collecting the highest value of each block surface as the surface height.

[0015] Furthermore, the specific logic for obtaining the part size deviation is: fitting the minimum bounding box of the part, calculating the deviation between the length, width, and height of the minimum bounding box of the part and the length, width, and height of the standard minimum bounding box of the part to generate the part size deviation; the specific formula for generating the part size deviation is:

[0016] P=(L-L0)(W-W0)(H-H0)

[0017] Among them, P is the part size deviation, L is the length of the minimum bounding box, W is the width of the minimum bounding box, H is the height of the minimum bounding box, L0 is the length of the standard minimum bounding box, W0 is the width of the standard minimum bounding box, and H0 is the height of the standard minimum bounding box.

[0018] Furthermore, the specific logic for generating the spatial standard coefficient is: extracting features from the surface data to obtain the surface flatness, and generating the spatial standard coefficient based on the part size deviation and the surface flatness; the specific formula for extracting the surface flatness is:

[0019]

[0020] Where Z is the surface flatness, is the mean surface height, Hm i is the surface height of the i-th surface, n is the number of surfaces; the specific formula for generating the spatial standard coefficient is:

[0021]

[0022] Among them, S is the space standard coefficient, P is the part size deviation, and Z is the surface flatness.

[0023] Furthermore, the specific logic for generating the part qualification index is as follows: the part property test results are subjected to the toughening treatment to obtain the part function coefficient, and the part qualification index is generated according to the space standard coefficient and the part function coefficient; the specific formula for generating the part function coefficient is as follows:

[0024]

[0025] Among them, Ab is the part function coefficient, Ha is the part hardness, R is the part toughness, and F is the tensile strength;

[0026] The specific formula for generating the part qualification index is:

[0027] QS=S*tan-1 Ab

[0028] Among them, QS is the part qualification index, Ab is the part function coefficient, and S is the space standard coefficient.

[0029] Furthermore, the specific logic for obtaining the part quality prediction coefficient is as follows: calculate the average part qualification index of all parts, set a part qualification threshold, compare the part qualification index of each part with the preset threshold, mark the part whose part qualification index is greater than the part qualification threshold as a qualified part, otherwise it is an unqualified part, count the number of qualified parts and the number of unqualified parts and calculate the part qualification rate based on the number of qualified parts and the number of unqualified parts, and obtain the part quality prediction coefficient based on the average part qualification index and the part qualification rate; the specific formula for generating the average part qualification index is as follows:

[0030]

[0031] in, is the average parts qualification index, QS i is the part qualification index of the ith part, N is the number of all parts;

[0032] The specific formula for generating the part qualification rate is:

[0033]

[0034] Among them, C is the qualified rate of parts, and M is the number of qualified parts;

[0035] The specific formula for generating the part quality prediction coefficient is:

[0036]

[0037] Among them, QP is the part quality prediction coefficient, C is the part qualification rate, is the average parts qualification index.

[0038] Furthermore, the LASSO regression method is used to construct the loss function, which is expressed as:

[0039]

[0040] Where K is the number of parts, y i is the part quality prediction coefficient of the i-th group of parts, O is the number of production data types, x ij represents the jth production data of the i-th group of parts, β j is the characteristic coefficient, is the part quality prediction coefficient, and λ is the regularization parameter.

[0041] The present invention further provides a device for predicting and evaluating the quality of a part, and the method for predicting and evaluating the quality of a part described in the device specifically comprises:

[0042] A data acquisition module is used to collect part production data and part quality data in historical data, wherein the part production data includes production temperature, machine tool spindle speed, processing time, cutting depth, feed rate and cutting speed, and the part quality data includes part size, surface data and part property test results;

[0043] A data processing module is used to extract the features of the part size to obtain the part size deviation, extract the features of the surface data to obtain the surface flatness, and generate a spatial standard coefficient according to the part size deviation and the surface flatness;

[0044] The data analysis module is used to perform a quantitative tempering process on the part property test results to obtain the part function coefficient, and generate a part qualification index based on the space standard coefficient and the part function coefficient;

[0045] The qualified judgment module is used to set the qualified threshold of the parts, compare the qualified index of each part with the preset threshold, and judge the qualified situation of the parts according to the comparison result; perform statistical analysis on the qualified situation of the parts to obtain the average qualified index and the qualified rate of the parts, and obtain the part quality prediction coefficient according to the average qualified index and the qualified rate of the parts;

[0046] Modeling and optimization module, which uses historical production data as a data set, part quality prediction coefficients as labels, uses the LASSO regression method to build a part quality prediction model, and uses the data set to train and optimize the part quality prediction model;

[0047] The prediction optimization module is used to input the production data to be predicted into the optimized part quality prediction model to obtain the part quality prediction coefficient, set the part quality prediction threshold, and if the part quality prediction coefficient is greater than the part quality prediction threshold, a batch of parts produced under the production conditions is defined as qualified, otherwise it is defined as unqualified. The main contributing production data of this batch of unqualified parts is determined according to the part quality prediction model, the main contributing production data is optimized, the optimized production data is input into the part quality prediction model and the results are compared with the threshold until they are qualified, and an optimization report is generated.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] The present invention directly quantifies the quality of historical finished parts as a basis for evaluating the quality of parts, and finds the relationship between production data and the quality of finished parts through a machine learning model to predict the quality of the parts to be predicted. When quantifying the quality of parts, multiple types of part quality-related data are taken into account, which can improve the credibility of part quality assessment decisions.

[0050] The present invention only needs to collect part production data when predicting part quality. The part quality can be directly predicted through the production data, which can avoid the decision lag caused by part parameter collection and the high scrap rate and low efficiency caused by such lag. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the overall method flow of the present invention.

[0052] Figure 2 It is a schematic diagram of the overall structure of the device of the present invention. DETAILED DESCRIPTION

[0053] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0054] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0055] Example:

[0056] See also Figure 1 , the present invention provides a technical solution:

[0057] A method for optimizing part quality prediction and evaluation, comprising the following steps:

[0058] Step 1: Collect part production data and part quality data from historical data, wherein the part production data includes production temperature, machine tool spindle speed, processing time, cutting depth, feed rate and cutting speed, and the part quality data includes part size, surface data and part property test results;

[0059] The specific logic for obtaining the surface data in this embodiment is: dividing the surface into n blocks of equal area, taking the lowest point of all the block surfaces as a reference, and collecting the highest value of each block surface as the surface height.

[0060] The properties tested by the part test include: part hardness, part toughness and tensile strength. The part hardness, part toughness and tensile strength of the part are obtained by sending the part to a laboratory for non-destructive testing.

[0061] Step 2: Extract the features of the part size to obtain the part size deviation, extract the features of the surface data to obtain the surface flatness, and generate the spatial standard coefficient according to the part size deviation and surface flatness;

[0062] The specific logic for obtaining the part size deviation is: fitting the minimum bounding box of the part, calculating the deviation between the length, width, and height of the minimum bounding box of the part and the length, width, and height of the standard minimum bounding box of the part to generate the part size deviation; the specific formula for generating the part size deviation is:

[0063] P=(L-L0)(W-W0)(H-H0)

[0064] Among them, P is the part size deviation, L is the length of the minimum bounding box, W is the width of the minimum bounding box, H is the height of the minimum bounding box, L0 is the length of the standard minimum bounding box, W0 is the width of the standard minimum bounding box, and H0 is the height of the standard minimum bounding box. The part size deviation P reflects the deviation between the part size and the ideal part size. The larger its value, the deviation between the part size and the ideal part size, and the worse the part quality. The generation of this feature can provide an important basis for the evaluation of the comprehensive quality of parts.

[0065] The specific logic for generating the spatial standard coefficient is: extracting features from surface data to obtain surface flatness, and generating the spatial standard coefficient based on part size deviation and surface flatness; the specific formula for extracting surface flatness is:

[0066]

[0067] Where Z is the surface flatness, is the mean surface height, Hm i is the surface height of the i-th surface, and n is the number of surfaces; the surface flatness Z reflects the flatness of the part surface. The smaller its value is, the flatter the part surface is and the better the part quality is. The generation of this feature can provide an important basis for the evaluation of part quality; the average surface height Reflects the general height of the part surface, the height of each small surface and the average surface height The larger the gap, the higher the probability that the part surface is uneven, and the average surface height The calculation of plays an important role in evaluating the extraction of surface flatness Z.

[0068] The specific formula for generating the spatial standard coefficient is:

[0069]

[0070] Among them, S is the spatial standard coefficient, P is the part size deviation, and Z is the surface flatness. The spatial standard coefficient S reflects the quality of the part from a spatial perspective. The larger its value, the better the quality of the part from a spatial perspective. The generation of this coefficient can provide an important basis for the evaluation of the comprehensive quality of the part.

[0071] Step 3: Perform a de-quantification and tempering process on the part property test results to obtain the part function coefficient, and generate the part qualification index based on the space standard coefficient and the part function coefficient;

[0072] The specific logic for generating the part qualification index is as follows: the part property test results are subjected to the tempering treatment to obtain the part function coefficient, and the part qualification index is generated according to the space standard coefficient and the part function coefficient; the specific formula for generating the part function coefficient is as follows:

[0073]

[0074] Among them, Ab is the part function coefficient, Ha is the part hardness, R is the part toughness, and F is the tensile strength; the part function coefficient Ab reflects the comprehensive performance of the part. The larger its value is, the better the part performance and the better the part quality. The generation of this coefficient can provide an important basis for the evaluation of the comprehensive quality of the part.

[0075] The specific formula for generating the part qualification index is:

[0076] QS=S*tan -1 Ab

[0077] Among them, QS is the part qualification index, Ab is the part function coefficient, and S is the space standard coefficient. The part qualification index QS reflects the comprehensive quality level of a single part. The larger the value, the higher the comprehensive quality of the single part. This coefficient can be directly used as the basis for part quality assessment.

[0078] Step 4: Set the part qualification threshold, compare the part qualification index of each part with the preset threshold, and judge the qualification of the part according to the comparison result; perform statistical analysis on the part qualification to obtain the average part qualification index and part qualification rate, and obtain the part quality prediction coefficient according to the average part qualification index and part qualification rate;

[0079] The specific logic for obtaining the part quality prediction coefficient is as follows: calculate the average part qualification index of all parts, set the part qualification threshold, compare the part qualification index of each part with the preset threshold, mark the parts with a part qualification index greater than the part qualification threshold as qualified parts, otherwise as unqualified parts, count the number of qualified parts and the number of unqualified parts, and calculate the part qualification rate based on the number of qualified parts and the number of unqualified parts, and obtain the part quality prediction coefficient based on the average part qualification index and the part qualification rate; the specific formula for generating the average part qualification index is as follows:

[0080]

[0081] in, is the average parts qualification index, QS i is the part qualification index of the ith part, N is the number of all parts; the average part qualification index It reflects the comprehensive quality of a batch of parts. The larger the value is, the higher the comprehensive quality of the batch of parts is.

[0082] The specific formula for generating the part qualification rate is:

[0083]

[0084] Among them, C is the parts qualification rate, and M is the number of qualified parts; the parts qualification rate C reflects the quality of parts that meet the standards under certain production conditions. The larger the value, the more parts that meet the quality standards under certain production conditions.

[0085] The specific formula for generating the part quality prediction coefficient is:

[0086]

[0087] Among them, QP is the part quality prediction coefficient, C is the part qualification rate, The part quality prediction coefficient QP reflects the quality level of a batch of parts under certain conditions. The larger the value, the higher the quality of a batch of parts under certain conditions. The generation of this coefficient can provide an important basis for the quality assessment of a batch of parts.

[0088] Step 5: Using historical production data as a data set and part quality prediction coefficients as labels, the LASSO regression method is used to build a part quality prediction model, and the data set is used to train and optimize the part quality prediction model;

[0089] A feedforward neural network is used, with the part quality prediction coefficient as the label, and the data set is used to train and optimize the part quality prediction model; specifically, it includes: input layer, hidden layer, output layer and activation function, and the loss function is constructed using the LASSO regression method; the input data is calculated once through the network to obtain the output result, and the loss function is calculated based on the predicted value and the true value. The gradient of the loss function for each weight and bias is calculated by the chain rule, and the weights and biases of the network are updated using the gradient descent algorithm to minimize the loss function.

[0090] The LASSO regression method is used to construct the loss function, which is expressed as:

[0091]

[0092] Where K is the number of parts, y i is the part quality prediction coefficient of the i-th group of parts, O is the number of production data types, x ij represents the jth production data of the i-th group of parts, β j is the characteristic coefficient, is the predicted part quality prediction coefficient, λ is the regularization parameter. In the loss function, the LASSO regression method expresses the predicted part quality prediction coefficient with a linear equation of various production data, and the characteristic coefficient β j is the coefficient of each production feature in the linear equation; and a regularization term is added for optimization; this regularization term is represented by the regularization parameter λ and the sum of the absolute values ​​of all feature coefficients. This can not only be used to improve the prediction accuracy of the model, but also automatically perform feature selection, by using the feature coefficient β in front of each production data j The size of determines the importance of the corresponding production data; the characteristic coefficient β j The larger it is, the greater the contribution of the production data to the part quality prediction coefficient.

[0093] Step 6: Input the production data to be predicted into the optimized part quality prediction model to obtain the part quality prediction coefficient, set the part quality prediction threshold, if the part quality prediction coefficient is greater than the part quality prediction threshold, then a batch of parts produced under this production condition is defined as qualified, otherwise it is defined as unqualified, and the main contributing production data of this batch of unqualified parts is determined according to the part quality prediction model, and the main contributing production data is optimized, and the optimized production data is input into the part quality prediction model and the results are compared with the threshold until they are qualified, and an optimization report is generated.

[0094] See also Figure 2 The present invention further provides a device for predicting and evaluating the quality of a part, and the method for predicting and evaluating the quality of a part described in the device specifically comprises:

[0095] A data acquisition module is used to collect part production data and part quality data in historical data, wherein the part production data includes production temperature, machine tool spindle speed, processing time, cutting depth, feed rate and cutting speed, and the part quality data includes part size, surface data and part property test results;

[0096] A data processing module is used to extract the features of the part size to obtain the part size deviation, extract the features of the surface data to obtain the surface flatness, and generate a spatial standard coefficient according to the part size deviation and the surface flatness;

[0097] The data analysis module is used to perform a quantitative tempering process on the part property test results to obtain the part function coefficient, and generate a part qualification index based on the space standard coefficient and the part function coefficient;

[0098] The qualified judgment module is used to set the qualified threshold of the parts, compare the qualified index of each part with the preset threshold, and judge the qualified situation of the parts according to the comparison result; perform statistical analysis on the qualified situation of the parts to obtain the average qualified index and the qualified rate of the parts, and obtain the part quality prediction coefficient according to the average qualified index and the qualified rate of the parts;

[0099] Modeling and optimization module, which uses historical production data as a data set, part quality prediction coefficients as labels, uses the LASSO regression method to build a part quality prediction model, and uses the data set to train and optimize the part quality prediction model;

[0100] The prediction optimization module is used to input the production data to be predicted into the optimized part quality prediction model to obtain the part quality prediction coefficient, set the part quality prediction threshold, and if the part quality prediction coefficient is greater than the part quality prediction threshold, a batch of parts produced under the production conditions is defined as qualified, otherwise it is defined as unqualified. The main contributing production data of this batch of unqualified parts is determined according to the part quality prediction model, the main contributing production data is optimized, the optimized production data is input into the part quality prediction model and the results are compared with the threshold until they are qualified, and an optimization report is generated.

[0101] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0102] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0103] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0104] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for optimizing part quality prediction and evaluation, characterized in that: The specific steps include: Step 1: Collect part production data and part quality data from historical data, wherein the part production data includes production temperature, machine tool spindle speed, processing time, cutting depth, feed rate and cutting speed, and the part quality data includes part size, surface data and part property test results; Step 2: Extract the features of the part size to obtain the part size deviation, extract the features of the surface data to obtain the surface flatness, and generate the spatial standard coefficient according to the part size deviation and surface flatness; Step 3: Perform a de-quantification and tempering process on the part property test results to obtain the part function coefficient, and generate the part qualification index based on the space standard coefficient and the part function coefficient; Step 4: Set the part qualification threshold, compare the part qualification index of each part with the preset threshold, and judge the qualification of the part according to the comparison result; perform statistical analysis on the part qualification to obtain the average part qualification index and part qualification rate, and obtain the part quality prediction coefficient according to the average part qualification index and part qualification rate; Step 5: Using historical production data as a data set and part quality prediction coefficients as labels, the LASSO regression method is used to build a part quality prediction model, and the data set is used to train and optimize the part quality prediction model; Step 6: Input the production data to be predicted into the optimized part quality prediction model to obtain the part quality prediction coefficient, set the part quality prediction threshold, if the part quality prediction coefficient is greater than the part quality prediction threshold, then a batch of parts produced under this production condition is defined as qualified, otherwise it is defined as unqualified, and the main contributing production data of this batch of unqualified parts is determined according to the part quality prediction model, and the main contributing production data is optimized, and the optimized production data is input into the part quality prediction model and the results are compared with the threshold until they are qualified, and an optimization report is generated; The specific logic for the spatial standard coefficient is: extract the surface data to obtain the surface flatness, and generate the spatial standard coefficient according to the part size deviation and surface flatness; the specific formula for extracting the surface flatness is: Where Z is the surface flatness, is the mean surface height, Hm i is the surface height of the i-th surface, and n is the number of surfaces; The specific formula for generating the spatial standard coefficient is: Among them, S is the space standard coefficient, P is the part size deviation, Z is the surface flatness, The specific logic for generating the part qualification index is as follows: the part property test results are subjected to the tempering treatment to obtain the part function coefficient, and the part qualification index is generated according to the space standard coefficient and the part function coefficient; the specific formula for generating the part function coefficient is as follows: Among them, Ab is the part function coefficient, Ha is the part hardness, R is the part toughness, and F is the tensile strength; The specific formula for generating the part qualification index is: <h2 style=";text-align:left;direction:ltr">QS = S*tan<h2 style=";text-align:left;direction:ltr"> -1 <h2 style=";text-align:left;direction:ltr"> Ab Among them, QS is the part qualification index, Ab is the part function coefficient, and S is the space standard coefficient; The specific logic for obtaining the part quality prediction coefficient is as follows: calculate the average part qualification index of all parts, set the part qualification threshold, compare the part qualification index of each part with the preset threshold, mark the parts with a part qualification index greater than the part qualification threshold as qualified parts, otherwise as unqualified parts, count the number of qualified parts and the number of unqualified parts, and calculate the part qualification rate based on the number of qualified parts and the number of unqualified parts, and obtain the part quality prediction coefficient based on the average part qualification index and the part qualification rate; the specific formula for generating the average part qualification index is as follows: in, is the average parts qualification index, QS i is the part qualification index of the ith part, N is the number of all parts; The specific formula for generating the part qualification rate is: Among them, C is the qualified rate of parts, and M is the number of qualified parts; The specific formula for generating the part quality prediction coefficient is: Among them, QP is the part quality prediction coefficient, C is the part qualification rate, is the average parts qualification index.

2. A part quality prediction and evaluation optimization method according to claim 1, characterized in that: The specific logic for obtaining the surface data is as follows: the surface is divided into n blocks of equal area, the lowest point of all the block surfaces is taken as a reference, and the highest value of each block surface is collected as the surface height.

3. A part quality prediction and evaluation optimization method according to claim 1, characterized in that: The specific logic for obtaining the part size deviation is: fitting the minimum bounding box of the part, calculating the deviation between the length, width, and height of the minimum bounding box of the part and the length, width, and height of the standard minimum bounding box of the part to generate the part size deviation; the specific formula for generating the part size deviation is: P=(L-L0)(W-W0)(H-H0) Among them, P is the part size deviation, L is the length of the minimum bounding box, W is the width of the minimum bounding box, H is the height of the minimum bounding box, L0 is the length of the standard minimum bounding box, W0 is the width of the standard minimum bounding box, and H0 is the height of the standard minimum bounding box.

4. A part quality prediction and evaluation optimization method according to claim 1, characterized in that: The LASSO regression method is used to construct the loss function, which is expressed as: Where K is the number of parts, y i is the part quality prediction coefficient of the i-th group of parts, O is the number of production data types, x ij represents the jth production data of the i-th group of parts, β j is the characteristic coefficient, is the part quality prediction coefficient, and λ is the regularization parameter.

5. A device for predicting and evaluating the quality of a part, characterized in that: The device is used to implement the part quality prediction and evaluation optimization method according to any one of claims 1 to 4, specifically comprising: A data acquisition module is used to collect part production data and part quality data in historical data, wherein the part production data includes production temperature, machine tool spindle speed, processing time, cutting depth, feed rate and cutting speed, and the part quality data includes part size, surface data and part property test results; A data processing module is used to extract the features of the part size to obtain the part size deviation, extract the features of the surface data to obtain the surface flatness, and generate a spatial standard coefficient according to the part size deviation and the surface flatness; The data analysis module is used to perform a quantitative tempering process on the part property test results to obtain the part function coefficient, and generate a part qualification index based on the space standard coefficient and the part function coefficient; The qualified judgment module is used to set the qualified threshold of the parts, compare the qualified index of each part with the preset threshold, and judge the qualified situation of the parts according to the comparison result; perform statistical analysis on the qualified situation of the parts to obtain the average qualified index and the qualified rate of the parts, and obtain the part quality prediction coefficient according to the average qualified index and the qualified rate of the parts; Modeling and optimization module, which uses historical production data as a data set, part quality prediction coefficients as labels, uses the LASSO regression method to build a part quality prediction model, and uses the data set to train and optimize the part quality prediction model; The prediction optimization module is used to input the production data to be predicted into the optimized part quality prediction model to obtain the part quality prediction coefficient, set the part quality prediction threshold, and if the part quality prediction coefficient is greater than the part quality prediction threshold, a batch of parts produced under the production conditions is defined as qualified, otherwise it is defined as unqualified. The main contributing production data of this batch of unqualified parts is determined according to the part quality prediction model, the main contributing production data is optimized, the optimized production data is input into the part quality prediction model and the results are compared with the threshold until they are qualified, and an optimization report is generated.

Images