CN117495070A - Technological parameter recommendation method and system for industrial assembly line - Google Patents
Technological parameter recommendation method and system for industrial assembly line Download PDFInfo
- Publication number
- CN117495070A CN117495070A CN202311815929.5A CN202311815929A CN117495070A CN 117495070 A CN117495070 A CN 117495070A CN 202311815929 A CN202311815929 A CN 202311815929A CN 117495070 A CN117495070 A CN 117495070A
- Authority
- CN
- China
- Prior art keywords
- data
- parameter recommendation
- production
- time sequence
- quality inspection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 82
- 238000004519 manufacturing process Methods 0.000 claims abstract description 112
- 238000007689 inspection Methods 0.000 claims abstract description 54
- 230000008569 process Effects 0.000 claims abstract description 42
- 238000010801 machine learning Methods 0.000 claims abstract description 36
- 238000013136 deep learning model Methods 0.000 claims abstract description 20
- 230000004927 fusion Effects 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 15
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 12
- 230000007246 mechanism Effects 0.000 claims description 10
- 238000010923 batch production Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 230000002068 genetic effect Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 239000004615 ingredient Substances 0.000 description 6
- 241001441571 Hiodontidae Species 0.000 description 5
- 239000002994 raw material Substances 0.000 description 5
- 238000005457 optimization Methods 0.000 description 4
- 238000009825 accumulation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0633—Workflow analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Strategic Management (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Operations Research (AREA)
- Educational Administration (AREA)
- Probability & Statistics with Applications (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Manufacturing & Machinery (AREA)
- Development Economics (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of industrial artificial intelligence, in particular to a process parameter recommendation method and a process parameter recommendation system for an industrial assembly line, wherein the method comprises the following steps: reading production data of a current batch of products of an industrial assembly line; analyzing the production data into continuous variable data, category variable data and time-ordered data, and carrying out data processing according to different data types, wherein the method comprises the following steps: taking continuous variable data and category variable data as input, inputting a plurality of trained machine learning models, inputting a fusion-trained machine learning model, and carrying out multi-model fusion prediction to obtain a quality inspection prediction result predicted by the machine learning model; and taking the time sequence data as input, inputting the trained deep learning model to obtain a time sequence model prediction result, and integrating the quality inspection prediction result and the time sequence prediction result predicted by the machine learning model to obtain a final quality inspection prediction result of the current batch of products combined with the time sequence data. The invention can predict the quality of the current production product.
Description
Technical Field
The invention relates to the technical field of industrial artificial intelligence, in particular to a process parameter recommendation method and system of an industrial assembly line.
Background
In the industrial flow line production process, the flow line production mode of processing and producing the same type of products is still mainstream by manually setting the parameters of a processing machine and the parameters of the process, and particularly for batch production lines, the parameters of the process need to be adjusted manually and continuously based on manual experience. The assembly line technological parameter setting based on artificial experience can reach a better qualified standard under the standard condition, but in the production process, a large number of random disturbance conditions exist, such as raw material quality fluctuation, more influences are caused on production and processing, the quality inspection result of a product cannot be obtained in time, and the artificial experience cannot be estimated accurately along with the random disturbance and the parameters are adjusted in time. Moreover, the quality inspection process in the mode is long, and the quality fluctuation of the produced product caused by the raw material difference and different parameters cannot be accurately perceived by manual quality inspection, so that the quality of the produced product is frequently fluctuated.
In the prior art, the manual implementation of the process parameter adjustment only focuses on the parameter recommendation of the next batch production or the unified specification parameter adjustment after the current specification production is completed. The method does not relate to predicting quality inspection results and recommending parameters in time for various production data, guiding the next batch of production, recommending specification process parameters through accumulated data and the like.
Disclosure of Invention
First, the technical problem to be solved
In view of the above-mentioned shortcomings and disadvantages of the prior art, the present invention provides a process parameter recommendation method and system for an industrial assembly line, which solves the technical problems that the existing process parameter adjustment by manual experience cannot be estimated more accurately along with random disturbance and the parameter can be adjusted in time.
(II) technical scheme
In order to achieve the above purpose, the main technical scheme adopted by the invention comprises the following steps:
in a first aspect, an embodiment of the present invention provides a process parameter recommendation method for an industrial assembly line, including the following steps:
reading production data of a current batch of products of an industrial assembly line;
analyzing the production data into continuous variable data, category variable data and time-ordered data, and carrying out data processing according to different data types, wherein the method comprises the following steps:
taking continuous variable data and category variable data as input, inputting a plurality of trained machine learning models, inputting a fusion-trained machine learning model, and carrying out multi-model fusion prediction to obtain a quality inspection prediction result predicted by the machine learning model;
and taking the time sequence data as input, inputting the trained deep learning model to obtain a time sequence model prediction result, and integrating the quality inspection prediction result and the time sequence prediction result predicted by the machine learning model to obtain a final quality inspection prediction result of the current batch of products combined with the time sequence data.
The technical parameter recommendation method of the industrial assembly line provided by the embodiment of the invention can be combined with processing of various different types of production data, and various different learning models are used for learning the data content, so that the prediction of the current production quality inspection result through the current assembly line production batch data is realized.
Optionally, a plurality of trained machine learning models are trained by:
taking continuous variable data, category variable data and quality inspection results in historical production data of an industrial assembly line as data sets, and fully learning by adopting a plurality of different first-type machine learning models to obtain a plurality of trained first-type machine learning models; continuously integrating and learning the quality inspection results learned by a plurality of first-class machine learning models in a multi-model fusion mode to obtain a machine learning model subjected to fusion training;
the deep learning model is obtained through training the following steps:
and sending the time sequence data and the quality inspection result in the historical production data of the industrial assembly line into a time sequence deep learning model based on a transducer as a data set, and finally obtaining a trained deep learning model by learning the correlation between the trend among a plurality of time sequence characteristic data and the quality inspection result.
Optionally, the production data includes product specifications, machine parameters, and process parameters.
Optionally, the method further comprises:
in the primary parameter recommendation stage, production data and quality inspection data of the same-specification products in the historical data are used as input data, and machine set parameter recommendation values in the next batch production under the same product specification are obtained through a primary parameter recommendation model;
in the second-level parameter recommendation stage, the production data and the real quality inspection data of the current batch of products are used as input, and the production process parameter recommendation value under the same product specification is obtained through a second-level parameter recommendation model.
Fusing the recommended values of the machine set parameters and the recommended values of the production process parameters which meet the constraint conditions to form the recommended values of the production process parameters under the same product specification. Production process parameter recommendations, which are commonly used in recent times under this specification, make necessary references for process adjustments by the process sector.
Optionally, the primary parameter recommendation model and the secondary parameter recommendation model are obtained through training of the following steps:
and taking the historical machine setting parameters, production process parameters, production data and quality inspection results under the same product specification as data sets, inputting parameter recommendation models based on genetic algorithm, and respectively training to obtain a primary parameter recommendation model taking the machine setting parameters as output and a secondary parameter recommendation model taking the production process parameters as output. Optionally, preprocessing the data in the data set prior to training each model includes:
for continuous variable data, performing data preprocessing by adopting feature extraction, feature screening, feature generation, repeated data removal, outlier removal, feature generation and/or normalization;
aiming at category type variable data, category coding, independent heat coding, feature coding and/or feature embedding are adopted to conduct data preprocessing on the data;
and aiming at time sequence data, performing data preprocessing by adopting time sequence segmentation, length complementation, data complementation, extraction of time sequence trend characteristics and/or normalization.
Optionally, extracting the time series trend feature includes adopting a period term trend term decomposition method, which assumes that the time series data is composed of a trend term and a period term, namely:
；
wherein y is the extracted time sequence trend characteristic; trend item T is obtained through sliding window and mean value obtaining mode t The original value and the trend term are differenced to obtain a period term S t And finally realizing trend decomposition.
Optionally, the first type of machine learning model includes XGBoost, catBoost, lightGBM and Gradient Boosting Regressor models.
Optionally, a time sequence deep learning model based on a transducer is adopted, and an attention mechanism-based deep learning model is adopted, wherein an attention mechanism formula is expressed as follows:
；
wherein the method comprises the steps ofFor a sparse matrix obtained by a sparsity measure, Q, K, V are the input data X multiplied by the learnable matrix W, respectively Q ,W K ,W V The input matrix d k Vector dimensions for Q and K;
the evaluation and calculation mode for sparsity is to calculate the relative entropy of the probability distribution of the corresponding attention moment array of Q by using KL (karma) sparsity, wherein the sparsity evaluation formula of the ith value is as follows:
；
wherein q i And k j For elements in Q and K in the attention mechanism, i and j are the sequence numbers of the elements, d k Vector dimension of K, L k Is the length of K, e is a constant, and the superscript T denotes the transpose of the matrix.
In a second aspect, an embodiment of the present invention provides a process parameter recommendation system for an industrial pipeline, including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing steps of any of the methods described above when executing the computer program.
(III) beneficial effects
The beneficial effects of the invention are as follows: the process parameter recommendation method and system of the industrial assembly line can be combined to process various different types of production data, various different learning models are used for learning data content, the current production quality inspection result is predicted through the current assembly line production batch data, the next batch production process parameter recommendation with the same specification is performed by combining the quality inspection prediction result and the production data, the specification parameter recommendation is performed by combining the corresponding production data and the real value of the quality inspection data through historical data accumulation, the specification parameter recommendation is comprehensively performed by combining the previously recommended parameters, and the process department is helped to adjust the production process parameters with different specifications.
Drawings
FIG. 1 is a flow chart of a process parameter recommendation method for an industrial pipeline according to an embodiment of the present invention;
FIG. 2 is a flow chart of recommending process parameter values for a production lot by production data according to an embodiment of the present invention.
Detailed Description
The invention will be better explained by the following detailed description of the embodiments with reference to the drawings.
The technical parameter recommendation method and the technical parameter recommendation system for the industrial assembly line can be used for predicting quality inspection of the batch product assembly line after one batch of production is completed according to various types of production data. And the parameter optimization recommendation can be performed, the next product batch parameter with the same specification is optimized, and the optimal parameter is guided to be set so as to achieve the production batch quality optimization. Meanwhile, through accumulation and summarization of a large amount of historical data, optimization recommendation of parameters can be performed according to different production specifications through an algorithm, so that a process department can adjust the pipeline parameters.
In order that the above-described aspects may be better understood, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
Example 1
Referring to fig. 1, an embodiment of the present invention provides a process parameter recommendation method for an industrial assembly line, including the following steps:
reading production data of a current batch of products of an industrial assembly line; in practice, the production data includes product specifications, machine parameters, and process parameters.
Analyzing the production data into continuous variable data, category variable data and time-ordered data, and carrying out data processing according to different data types, wherein the method comprises the following steps:
taking continuous variable data and category variable data as input, inputting a plurality of trained machine learning models, inputting a fusion-trained machine learning model, and carrying out multi-model fusion prediction to obtain a quality inspection prediction result predicted by the machine learning model;
and taking the time sequence data as input, inputting the trained deep learning model to obtain a time sequence model prediction result, and integrating the quality inspection prediction result and the time sequence prediction result predicted by the machine learning model to obtain a final quality inspection prediction result of the current batch of products combined with the time sequence data.
Preprocessing data in a data set prior to training of each model, comprising:
for continuous variable data, performing data preprocessing by adopting feature extraction, feature screening, feature generation, repeated data removal, outlier removal, feature generation and/or normalization;
aiming at category type variable data, category coding, independent heat coding, feature coding and/or feature embedding are adopted to conduct data preprocessing on the data;
and aiming at time sequence data, performing data preprocessing by adopting time sequence segmentation, length complementation, data complementation, extraction of time sequence trend characteristics and/or normalization. The extracting the time sequence trend feature comprises adopting a period term trend term decomposition method, wherein the time sequence data is assumed to be composed of a trend term and a period term, namely:
；
wherein y is the extracted time sequence trend characteristic; trend item T is obtained through sliding window and mean value obtaining mode t The original value and the trend term are differenced to obtain a period term S t And finally realizing trend decomposition.
In this embodiment, a plurality of trained machine learning models are obtained by training the following steps:
taking continuous variable data, category variable data and quality inspection results in historical production data of an industrial assembly line as data sets, and fully learning by adopting a plurality of different first-type machine learning models to obtain a plurality of trained first-type machine learning models; and continuously performing integrated learning on the quality inspection results after learning the plurality of first-class machine learning models in a multi-model fusion mode to obtain a machine learning model after fusion training. In this implementation, the first type of machine learning model includes XGBoost, catBoost, lightGBM and other models such as Gradient Boosting Regressor integrated decision tree models. The machine learning model mainly characterizes the relationship between production data and quality inspection results.
The deep learning model is obtained through training the following steps:
and sending the time sequence data and the quality inspection result in the historical production data of the industrial assembly line into a time sequence deep learning model based on a transducer as a data set, and finally obtaining a trained deep learning model by learning the correlation between the trend among a plurality of time sequence characteristic data and the quality inspection result. The time sequence deep learning model based on the transducer adopts a deep learning model based on an attention mechanism, and the attention mechanism formula is expressed as follows:
；
considering sparsity of production time sequence data, an original model is improved by adopting a sparse attention mechanism, and an improved attention mechanism formula is expressed as follows:
；
wherein the method comprises the steps ofFor a sparse matrix obtained by a sparsity measure, Q, K, V are the input data X multiplied by the learnable matrix W, respectively Q ,W K ,W V The input matrix d k Vector dimensions for Q and K;
the evaluation and calculation mode for sparsity is to calculate the relative entropy of the probability distribution of the corresponding attention moment array of Q by using KL (karma) sparsity, wherein the sparsity evaluation formula of the ith value is as follows:
；
wherein q i And k j I and j are elements in Q and K in the attention mechanismSequence number of element d k Vector dimension of K, L k Is the length of K, e is a constant, and the superscript T denotes the transpose of the matrix.
The technical parameter recommendation method of the industrial assembly line provided by the embodiment of the invention can be combined with processing of various different types of production data, and various different learning models are used for learning the data content, so that the prediction of the current production quality inspection result through the current assembly line production batch data is realized. After the production of the current batch of products is finished, the real quality inspection result data can be used as new data to be updated into the model for continuous learning and optimization of the model.
Referring to fig. 2, in addition to predicting the production quality inspection result, the present embodiment also provides an example of parameter recommendation, which may be performed by first-level parameter recommendation, second-level parameter recommendation, and fusion parameter recommendation.
Through the steps, the final quality inspection prediction result of the combination time sequence data of the current batch of products is obtained through prediction. The following recommendations may also be made:
in the primary parameter recommendation stage, production data and quality inspection data of products with the same specification in the historical data are used as input data, and machine set parameter recommendation values in the next batch production under the same product specification are obtained through a primary parameter recommendation model. The machine setting parameter recommended value (primary parameter recommended value) can be used for guiding the production machine parameter setting when the next batch with the same specification is produced.
In the second-level parameter recommendation stage, the production data and the real quality inspection data of the current batch of products are used as input, and the production process parameter recommendation value under the same product specification is obtained through a second-level parameter recommendation model.
Fusing the recommended values of the machine set parameters and the recommended values of the production process parameters which meet the constraint conditions to form the recommended values of the production process parameters under the same product specification. The complete parameter recommendation under the specification is obtained and is used for the process department to adjust the parameter setting of the production line.
In this embodiment, the primary parameter recommendation model and the secondary parameter recommendation model are obtained through training:
and taking the historical machine setting parameters, production process parameters, production data and quality inspection results under the same product specification as data sets, inputting parameter recommendation models based on genetic algorithm, and respectively training to obtain a primary parameter recommendation model taking the machine setting parameters as output and a secondary parameter recommendation model taking the production process parameters as output.
In a second aspect, an embodiment of the present invention provides a process parameter recommendation system for an industrial pipeline, including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing steps of any of the methods described above when executing the computer program.
Example 2
The method of the embodiment 1 is applied to a rubber banburying production line and comprises the following steps:
obtaining production data of a rubber banburying production line, comprising: non-time sequence data such as production specification, specific ingredient types, weight of each ingredient, production time, production set temperature and set pressure, and time sequence data such as temperature, pressure, rotating speed, current, power, upper bolt position and the like when the internal mixer is produced;
inputting non-time sequence data such as the processed production specification, the specific ingredient type, the weight of each ingredient, the production time and the like into a machine learning model, and comprehensively predicting to obtain a quality inspection prediction result (a Mooney index); the processed time sequence data (mainly comprising temperature, pressure, current, rotor rotating speed, power, upper bolt position and the like) are input into a deep learning model, and a Mooney index (rubber quality evaluation index) is obtained by integrating the two models.
Further, in the primary parameter recommendation stage, production data (including non-time series data such as production specification, specific ingredient type, weight of each ingredient, production time, production set temperature and set pressure, time series data such as temperature, pressure, rotation speed and current, power and upper bolt position during the production of an internal mixer) and quality inspection data (Mooney index) of the same specification product (rubber banburying production line) in the historical data are taken as input data, and machine set parameter recommendation values during the next batch production under the same product specification are obtained through a primary parameter recommendation model, wherein the method comprises the following steps: setting temperature, setting rotation speed, setting time and setting pressure value. The primary parameter recommended value can be used for guiding the parameter setting of the production machine when the next batch with the same specification is produced.
In the second-level parameter recommendation stage, the production data and the real quality inspection data of the current batch of products are used as input, and the production process parameter recommendation value under the same product specification is obtained through a second-level parameter recommendation model.
Fusing the recommended values of the machine set parameters and the recommended values of the production process parameters which meet the constraint conditions to form the recommended values of the production process parameters under the same product specification. The complete parameter recommendation under the specification is obtained and is used for the process department to adjust the parameter setting of the production line.
For example, the following parameter recommendations under the same specification of the complete rubber banburying production line are obtained:
production time: 2023-07-21 09:29:18;
production specification: l-1MB-J367-08;
production batch: 237L090258013;
the production raw materials are as follows: a rubber raw material;
the weight of the production raw materials is as follows: 329.5;
the production state is as follows: 2;
production temperature: 143.0;
production stops working: 673.0;
setting a Mooney time: 79;
setting banburying time: 99, a step of;
setting a production temperature: 173.3;
setting the rotating speed: 53;
setting production and working: 171;
the Mooney quality inspection index: 95.3.
compared with the prior art, the process parameter recommendation method and system of the industrial assembly line can process various types of data on the production line, rapidly predicts the production quality inspection prediction result of batches of the assembly line by means of multiple models, does not need to wait for a lengthy quality inspection process, simultaneously generates primary parameter recommendation, can timely recommend and adjust process parameters after one batch of production is finished by means of the primary parameter recommendation, effectively helps the production line to improve production quality and stability, and finally can accumulate a large amount of historical production and quality inspection data to realize specification parameter recommendation and provide reference for the process department to adjust production process parameters.
In the description of the present invention, it should be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
In the present invention, unless explicitly specified and limited otherwise, the terms "mounted," "connected," "secured," and the like are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium; may be a communication between two elements or an interaction between two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In the present invention, unless expressly stated or limited otherwise, a first feature is "on" or "under" a second feature, which may be in direct contact with the first and second features, or in indirect contact with the first and second features via an intervening medium. Moreover, a first feature "above," "over" and "on" a second feature may be a first feature directly above or obliquely above the second feature, or simply indicate that the first feature is higher in level than the second feature. The first feature being "under", "below" and "beneath" the second feature may be the first feature being directly under or obliquely below the second feature, or simply indicating that the first feature is level lower than the second feature.
In the description of the present specification, the terms "one embodiment," "some embodiments," "examples," "particular examples," or "some examples," etc., refer to particular features, structures, materials, or characteristics described in connection with the embodiment or example as being included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that alterations, modifications, substitutions and variations may be made in the above embodiments by those skilled in the art within the scope of the invention.
Claims (10)
1. The technological parameter recommendation method of the industrial assembly line is characterized by comprising the following steps of:
reading production data of a current batch of products of an industrial assembly line;
analyzing the production data into continuous variable data, category variable data and time-ordered data, and carrying out data processing according to different data types, wherein the method comprises the following steps:
taking continuous variable data and category variable data as input, inputting a plurality of trained machine learning models, inputting a fusion-trained machine learning model, and carrying out multi-model fusion prediction to obtain a quality inspection prediction result predicted by the machine learning model;
and taking the time sequence data as input, inputting the trained deep learning model to obtain a time sequence model prediction result, and integrating the quality inspection prediction result and the time sequence prediction result predicted by the machine learning model to obtain a final quality inspection prediction result of the current batch of products combined with the time sequence data.
2. The process parameter recommendation method of an industrial pipeline of claim 1, wherein the plurality of trained machine learning models are trained by:
taking continuous variable data, category variable data and quality inspection results in historical production data of an industrial assembly line as data sets, and fully learning by adopting a plurality of different first-type machine learning models to obtain a plurality of trained first-type machine learning models; continuously integrating and learning the quality inspection results learned by a plurality of first-class machine learning models in a multi-model fusion mode to obtain a machine learning model subjected to fusion training;
the deep learning model is obtained through training the following steps:
and sending the time sequence data and the quality inspection result in the historical production data of the industrial assembly line into a time sequence deep learning model based on a transducer as a data set, and finally obtaining a trained deep learning model by learning the correlation between the trend among a plurality of time sequence characteristic data and the quality inspection result.
3. The process parameter recommendation method of an industrial pipeline of claim 1, wherein said production data includes product specifications, machine parameters, and process parameters.
4. A process parameter recommendation method for an industrial pipeline as claimed in any one of claims 1 to 3, further comprising:
in the primary parameter recommendation stage, production data and quality inspection data of the same-specification products in the historical data are used as input data, and machine set parameter recommendation values in the next batch production under the same product specification are obtained through a primary parameter recommendation model;
in the second-level parameter recommendation stage, the production data and the real quality inspection data of the current batch of products are used as input, and the production process parameter recommendation value under the same product specification is obtained through a second-level parameter recommendation model;
fusing the recommended values of the machine set parameters and the recommended values of the production process parameters which meet the constraint conditions to form the recommended values of the production process parameters under the same product specification.
5. The process parameter recommendation method of an industrial pipeline of claim 4, wherein the primary parameter recommendation model and the secondary parameter recommendation model are trained by:
and taking the historical machine setting parameters, production process parameters, production data and quality inspection results under the same product specification as data sets, inputting parameter recommendation models based on genetic algorithm, and respectively training to obtain a primary parameter recommendation model taking the machine setting parameters as output and a secondary parameter recommendation model taking the production process parameters as output.
6. The process parameter recommendation method of an industrial pipeline of claim 5, wherein preprocessing data in the data set prior to model training comprises:
for continuous variable data, performing data preprocessing by adopting feature extraction, feature screening, feature generation, repeated data removal, outlier removal, feature generation and/or normalization;
aiming at category type variable data, category coding, independent heat coding, feature coding and/or feature embedding are adopted to conduct data preprocessing on the data;
and aiming at time sequence data, performing data preprocessing by adopting time sequence segmentation, length complementation, data complementation, extraction of time sequence trend characteristics and/or normalization.
7. The process parameter recommendation method of an industrial pipeline of claim 6, wherein extracting the time series trend feature comprises employing a period term trend term decomposition method assuming that the time series data is composed of a trend term and a period term, namely:
；
wherein y is the extracted time sequence trend characteristic; trend item T is obtained through sliding window and mean value obtaining mode t The original value and the trend term are differenced to obtain a period term S t And finally realizing trend decomposition.
8. The process parameter recommendation method of an industrial pipeline of claim 2, wherein the first type of machine learning model comprises XGBoost, catBoost, lightGBM and Gradient Boosting Regressor models.
9. The process parameter recommendation method of an industrial pipeline according to claim 2, wherein the time sequence deep learning model based on a transducer adopts a deep learning model based on an attention mechanism, and an attention mechanism formula is expressed as follows:
；
wherein the method comprises the steps ofFor a sparse matrix obtained by a sparsity measure, Q, K, V are the input data X multiplied by the learnable matrix W, respectively Q ,W K ,W V The input matrix d k Vector dimensions for Q and K;
the evaluation and calculation mode for sparsity is to calculate the relative entropy of the probability distribution of the corresponding attention moment array of Q by using KL (karma) sparsity, wherein the sparsity evaluation formula of the ith value is as follows:
；
wherein q i And k j For elements in Q and K in the attention mechanism, i and j are the sequence numbers of the elements, d k Vector dimension of K, L k Is the length of K, e is a constant, and the superscript T denotes the transpose of the matrix.
10. A process parameter recommendation system for an industrial pipeline comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method of any of the preceding claims 1 to 9 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311815929.5A CN117495070A (en) | 2023-12-27 | 2023-12-27 | Technological parameter recommendation method and system for industrial assembly line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311815929.5A CN117495070A (en) | 2023-12-27 | 2023-12-27 | Technological parameter recommendation method and system for industrial assembly line |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117495070A true CN117495070A (en) | 2024-02-02 |
Family
ID=89678549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311815929.5A Pending CN117495070A (en) | 2023-12-27 | 2023-12-27 | Technological parameter recommendation method and system for industrial assembly line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117495070A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117762113A (en) * | 2024-02-22 | 2024-03-26 | 宁波数益工联科技有限公司 | Automatic monitoring iterative parameter adjusting method and system based on integrated model |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764597A (en) * | 2018-04-02 | 2018-11-06 | 华南理工大学 | A kind of product quality control method based on integrated study |
| CN111709630A (en) * | 2020-06-08 | 2020-09-25 | 深圳乐信软件技术有限公司 | Voice quality inspection method, device, equipment and storage medium |
| CN112464094A (en) * | 2020-11-30 | 2021-03-09 | 泰康保险集团股份有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113176761A (en) * | 2021-04-28 | 2021-07-27 | 西安电子科技大学 | Machine learning-based multi-feature thin plate part quality prediction and process parameter optimization |
| CN113435699A (en) * | 2021-05-24 | 2021-09-24 | 深圳大学 | Intelligent quality control method and system |
| CN114519469A (en) * | 2022-02-22 | 2022-05-20 | 重庆大学 | Construction method of multivariate long sequence time sequence prediction model based on Transformer framework |
| CN115456460A (en) * | 2022-09-30 | 2022-12-09 | 昆明理工大学 | Multi-quality index output prediction method and system for tobacco shred loosening and conditioning process |
| CN116307289A (en) * | 2023-05-19 | 2023-06-23 | 苏州益诚隆纺织科技有限公司 | Textile processing procedure parameter detection and prediction method, system and storage medium |
| CN116662925A (en) * | 2023-04-23 | 2023-08-29 | 沈阳化工大学 | Industrial process soft measurement method based on weighted sparse neural network |
| CN117009859A (en) * | 2023-09-26 | 2023-11-07 | 深圳市魔数智擎人工智能有限公司 | Feature stitching method and system based on memory calculation |
-
2023
- 2023-12-27 CN CN202311815929.5A patent/CN117495070A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764597A (en) * | 2018-04-02 | 2018-11-06 | 华南理工大学 | A kind of product quality control method based on integrated study |
| CN111709630A (en) * | 2020-06-08 | 2020-09-25 | 深圳乐信软件技术有限公司 | Voice quality inspection method, device, equipment and storage medium |
| CN112464094A (en) * | 2020-11-30 | 2021-03-09 | 泰康保险集团股份有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113176761A (en) * | 2021-04-28 | 2021-07-27 | 西安电子科技大学 | Machine learning-based multi-feature thin plate part quality prediction and process parameter optimization |
| CN113435699A (en) * | 2021-05-24 | 2021-09-24 | 深圳大学 | Intelligent quality control method and system |
| CN114519469A (en) * | 2022-02-22 | 2022-05-20 | 重庆大学 | Construction method of multivariate long sequence time sequence prediction model based on Transformer framework |
| CN115456460A (en) * | 2022-09-30 | 2022-12-09 | 昆明理工大学 | Multi-quality index output prediction method and system for tobacco shred loosening and conditioning process |
| CN116662925A (en) * | 2023-04-23 | 2023-08-29 | 沈阳化工大学 | Industrial process soft measurement method based on weighted sparse neural network |
| CN116307289A (en) * | 2023-05-19 | 2023-06-23 | 苏州益诚隆纺织科技有限公司 | Textile processing procedure parameter detection and prediction method, system and storage medium |
| CN117009859A (en) * | 2023-09-26 | 2023-11-07 | 深圳市魔数智擎人工智能有限公司 | Feature stitching method and system based on memory calculation |
Non-Patent Citations (2)
| Title |
|---|
| 温兰钰: "基于任务自适应采样网络的通用长序列建模", 《中国学位论文全文数据库》, 5 June 2023 (2023-06-05), pages 21 - 24 * |
| 王青天等编著: "《Python金融大数据风控建模实践：基于机器学习》", vol. 2020, 30 June 2020, 北京：机械工业出版社, pages: 22 - 23 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117762113A (en) * | 2024-02-22 | 2024-03-26 | 宁波数益工联科技有限公司 | Automatic monitoring iterative parameter adjusting method and system based on integrated model |
| CN117762113B (en) * | 2024-02-22 | 2024-05-10 | 宁波数益工联科技有限公司 | Automatic monitoring iterative parameter adjusting method and system based on integrated model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108959728B (en) | Radio frequency device parameter optimization method based on deep learning | |
| CN117495070A (en) | Technological parameter recommendation method and system for industrial assembly line | |
| CN110685868A (en) | Wind turbine generator fault detection method and device based on improved gradient elevator | |
| Jiao et al. | A novel learning rate function and its application on the SVD++ recommendation algorithm | |
| CN112987666A (en) | Power plant unit operation optimization regulation and control method and system | |
| CN103605909A (en) | Water quality predication method based on grey theory and support vector machine | |
| CN115878603A (en) | Water quality missing data interpolation algorithm based on K nearest neighbor algorithm and GAN network | |
| CN111985825A (en) | Crystal face quality evaluation method for roller mill orientation instrument | |
| CN108390380A (en) | A kind of transformer state parameter trend forecasting method and system | |
| Tembusai et al. | K-nearest neighbor with k-fold cross validation and analytic hierarchy process on data classification | |
| CN111949640A (en) | Intelligent parameter adjusting method and system based on industrial big data | |
| CN115879369A (en) | Coal mill fault early warning method based on optimized LightGBM algorithm | |
| CN116956120A (en) | Prediction method for water quality non-stationary time sequence based on improved TFT model | |
| CN115169453A (en) | Hot continuous rolling width prediction method based on density clustering and depth residual error network | |
| CN115936241A (en) | Method and system for predicting rhythm of electric power construction operation | |
| CN108491477B (en) | Neural network recommendation method based on multi-dimensional cloud and user dynamic interest | |
| CN114429238A (en) | Wind turbine generator fault early warning method based on space-time feature extraction | |
| CN116562454A (en) | Manufacturing cost prediction method applied to BIM long-short-time attention mechanism network | |
| CN116662832A (en) | Training sample selection method based on clustering and active learning | |
| CN111582619B (en) | Adaptive design method based on correlation and dependency analysis | |
| CN110955811B (en) | Power data classification method and system based on naive Bayes algorithm | |
| CN115687899B (en) | Hybrid feature selection method based on high-dimensional spinning data | |
| CN116722548B (en) | Photovoltaic power generation prediction method based on time sequence model and related equipment | |
| Wang et al. | KERNEL PRINCIPAL COMPONENT ANALYSIS: RADIAL BASIS FUNCTION NEURAL NETWORKS–BASED SOFT-SENSOR MODELING OF POLYMERIZING PROCESS OPTIMIZED BY CULTURAL DIFFERENTIAL EVOLUTION ALGORITHM | |
| Xu et al. | A Top-N recommendation algorithm based on graph convolutional network that integrates basic user information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |