CN116956747A - Method for building machine learning modeling platform based on AI (advanced technology attachment) capability - Google Patents
Method for building machine learning modeling platform based on AI (advanced technology attachment) capability Download PDFInfo
- Publication number
- CN116956747A CN116956747A CN202311087327.2A CN202311087327A CN116956747A CN 116956747 A CN116956747 A CN 116956747A CN 202311087327 A CN202311087327 A CN 202311087327A CN 116956747 A CN116956747 A CN 116956747A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- algorithm
- machine learning
- capability
- 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 44
- 238000010801 machine learning Methods 0.000 title claims abstract description 24
- 238000005516 engineering process Methods 0.000 title description 9
- 230000000694 effects Effects 0.000 claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000004140 cleaning Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000002360 preparation method Methods 0.000 claims abstract description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 64
- 230000008569 process Effects 0.000 claims description 15
- 238000012544 monitoring process Methods 0.000 claims description 9
- 230000008439 repair process Effects 0.000 claims description 8
- 238000011156 evaluation Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 230000000007 visual effect Effects 0.000 claims description 7
- 238000004519 manufacturing process Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 239000013589 supplement Substances 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000003066 decision tree Methods 0.000 claims description 5
- 238000012417 linear regression Methods 0.000 claims description 5
- 238000012706 support-vector machine Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 230000006399 behavior Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 3
- 230000009469 supplementation Effects 0.000 claims description 3
- 238000012300 Sequence Analysis Methods 0.000 claims description 2
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012731 temporal analysis Methods 0.000 description 3
- 238000000700 time series analysis Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000004064 recycling Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a method for building a machine learning modeling platform based on AI capability, which belongs to the field of building methods of modeling platforms, and comprises the steps of collecting data and acquiring effective modeling data; selecting a model frame, and fully knowing the preparation items so as to determine the model frame of the prediction task; analyzing the data; preparing data, cleaning and preprocessing the data and performing characteristic engineering; training and evaluating a model, training the model and evaluating the effect of the machine learning model used on the data; lifting a model result; and determining a model, and taking the complexity and effect factors of the model into consideration. According to the application, the model type most suitable for the prediction task is automatically determined by comparing and matching the models of the previous same prediction task, so that the problem that a common business person does not know how to select a proper model for modeling is solved, the modeling speed is further improved, and the built model has a better effect.
Description
Technical Field
The application relates to a method for building a modeling platform, in particular to a method for building a machine learning modeling platform based on AI (advanced technology attachment) capability.
Background
With the rapid development of internet technology and the rapid progress of enterprise digitization, enterprises and institutions accumulate a large amount of business data, and how to generate value for the accumulated large data becomes one of the important research subjects of enterprise institutions in recent years. With the development of technology in the fields of artificial intelligence and machine learning, data modeling provides a new technical means for data enabling. By modeling and analyzing the mass data, service support and high-accuracy strategy recommendation can be provided for multiple scenes such as accurate marketing, risk prevention and control, financial trust and the like. Through a standardized, procedural and visual modeling platform, common business personnel can realize a main modeling procedure, and the data enabling cost is reduced.
In the prior related technology, in business intelligent application and a visual data analysis platform, business intelligent functions such as data statistics analysis, chart generation, report making, data large screen and the like can be realized through a visual data analysis component; in the visual modeling platform, a modeling flow can be realized through a drag type or modularized operation mode, and a model file is output to build a model. After the model is built, training and evaluating the model are needed, the training and evaluating model belongs to the prior known public technology, the most common model training and evaluating method is to enable the model to train and learn on the whole training set, then predict the whole training set again, calculate the prediction error of the prediction result and the label by means of Mean_squared_error (MSE), square to obtain the RMSE of the regression model on the whole training set, compare the obtained RMSE with a preset reference value, and indicate that the model effect evaluation reaches the standard if the RMSE is smaller than the reference value.
However, the existing modeling related technology does not consider the problem that a common business person does not clearly select a proper model for modeling, so that the modeling speed is low, and the built model effect is very general. Therefore, a person skilled in the art provides a method for building a machine learning modeling platform based on AI capability, so as to solve the problems set forth in the background art.
Disclosure of Invention
The application aims to provide a method for building a machine learning modeling platform based on AI capability, which is used for automatically determining the model type most suitable for the prediction task through comparing and matching the models of the previous same prediction task, and avoiding the error selection of models by common service personnel, so as to solve the problem that the common service personnel cannot clearly select the proper models for modeling, further improve the modeling speed, and enable the built models to have better effect, thereby solving the problems in the background technology.
In order to achieve the above purpose, the present application provides the following technical solutions:
a method for constructing a machine learning modeling platform based on AI capability comprises the following steps:
collecting data, and acquiring effective modeling data;
selecting a model frame, and fully knowing the preparation items so as to determine the model frame of the prediction task;
analyzing the data, and knowing the data to be modeled by using a descriptive statistical method and a visual method;
preparing data, cleaning and preprocessing the data and performing characteristic engineering, and converting the data into a form in which a model can be trained;
training and evaluating a model, training the model and evaluating the effect of the machine learning model on data, and presetting a reference value for comparing the effects after the subsequent model debugging;
the model result is promoted, and the model algorithm is debugged and optimized to obtain the best result;
determining a model, taking the complexity and effect factors of the model into consideration to determine the model to be used finally, deploying the model into a production system, and formulating a corresponding monitoring system.
As a further scheme of the application: the model framework for determining the prediction task in the selection of the model framework is specifically as follows:
the obtained modeling data is subjected to feature recognition, the identified feature is marked as Pi, i=1· n;
after the project is fully known to be prepared, determining that the project prediction task is Q;
collecting and counting a predictive model with a previous predictive task of Q, it is marked as Wj and, j=1· n;
feature data of Wj is collected and counted, this is marked as pi and is referred to as pi, i=1· n;
the accuracy of the predicted results of Wj is collected and counted, this is marked as Gj and, j=1· n;
matching Pi of each Wj with the identification feature Pi, and screening the Wj corresponding to the Pi if the matching degree of the Pi and the identification feature Pi reaches a preset value;
counting all the screened Wj and marking the model types;
the accuracy of the prediction results corresponding to all the screened Wj is called, and a accuracy list is generated, wherein the ranking higher the accuracy of the prediction results is, the higher the ranking is;
classifying the Wj of the first ten names in the precision list according to the model types marked by the Wj;
and determining the model type with the largest number as a model frame of the prediction task according to the ten classified Wj.
As still further aspects of the application: the model types comprise a linear regression class prediction model, a decision tree class prediction model, a neural network class prediction model, a support vector machine class prediction model and a time sequence analysis class prediction model.
As still further aspects of the application: the data cleaning and preprocessing comprises the steps of removing repeated data, processing missing data and normalizing data.
As still further aspects of the application: in the process of processing missing data, evaluating the missing part of the data, discarding the data if the evaluation result is that the missing degree is larger than a preset value, repairing the data if the evaluation result is that the missing degree is smaller than or equal to the preset value, and recycling the repaired data.
As still further aspects of the application: the specific process of data restoration is as follows:
the part of the data which is not deleted is marked as A, and the part which is deleted is marked as a;
similarity matching is carried out on the A and other complete data, complete data with similarity reaching more than 90 percent are screened out, the mark is Bi, and the mixture is prepared from the following components, i=1·····n;
and then marking the partial data corresponding to a in Bi as Ci, i=1· n;
marking the data with the largest repetition number in Ci as C supplement;
the repair data for a was supplemented with C, i.e. the data after repair was (a+c supplementation).
As still further aspects of the application: the specific process of the model algorithm debugging is as follows: based on control variable data of modeling data, different event functions are written on corresponding prediction models, the prediction models are driven to operate, parameters are adjusted according to the operating states of field device components, and the prediction models and the corresponding field device components operate synchronously.
As still further aspects of the application: the specific process of the model algorithm tuning is as follows:
parameter adjustment: the behavior of the algorithm is changed by adjusting the parameters of the algorithm, so that the performance of the algorithm is improved;
problem modeling: for different problems, different algorithms are needed, so that when the algorithm is selected, the problem needs to be modeled first, and the algorithm which is most suitable for the characteristics of the problem is selected;
algorithm combination: by combining a plurality of algorithms, the performance of the algorithms is improved;
local search: when the algorithm falls into a local optimal solution, adding a local search strategy into the algorithm, and improving the global search capability of the algorithm;
parallel computing: through parallel calculation, the running speed of the algorithm is increased, so that the performance of the algorithm is improved;
mixing and optimizing: by combining the algorithm with other optimization methods, the performance of the algorithm can be improved.
As still further aspects of the application: the specific process of determining the model to be used finally is as follows:
assigning a weight value X to the complexity of the model and assigning a weight value Y to the effect factors of the model;
the complexity of the model is digitalized and marked as S, and the effect factors of the model are digitalized and marked as U;
calculating the cost performance V of the model, wherein V=U×Y-S×X;
and screening out the model with the maximum V value as the model to be used finally.
As still further aspects of the application: the monitoring system comprises a monitoring camera arranged in the production system.
Compared with the prior art, the application has the beneficial effects that:
1. according to the application, the model type most suitable for the prediction task is automatically determined by comparing and matching the models of the previous same prediction task, so that the problem that a common business person cannot clearly select a proper model to model is solved, the modeling speed is further improved, and the built model has a better effect.
2. According to the application, the missing data can be evaluated, the missing data with the missing degree smaller than or equal to the preset value is repaired, the missing part of the missing data is complemented by comparing and matching the missing data with the complete data, so that the missing part of the missing data tends to be complete, the repairable data is effectively recovered and used, and the number of sample data is increased to improve the accuracy of model prediction.
3. The application can further improve the performance of the algorithm through the set model algorithm tuning, and when determining the model to be used finally, the application selects the proper model as the model to be used finally by comprehensively considering the complexity and effect factors of the model.
Drawings
Fig. 1 is a flowchart of a method for building a machine learning modeling platform based on AI capabilities.
Detailed Description
In order to better understand the above technical solutions, the following detailed description will refer to the accompanying drawings and specific embodiments.
Referring to fig. 1, in an embodiment of the present application, a method for building a machine learning modeling platform based on AI capability includes the following steps:
collecting data, and acquiring effective modeling data;
selecting a model frame, and fully knowing the preparation items so as to determine the model frame of the prediction task;
analyzing the data, and knowing the data to be modeled by using a descriptive statistical method and a visual method;
preparing data, cleaning and preprocessing the data and performing characteristic engineering, and converting the data into a form in which a model can be trained;
training and evaluating a model, training the model and evaluating the effect of the machine learning model on data, and presetting a reference value for comparing the effects after the subsequent model debugging;
the model result is promoted, and the model algorithm is debugged and optimized to obtain the best result;
determining a model, taking the complexity and effect factors of the model into consideration to determine the model to be used finally, deploying the model into a production system, and formulating a corresponding monitoring system.
According to the application, the model type most suitable for the prediction task is automatically determined by comparing and matching the models of the previous same prediction task, so that the problem that a common business person cannot clearly select a proper model to model is solved, the modeling speed is further improved, and the built model has a better effect.
In this embodiment: the model framework for determining the prediction task in the selection of the model framework is specifically as follows: the obtained modeling data is subjected to feature recognition, the identified feature is marked as Pi, i=1· n; after the project is fully known to be prepared, determining that the project prediction task is Q; collecting and counting a predictive model with a previous predictive task of Q, it is marked as Wj and, j=1· n; feature data of Wj is collected and counted, this is marked as pi and is referred to as pi, i=1· n; the accuracy of the predicted results of Wj is collected and counted, this is marked as Gj and, j=1· n; matching Pi of each Wj with the identification feature Pi, and screening the Wj corresponding to the Pi if the matching degree of the Pi and the identification feature Pi reaches a preset value; counting all the screened Wj and marking the model types; the accuracy of the prediction results corresponding to all the screened Wj is called, and a accuracy list is generated, wherein the ranking higher the accuracy of the prediction results is, the higher the ranking is; classifying the Wj of the first ten names in the precision list according to the model types marked by the Wj; and determining the model type with the largest number as a model frame of the prediction task according to the ten classified Wj. In the setting, the application can rapidly determine the model type conforming to the current prediction task, thereby facilitating modeling by common business personnel.
In this embodiment: model classes include linear regression class prediction models, decision tree class prediction models, neural network class prediction models, support vector machine class prediction models, and time series analysis class prediction models. Linear regression is a model based on linear equations that predicts future values of a target variable by fitting an independent variable to a dependent variable. The decision tree predicts by dividing the data set into subsets and building corresponding decision rules. The neural network is a model for simulating human brain neurons, and predicts by learning the rules in the data set. The support vector machine predicts by mapping the data into a high-dimensional space and creating separate hyperplanes in this space. Time series analysis predicts future trends and results by analysis and processing of historical data. Each predictive model has its advantages and disadvantages. Linear regression is applicable where the dataset is relatively simple and the independent and dependent variables are in linear relationship. The decision tree is suitable for the situations of multiple attributes and complex data sets, but the phenomenon of over fitting is easy to occur. The neural network has strong learning ability and adaptability, but is easy to generate local optimal solution. Support vector machines perform well in handling high-dimensional data and non-linear relationships, but require significant computational resources. Time series analysis is applicable to datasets with significant periodic variations, but it is necessary to ensure the accuracy and integrity of the data.
In this embodiment: the data cleaning and preprocessing comprises removing repeated data, processing missing data and normalizing data. The setting effectively filters invalid data, and avoids affecting the accuracy of the prediction result.
In this embodiment: in the process of processing missing data, evaluating the missing part of the data, discarding the data if the evaluation result is that the missing degree is larger than a preset value, repairing the data if the evaluation result is that the missing degree is smaller than or equal to the preset value, and recycling the repaired data. The device can effectively recycle repairable data, and increase the number of sample data to improve the accuracy of model prediction.
In this embodiment: the specific process of data repair is as follows: the part of the data which is not deleted is marked as A, and the part which is deleted is marked as a; similarity matching is carried out on the A and other complete data, complete data with similarity reaching more than 90 percent are screened out, the mark is Bi, and the mixture is prepared from the following components, i=1·····n; and then marking the partial data corresponding to a in Bi as Ci, i=1· n; marking the data with the largest repetition number in Ci as C supplement; the repair data for a was supplemented with C, i.e. the data after repair was (a+c supplementation). Data repair can quickly supplement partially missing data, and then supplement similar data to data with a low degree of missing, so that the data tends to be complete.
In this embodiment: the specific process of model algorithm debugging is as follows: based on control variable data of modeling data, different event functions are written on corresponding prediction models, the prediction models are driven to operate, parameters are adjusted according to the operating states of field device components, and the prediction models and the corresponding field device components operate synchronously. The adaptability and the accuracy of the model can be improved by debugging the model algorithm.
In this embodiment: the specific process of model algorithm tuning is as follows: parameter adjustment: the behavior of the algorithm is changed by adjusting the parameters of the algorithm, so that the performance of the algorithm is improved; problem modeling: for different problems, different algorithms are needed, so that when the algorithm is selected, the problem needs to be modeled first, and the algorithm which is most suitable for the characteristics of the problem is selected; algorithm combination: by combining a plurality of algorithms, the performance of the algorithms is improved; local search: when the algorithm falls into a local optimal solution, adding a local search strategy into the algorithm, and improving the global search capability of the algorithm; parallel computing: through parallel calculation, the running speed of the algorithm is increased, so that the performance of the algorithm is improved; mixing and optimizing: by combining the algorithm with other optimization methods, the performance of the algorithm can be improved. Model algorithm tuning can further improve the performance of the algorithm.
In this embodiment: the specific process of determining the last model to be used is: assigning a weight value X to the complexity of the model and assigning a weight value Y to the effect factors of the model; the complexity of the model is digitalized and marked as S, and the effect factors of the model are digitalized and marked as U; calculating the cost performance V of the model, wherein V=U×Y-S×X; and screening out the model with the maximum V value as the model to be used finally. The setting can integrate the complexity and effect factors of the model, and select a proper model as the model to be used finally.
In this embodiment: the monitoring system comprises a monitoring camera arranged in the production system.
According to the application, the model type most suitable for the prediction task is automatically determined by comparing and matching the models of the previous same prediction task, so that the problem that a common business person cannot clearly select a proper model to model is solved, the modeling speed is further improved, and the built model has a better effect. In addition, the application can evaluate the missing data, restore the missing data with the missing degree smaller than or equal to the preset value, supplement the missing part of the missing data by comparing and matching the missing data with the complete data, so that the missing part of the missing data tends to be complete, further effectively recover the repairable data for use, and increase the number of sample data to improve the accuracy of model prediction. The performance of the algorithm can be further improved through the set model algorithm tuning, and when the model to be used finally is determined, a proper model is selected as the model to be used finally by comprehensively considering the complexity and effect factors of the model.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
The foregoing description is only a preferred embodiment of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art, who is within the scope of the present application, should make equivalent substitutions or modifications according to the technical solution of the present application and the inventive concept thereof, and should be covered by the scope of the present application.
Claims (10)
1. The method for constructing the machine learning modeling platform based on the AI capability is characterized by comprising the following steps of:
collecting data, and acquiring effective modeling data;
selecting a model frame, and fully knowing the preparation items so as to determine the model frame of the prediction task;
analyzing the data, and knowing the data to be modeled by using a descriptive statistical method and a visual method;
preparing data, cleaning and preprocessing the data and performing characteristic engineering, and converting the data into a form in which a model can be trained;
training and evaluating a model, training the model and evaluating the effect of the machine learning model on data, and presetting a reference value for comparing the effects after the subsequent model debugging;
the model result is promoted, and the model algorithm is debugged and optimized to obtain the best result;
determining a model, taking the complexity and effect factors of the model into consideration to determine the model to be used finally, deploying the model into a production system, and formulating a corresponding monitoring system.
2. The method for constructing a machine learning modeling platform based on AI capability according to claim 1, wherein the model framework for determining the prediction task in the selection of the model framework is specifically:
the obtained modeling data is subjected to feature recognition, the identified feature is marked as Pi, i=1· n;
after the project is fully known to be prepared, determining that the project prediction task is Q;
collecting and counting a predictive model with a previous predictive task of Q, it is marked as Wj and, j=1· n;
feature data of Wj is collected and counted, this is marked as pi and is referred to as pi, i=1· n;
the accuracy of the predicted results of Wj is collected and counted, this is marked as Gj and, j=1· n;
matching Pi of each Wj with the identification feature Pi, and screening the Wj corresponding to the Pi if the matching degree of the Pi and the identification feature Pi reaches a preset value;
counting all the screened Wj and marking the model types;
the accuracy of the prediction results corresponding to all the screened Wj is called, and a accuracy list is generated, wherein the ranking higher the accuracy of the prediction results is, the higher the ranking is;
classifying the Wj of the first ten names in the precision list according to the model types marked by the Wj;
and determining the model type with the largest number as a model frame of the prediction task according to the ten classified Wj.
3. The method for building the AI-capability-based machine learning modeling platform according to claim 2, wherein the model types include a linear regression class prediction model, a decision tree class prediction model, a neural network class prediction model, a support vector machine class prediction model, and a time sequence analysis class prediction model.
4. The method for constructing a machine learning modeling platform based on AI capabilities according to claim 1, wherein the data cleaning and preprocessing includes removing duplicate data, processing missing data, and normalizing data.
5. The method for building the machine learning modeling platform based on the AI capability as claimed in claim 4, wherein in the process of processing the missing data, the missing part of the data is evaluated, if the evaluation result is that the missing degree is greater than a preset value, the data is abandoned, if the evaluation result is that the missing degree is less than or equal to the preset value, the data is repaired, and the repaired data is recycled.
6. The method for constructing a machine learning modeling platform based on AI capability according to claim 5, wherein the specific process of data restoration is as follows:
the part of the data which is not deleted is marked as A, and the part which is deleted is marked as a;
similarity matching is carried out on the A and other complete data, complete data with similarity reaching more than 90 percent are screened out, the mark is Bi, and the mixture is prepared from the following components, i=1·····n;
and then marking the partial data corresponding to a in Bi as Ci, i=1· n;
marking the data with the largest repetition number in Ci as C supplement;
the repair data for a was supplemented with C, i.e. the data after repair was (a+c supplementation).
7. The method for constructing the machine learning modeling platform based on the AI capability as claimed in claim 1, wherein the specific process of the model algorithm debugging is as follows: based on control variable data of modeling data, different event functions are written on corresponding prediction models, the prediction models are driven to operate, parameters are adjusted according to the operating states of field device components, and the prediction models and the corresponding field device components operate synchronously.
8. The method for constructing the machine learning modeling platform based on the AI capability as claimed in claim 1, wherein the specific process of optimizing the model algorithm is as follows:
parameter adjustment: the behavior of the algorithm is changed by adjusting the parameters of the algorithm, so that the performance of the algorithm is improved;
problem modeling: for different problems, different algorithms are needed, so that when the algorithm is selected, the problem needs to be modeled first, and the algorithm which is most suitable for the characteristics of the problem is selected;
algorithm combination: by combining a plurality of algorithms, the performance of the algorithms is improved;
local search: when the algorithm falls into a local optimal solution, adding a local search strategy into the algorithm, and improving the global search capability of the algorithm;
parallel computing: through parallel calculation, the running speed of the algorithm is increased, so that the performance of the algorithm is improved;
mixing and optimizing: by combining the algorithm with other optimization methods, the performance of the algorithm can be improved.
9. The method for building the machine learning modeling platform based on the AI capability according to claim 1, wherein the specific process of determining the model to be used finally is as follows:
assigning a weight value X to the complexity of the model and assigning a weight value Y to the effect factors of the model;
the complexity of the model is digitalized and marked as S, and the effect factors of the model are digitalized and marked as U;
calculating the cost performance V of the model, wherein V=U×Y-S×X;
and screening out the model with the maximum V value as the model to be used finally.
10. The method for building an AI-capability-based machine learning modeling platform of claim 1, wherein the monitoring architecture includes a monitoring camera disposed in a production system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311087327.2A CN116956747A (en) | 2023-08-28 | 2023-08-28 | Method for building machine learning modeling platform based on AI (advanced technology attachment) capability |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311087327.2A CN116956747A (en) | 2023-08-28 | 2023-08-28 | Method for building machine learning modeling platform based on AI (advanced technology attachment) capability |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116956747A true CN116956747A (en) | 2023-10-27 |
Family
ID=88458500
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311087327.2A Pending CN116956747A (en) | 2023-08-28 | 2023-08-28 | Method for building machine learning modeling platform based on AI (advanced technology attachment) capability |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116956747A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112801304A (en) * | 2021-03-17 | 2021-05-14 | 中奥智能工业研究院(南京)有限公司 | Automatic data analysis and modeling process |
| CN113609779A (en) * | 2021-08-16 | 2021-11-05 | 深圳力维智联技术有限公司 | Modeling method, device and equipment for distributed machine learning |
| CN114066110A (en) * | 2020-07-29 | 2022-02-18 | 太极计算机股份有限公司 | System for providing machine learning service for user |
| CN114154406A (en) * | 2021-11-22 | 2022-03-08 | 厦门深度赋智科技有限公司 | AI model automatic modeling system based on black box optimizer |
| WO2022083624A1 (en) * | 2020-10-21 | 2022-04-28 | 华为技术有限公司 | Model acquisition method, and device |
| CN114912082A (en) * | 2021-12-15 | 2022-08-16 | 许磊 | General computing task collaboration system |
| CN115080021A (en) * | 2022-05-13 | 2022-09-20 | 北京思特奇信息技术股份有限公司 | Zero code modeling method and system based on automatic machine learning |
| CN115409153A (en) * | 2022-08-08 | 2022-11-29 | 南京丰顿科技股份有限公司 | Attention LSTM-based animal husbandry index prediction method and prediction system |
| CN115413346A (en) * | 2020-02-03 | 2022-11-29 | 强力交易投资组合2018有限公司 | Artificial intelligence selection and configuration |
| CN116629352A (en) * | 2023-04-10 | 2023-08-22 | 苏州互微智速科技有限公司 | Hundred million-level parameter optimizing platform |
-
2023
- 2023-08-28 CN CN202311087327.2A patent/CN116956747A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115413346A (en) * | 2020-02-03 | 2022-11-29 | 强力交易投资组合2018有限公司 | Artificial intelligence selection and configuration |
| CN114066110A (en) * | 2020-07-29 | 2022-02-18 | 太极计算机股份有限公司 | System for providing machine learning service for user |
| WO2022083624A1 (en) * | 2020-10-21 | 2022-04-28 | 华为技术有限公司 | Model acquisition method, and device |
| CN112801304A (en) * | 2021-03-17 | 2021-05-14 | 中奥智能工业研究院(南京)有限公司 | Automatic data analysis and modeling process |
| CN113609779A (en) * | 2021-08-16 | 2021-11-05 | 深圳力维智联技术有限公司 | Modeling method, device and equipment for distributed machine learning |
| CN114154406A (en) * | 2021-11-22 | 2022-03-08 | 厦门深度赋智科技有限公司 | AI model automatic modeling system based on black box optimizer |
| CN114912082A (en) * | 2021-12-15 | 2022-08-16 | 许磊 | General computing task collaboration system |
| CN115080021A (en) * | 2022-05-13 | 2022-09-20 | 北京思特奇信息技术股份有限公司 | Zero code modeling method and system based on automatic machine learning |
| CN115409153A (en) * | 2022-08-08 | 2022-11-29 | 南京丰顿科技股份有限公司 | Attention LSTM-based animal husbandry index prediction method and prediction system |
| CN116629352A (en) * | 2023-04-10 | 2023-08-22 | 苏州互微智速科技有限公司 | Hundred million-level parameter optimizing platform |
Non-Patent Citations (1)
| Title |
|---|
| 林友芳: "交通大数据", pages: 106 - 109 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110232203B (en) | Knowledge distillation optimization RNN short-term power failure prediction method, storage medium and equipment | |
| CN110571792A (en) | Analysis and evaluation method and system for operation state of power grid regulation and control system | |
| CN112910690A (en) | Network traffic prediction method, device and equipment based on neural network model | |
| CN111178585A (en) | Fault reporting amount prediction method based on multi-algorithm model fusion | |
| CN111126820A (en) | Electricity stealing prevention method and system | |
| CN110757510A (en) | Method and system for predicting remaining life of robot | |
| CN113537807A (en) | Enterprise intelligent wind control method and device | |
| JP2023504103A (en) | MODEL UPDATE SYSTEM, MODEL UPDATE METHOD AND RELATED DEVICE | |
| CN113627032A (en) | Intelligent decision method for equipment design/maintenance scheme based on digital twin | |
| CN115545473A (en) | LSTM-based intelligent prediction method for domestic garbage throwing trend | |
| CN115221942A (en) | Equipment defect prediction method and system based on time sequence fusion and neural network | |
| CN110781206A (en) | Method for predicting whether electric energy meter in operation fails or not by learning meter-dismantling and returning failure characteristic rule | |
| CN114154716A (en) | Enterprise energy consumption prediction method and device based on graph neural network | |
| CN117315380A (en) | Deep learning-based pneumonia CT image classification method and system | |
| CN113159441A (en) | Prediction method and device for implementation condition of banking business project | |
| CN109635008B (en) | Equipment fault detection method based on machine learning | |
| CN117252459A (en) | Fruit quality evaluation system based on deep learning | |
| CN109934352B (en) | Automatic evolution method of intelligent model | |
| Chernyshev et al. | Integration of building information modeling and artificial intelligence systems to create a digital twin of the construction site | |
| CN116956747A (en) | Method for building machine learning modeling platform based on AI (advanced technology attachment) capability | |
| Pattnaik et al. | A survey on machine learning techniques used for software quality prediction | |
| CN114298413A (en) | Hydroelectric generating set runout trend prediction method | |
| Zakeri et al. | Automation and Computational Intelligence for Road Maintenance and Management: Advances and Applications | |
| CN110427973A (en) | A kind of classification method towards ambiguity tagging sample | |
| Vijayaganth et al. | Ensemble Machine Learning based Weather Prediction System |
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 |