CN116738551A - Intelligent processing method for acquired data of BIM model - Google Patents
Intelligent processing method for acquired data of BIM model Download PDFInfo
- Publication number
- CN116738551A CN116738551A CN202310996803.6A CN202310996803A CN116738551A CN 116738551 A CN116738551 A CN 116738551A CN 202310996803 A CN202310996803 A CN 202310996803A CN 116738551 A CN116738551 A CN 116738551A
- Authority
- CN
- China
- Prior art keywords
- interval
- scale
- data
- building data
- parameter
- 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.)
- Granted
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 54
- 239000011159 matrix material Substances 0.000 claims abstract description 31
- 238000012545 processing Methods 0.000 claims abstract description 16
- 238000012795 verification Methods 0.000 claims abstract description 16
- 230000002159 abnormal effect Effects 0.000 claims description 41
- 238000000034 method Methods 0.000 claims description 26
- 238000012706 support-vector machine Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 description 88
- 239000011521 glass Substances 0.000 description 14
- 230000000694 effects Effects 0.000 description 12
- 238000012216 screening Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 7
- 238000010276 construction Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008689 nuclear function Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Structural Engineering (AREA)
- Computing Systems (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Architecture (AREA)
- Mathematical Analysis (AREA)
- Civil Engineering (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of data processing, in particular to an intelligent processing method for collected data of a BIM model, which comprises the following steps: building data of different building types are collected, a data set is constructed according to historical building data, all scale parameters in a scale interval are obtained, a corresponding Gaussian kernel function matrix is obtained according to a training set of each scale parameter, an initial proper value of each scale parameter is obtained according to the Gaussian kernel function matrix corresponding to each scale parameter, the optimal scale parameters are screened, cross verification is carried out on all the optimal scale parameters, classification accuracy of each optimal scale parameter is obtained, further a better range is obtained, optimal scale parameters of a target building type are obtained according to the better range, an optimal SVM hyperplane model is built, building data are classified, and a BIM building model is built according to classification results. The invention improves the efficiency and accuracy of the establishment of the optimal SVM hyperplane model, and improves the accuracy of the BIM building model.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an intelligent processing method for collected data of a BIM model.
Background
In the digital transformation and intelligent development of the building industry, an intelligent processing method for collected data of a BIM model is becoming an important field gradually. BIM technology is an information management and collaborative work platform for building and infrastructure projects. BIM technology integrates a variety of information including geometric data, component properties, materials, costs, and progress, etc., by creating a three-dimensional, digitized building model.
When building data is collected, all the building data enter a BIM building information model, and data modeling is carried out on the collected building data, but the collected building data contains abnormal data due to errors or errors in the building data collecting process inevitably, so that the accuracy of the established model is affected. In order to prevent low accuracy of a final model caused by errors or errors in the building data acquisition process, data analysis processing is required to be performed on the acquired building data, so that abnormal classification monitoring on abnormal acquired data is realized, and the reliability of the data entering the BIM building information model is ensured.
At present, abnormal classification monitoring of abnormal collected data is usually carried out through an SVM support vector machine, but when the collected building data is classified through the existing SVM support vector machine, if the scale parameter corresponding to a kernel function is improperly set, the super-plane construction effect is poor, the accuracy of abnormal classification of the collected building data is affected, and the accuracy of a BIM building model is further affected.
Disclosure of Invention
The invention provides an intelligent processing method for acquired data of a BIM model, which aims to solve the existing problems.
The intelligent processing method for the acquired data of the BIM model adopts the following technical scheme:
an embodiment of the present invention provides an intelligent processing method for collected data of a BIM model, including the steps of:
collecting building data of different building types; taking any building type as a target building type, and constructing a data set of an SVM support vector machine according to historical building data of the target building type;
presetting a scale interval and acquiring all scale parameters in the scale interval; acquiring a training set of each scale parameter according to a data set of the SVM support vector machine, and acquiring a Gaussian kernel function matrix corresponding to each scale parameter according to the training set of each scale parameter; acquiring an initial proper value of each scale parameter according to the Gaussian kernel function matrix corresponding to each scale parameter;
acquiring a preferred scale parameter according to the initial proper value of each scale parameter; cross-verifying all the preferred scale parameters to obtain the classification accuracy of each preferred scale parameter;
acquiring a more optimal section according to the initial proper values and the classification accuracy of all the preferred scale parameters in the scale section, and acquiring the optimal scale parameters of the target building type according to the more optimal section;
And constructing an optimal SVM hyperplane model of the target building type according to the optimal scale parameters, classifying building data according to the optimal SVM hyperplane model, and constructing a BIM building model according to classification processing results.
Preferably, the construction of the data set of the SVM support vector machine according to the historical building data of the target building type includes the following specific steps:
acquiring a plurality of historical building data of a target building type, marking all the historical building data by adopting a manual marking method, marking abnormal historical building data as-1, marking normal historical building data as 1, and obtaining a label of each historical building data; and taking all the historical building data of the target building type and the corresponding labels as a data set of the SVM support vector machine.
Preferably, the step of obtaining the gaussian kernel function matrix corresponding to each scale parameter according to the training set of each scale parameter includes the following specific steps:
substituting each scale parameter into a Gaussian kernel function, and taking a training set corresponding to each scale parameter as input of a corresponding Gaussian kernel function to obtain a Gaussian kernel function matrix corresponding to each scale parameter.
Preferably, the obtaining the initial suitable value of each scale parameter according to the gaussian kernel function matrix corresponding to each scale parameter includes the following specific steps:
Taking any one scale parameter as a target scale parameter, and acquiring a kernel function value sequence of each historical building data in a training set of the target scale parameter; acquiring an initial proper value of the target scale parameter for each historical building data in the training set according to the kernel function value sequence of each historical building data in the training set of the target scale parameter; and taking the average value of the initial proper values of the target scale parameters for all the historical building data in the training set as the initial proper value of the target scale parameters.
Preferably, the step of obtaining the kernel function value sequence of each historical building data in the training set of the target scale parameter includes the following specific steps:
acquiring training set of target scale parameters, the firstHistorical building data is AND-division number in Gaussian kernel function matrixThe kernel function value between each of the history building data other than the history building data constitutes +.>A sequence of kernel function values for the historical building data.
Preferably, the obtaining the initial suitable value of the target scale parameter for each historical building data in the training set includes the following specific steps:
wherein ,representing the target scale parameter +.>Initial fit values for the individual historical building data; Training set representing target scale parameter +.>Historical building data; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, and +.>The historical building data are the minimum value of the kernel function values of all normal historical building data with the same specification; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, and +.>The historical building data are the maximum value of the kernel function values of all normal historical building data with different specifications; />Training set representing target scale parameter +.>In the kernel function value sequence of each history building data, all abnormal historiesMaximum value of kernel function values corresponding to building data; />Training set representing target scale parameter +.>Tags for the historical building data; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, the minimum value in the kernel function values corresponding to all abnormal historical building data; />Training set representing target scale parameter +.>And in the kernel function value sequence of each history building data, the maximum value in the kernel function values corresponding to all normal history building data.
Preferably, the obtaining the better interval according to the initial suitable values and the classification accuracy of all the preferred scale parameters in the scale interval includes the following specific steps:
the method comprises the steps of obtaining pearson correlation coefficients between initial suitable values and classification accuracy of all preferred scale parameters in a scale interval, and taking the pearson correlation coefficients as correlation values between the initial suitable values and the classification accuracy; taking the preferred scale parameter corresponding to the maximum classification accuracy in the scale interval as the center parameter of the scale interval; acquiring a first interval and a second interval of the scale interval and classification accuracy of the first interval and the second interval according to the central parameter;
acquiring all scale parameters in a first interval and a second interval, and acquiring initial proper values of each scale parameter in the first interval and the second interval; taking the average value of the initial proper values of all the scale parameters in the first interval as the average initial proper value of the first interval, and taking the average value of the initial proper values of all the scale parameters in the second interval as the average initial proper value of the second interval;
acquiring the preference of the first interval and the second interval according to the average initial fit value and the classification accuracy of the first interval and the second interval; and taking the section with the highest preference degree of the first section and the second section as a more preferred section.
Preferably, the step of obtaining the classification accuracy of the first interval and the second interval and the first interval and the second interval of the scale interval according to the central parameter includes the following specific steps:
taking the optimal scale parameter which is smaller than the central parameter and has the smallest difference value with the central parameter as the left boundary of the first interval, and taking the central parameter as the right boundary of the first interval to obtain the first interval of the scale interval; taking the optimal scale parameter which is larger than the central parameter and has the smallest difference value with the central parameter as the right boundary of the second interval, and taking the central parameter as the left boundary of the second interval to obtain the second interval of the scale interval; taking the average value of the classification accuracy rates corresponding to the left boundary and the right boundary of the first interval as the classification accuracy rate of the first interval, and taking the average value of the classification accuracy rates corresponding to the left boundary and the right boundary of the second interval as the classification accuracy rate of the second interval.
Preferably, the obtaining the preference degree of the first section and the second section according to the average initial suitable value and the classification accuracy of the first section and the second section respectively includes the following specific steps:
wherein , and />Respectively are firstPreference of interval and second interval, +.> and />Average initial fit value of the first interval and the second interval, respectively,/respectively>For the correlation value between the initial fit value and the classification accuracy> and />The classification accuracy rates of the first interval and the second interval are respectively.
Preferably, the obtaining the optimal scale parameter of the target building type according to the better interval includes the following specific steps:
performing a better interval update operation, including: acquiring the preferred scale parameters in the preferred interval according to the initial suitable value of each scale parameter in the preferred interval, performing cross verification on all the preferred scale parameters in the preferred interval to obtain the classification accuracy of each preferred scale parameter in the preferred interval, and acquiring a new preferred interval according to the initial suitable value and the classification accuracy of all the preferred scale parameters in the preferred interval;
repeating the updating operation of the more optimal section until the obtained first section and the obtained second section of the latest more optimal section have no scale parameter with higher classification accuracy than the classification accuracy of the central parameter of the latest more optimal section, and stopping iteration; taking the center parameter of the latest more optimal interval as the optimal scale parameter.
The technical scheme of the invention has the beneficial effects that: according to the invention, by collecting building data of different building types, taking historical building data as a data set, acquiring all scale parameters in a scale interval, acquiring a corresponding Gaussian kernel function matrix according to a training set of each scale parameter, acquiring an initial proper value of each scale parameter according to the Gaussian kernel function matrix corresponding to each scale parameter, primarily screening out the scale parameters of the kernel functions conforming to the distribution characteristics of the data of different building types, taking the scale parameters as optimized scale parameters, carrying out cross verification on all the optimized scale parameters to obtain the classification accuracy of each optimized scale parameter, acquiring a better interval according to the correlation value of the initial proper value and the classification accuracy, and searching the optimal scale parameter by continuous iteration to establish an optimal SVM hyperplane model. The pre-screening of the scale parameters is carried out according to the initial proper value, the calculation amount of subsequent cross verification is reduced, the efficiency of establishing the optimal SVM hyperplane model is improved, the accuracy of the optimal SVM hyperplane model is ensured by continuously and iteratively searching the optimal scale parameters, the classification accuracy of building data is improved, and the BIM building model constructed according to the classified building data is more accurate.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of the method for intelligent processing of acquired data for BIM model of the present invention;
FIG. 2 is a corresponding Gaussian kernel function matrix for a scale parameter of 0.1;
FIG. 3 is a corresponding Gaussian kernel function matrix for a scale parameter of 5.1;
FIG. 4 is a corresponding Gaussian kernel function matrix for a scale parameter of 95.1;
FIG. 5 is a diagram showing the variation of the initial suitable values of scale parameters and classification accuracy.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, characteristics and effects of the method for intelligently processing the acquired data for the BIM model according to the present invention in combination with the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the collected data intelligent processing method for a BIM model provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of a method for intelligently processing collected data for a BIM model according to an embodiment of the present invention is shown, and the method includes the following steps:
s001, building data of different building types for constructing the BIM building model are collected, and a data set of the SVM support vector machine is obtained.
The aim of the embodiment is to realize the optimization of the BIM building model by analyzing the building data for constructing the BIM building model to screen out abnormal data in the building data, and avoiding the interference of the abnormal data on the construction of the BIM building model.
When analyzing building data, because of large difference among building data of different building types, in order to screen out abnormal data, the embodiment selects to conduct abnormal analysis on the building data of the same building type, and then the abnormal screening process of the building data of the same building type is commonly applied to the building data of all building types, so that the generalization capability requirement on the SVM support vector machine is reduced, the accuracy of the building data for constructing the BIM building model is improved, and the constructed BIM building model is more accurate. Building data for different building types refers to data materials for different building types in the building process, such as data for different building types of tiles, glass, etc.
In this embodiment, building data for different building types is collected, each building type of building data comprising a plurality of data dimensions, such as length, width, thickness, price, brand, etc. of glass. The category data in the building data of each building type is subjected to digital processing, for example, brands of glass are category data, and different brands can be numbered to realize the digitization.
When the building data of the same building type are subjected to anomaly analysis, the building data of the same building type have different specification requirements, and because the building volume of the building model is large, the building data of the same specification are often not minority, so when the building data of a certain specification is very little, the building data of the specification are often abnormal data, for example, in the building data of glass, the length, width, thickness and price of all the glasses corresponding to the brand A are the same, the length, width, thickness and price of all the glasses corresponding to the brand B are the same, all the glasses corresponding to the brand A are the same, the length, width, thickness and price of all the glasses corresponding to the brand B are different, and the glass corresponding to the brand A and the glass corresponding to the brand B are different specifications. The length, width, thickness and price of most of the glasses corresponding to the brand C are the same, the length, width, thickness and price of the glasses corresponding to the brand C are different, and the building data corresponding to the glasses with different lengths, widths, thicknesses and prices of the glasses corresponding to the brand C are abnormal data.
The building data with the same specification can be compared with the standard specification data to judge whether the building data with the current specification is qualified or not, but if the building data are screened according to the standard specification data, errors occur between the finally constructed BIM building model and an actual building body.
In the present embodiment, any one building type is taken as a target building type, and the target building type is acquiredBar history building data, wherein->Is preset toIn the present embodiment, n=200 is taken as an example, and the data amount is not limited, and the practitioner can set +_according to the actual implementation>Is a value of (2).
For the type of building of interestMarking the pieces of historical building data by adopting a manual marking method, marking the historical building data belonging to abnormal data, namely the abnormal historical building data as-1, marking the historical building data belonging to normal data, namely the normal historical building data as 1, and obtaining +_part>Tags for the historical building data. The +.>The bar history building data and the corresponding labels are used as a data set of the SVM support vector machine.
Thus, the collection of building data of different building types is realized, and the data set of the SVM support vector machine is obtained.
S002, acquiring an initial proper value of the scale parameter according to the data set of the SVM support vector machine.
When the SVM support vector machine is used for super-plane construction, a Gaussian kernel function is adopted to have better local anomaly detection capability compared with other kernel functions, when the building data for constructing the BIM building model is subjected to anomaly detection, the anomaly data in the building data are few data, and a better few anomaly data detection effects can be achieved by selecting the Gaussian kernel function, so that the super-plane construction is performed by adopting the Gaussian kernel function.
When the SVM support vector machine is used for carrying out abnormal classification, the interval value between the support vector and the hyperplane is maximized according to the Gaussian kernel function matrix, so that the hyperplane is constructed. When the hyperplane is constructed by utilizing the Gaussian kernel function, the scale parameters of the Gaussian kernel function often determine the classification effect of the constructed hyperplane, and different Gaussian kernel function matrixes can be constructed by the scale parameters of different sizes of the Gaussian kernel function, so that the hyperplane finally solved by the convex optimization method has flat or local sharp characteristics. The larger scale parameters can make the local of the hyperplane constructed by the Gaussian kernel function sharper, so that the hyperplane model is more suitable for data with higher complexity, is favorable for accurately distinguishing the nuances between normal data and abnormal data, but is easy to cause over-fitting and has poorer generalization capability; in contrast, the smaller scale parameters can enable the hyperplane constructed by the Gaussian kernel function to be flatter, so that the influence range of the hyperplane model is wider, the hyperplane model has higher generalization capability, but excessive smoothing of the hyperplane model can be caused, details and local features in a data set can not be captured well, and under-fitting is easy to occur particularly under the conditions of larger data quantity and more complex distribution.
In the existing SVM support vector machine, a plurality of initial scale parameters of a Gaussian kernel function are set, and the optimal scale parameters of the Gaussian kernel function are selected by using a cross verification method. However, since the distribution state of the historical building data is unknown, there may be more redundant scale parameter values after the initial scale parameter setting, and meanwhile, since the data size of the historical building data is relatively large, the redundant scale parameter values may cause waste of calculation resources. Therefore, the embodiment realizes the preliminary screening of the scale parameters by analyzing the Gaussian kernel function matrix and combining the distribution characteristics of the historical building data of the target building type. And then cross verification is carried out according to the scale parameters after preliminary screening, and better scale parameters are selected according to the historical building data misclassified in the cross verification, so that the classification accuracy of the SVM support vector machine on the building data is improved, and further the BIM building model constructed according to the classified building data is more accurate.
In this embodiment, a scale interval is presetAnd a step +.>In this embodiment->=0.1、/>=100、/>The embodiment of =5 is described as an example, and the embodiment is not limited to this, and the implementation personnel can be set according to the specific situation. Within the scale interval, starting with the left boundary of the scale interval, every interval +. >And acquiring a numerical value, and taking the acquired numerical values as a scale parameter in a scale interval respectively. Thus obtaining a plurality of scale parameters in a scale interval: 0.1, 5.1, 10.1, …, 95.1, the number of the obtained scale parameters was noted as N.
Firstly, constructing Gaussian kernel function matrixes with different scale parameters, and screening out scale parameters which are not suitable for building data classification of a target building type through preliminary analysis of the Gaussian kernel function matrixes, wherein the method comprises the following specific steps:
for each scale parameter, extracting from the dataset of the SVM support vector machineIs used as training set, the remaining +.>As a validation set for cross-validation of the current scale parameter corresponding hyperplane model. Although the data set is the historical building data of the same building type, because the historical building data of the same building type have different specifications, when the training set and the verification set are extracted, the normal historical building data and the abnormal historical building data of different specifications are randomly extracted, so that the SVM support vector machine has decision boundaries for the building data of different specifications when the hyperplane model is constructed.
Substituting each scale parameter into a Gaussian kernel function, and taking a training set corresponding to each scale parameter as input of a corresponding Gaussian kernel function to obtain a Gaussian kernel function matrix corresponding to each scale parameter. Fig. 2, 3 and 4 are gaussian kernel function matrices corresponding to scale parameters of 0.1, 5.1 and 95.1, respectively.
It should be noted that, the function of the kernel function is to calculate the kernel function value between two data in the original space, so as to represent the inner product of the two data in the high-dimensional space, so that the calculation amount can be reduced, and the hyperplane model with the maximum interval value corresponding to the support vector can be solved by performing quadratic programming optimization on the vector in the gaussian kernel function matrix, wherein the support vector is the vector closest to the decision boundary (i.e. hyperplane).
If a certain scale parameter is suitable for classifying the historical building data of the target building type, in the Gaussian kernel function matrix corresponding to the scale parameter, the historical building data of different specifications should have larger difference values, and the historical building data of the same specification should be more approximate.
In this embodiment, any one scale parameter is used as a target scale parameter, and a training set of the target scale parameter is obtained, and the firstHistorical building data and division +.>The kernel function value between each of the history building data other than the history building data constitutes +.>Nuclear function value sequence of individual history building data +.>. First->Nuclear function value sequence of individual history building data +.>Is the number of historical building data minus one in the training set for the target scale parameter.
In the first placeNuclear function value sequence of individual history building data +.>In (1)/(2)>The kernel function value corresponding to the historical building data with the same specification of the historical building data should be larger, and the maximum is 1. If a certain historical building data and +>The more similar, if the values of the historical building data in the gaussian kernel function matrix are, the historical building data are +.>The higher the probability that the individual historical building data will be subsequently classified into the same class. In->Nuclear function value sequence of individual history building data +.>In, with the firstThe kernel function value corresponding to the historical building data of which the historical building data does not belong to the same specification should be small, and at the same time, the kernel function value is the same as the +.>The kernel function value corresponding to the abnormal historical building data of which the historical building data belongs to the same specification should also be small.
And similarly, acquiring a kernel function value sequence of each historical building data in the training set of the target scale parameters.
Although the kernel function between the historical building data of different specifications should have a lower value, a larger kernel function value still exists in the gaussian kernel function matrix due to the presence of the historical building data of the same specification. In order to reduce the false detection probability of abnormal data, the abnormal data and the rest of the historical building data should have lower kernel function values in the Gaussian kernel function matrix.
Obtaining initial suitable values of the target scale parameters for each historical building data in the training set:
wherein ,representing the target scale parameter +.>Initial fit values for the individual historical building data;training set representing target scale parameter +.>Historical building data; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, and +.>The historical building data is the minimum value of the kernel function values of all the normal historical building data with the same specification, wherein the minimum value is selected because the minimum value determines the lower limit of the kernel function value between the normal historical building data with the same specification so as to prevent partial positiveOften historical building data is misclassified; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, and +.>The maximum value is selected because the maximum value determines the upper limit of the kernel function value between the normal historical building data with different specifications so as to prevent the normal historical building data with different specifications from being mistakenly classified into one type, so that the generalization capability of the hyperplane model corresponding to the normal historical building data with different specifications in classification is poor, and 1 minus × is used for the generalization capability >Performing negative correlation mapping; />Training set representing target scale parameter +.>In the sequence of kernel function values of the historical building data, the maximum value of the kernel function values corresponding to all abnormal historical building data is selected because the maximum value determines +.>The upper limit of the kernel function value between the historical building data and all the abnormal historical building data is used for preventing the normal historical building data and the abnormal historical building data from being mistakenly classified into one type, so that the classification effect of the corresponding hyperplane model is poor in classification, the BIM building model established in the BIM modeling model modeling is inaccurate, and the 1 minus->Performing negative correlation mapping; />Training set of target scale parametersA tag of historical building data, when the tag is 1 +.>The history building data is normal history building data, and +.>Dividing by 3 to obtain the average value +.>The initial proper value of the historical building data is larger, and the classification effect is better for the mth historical building data under the target scale parameter; when the tag is-1, < th->The historical building data is abnormal historical building data, and the data of the same specification is not required to be searched for, namely +. >Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, the minimum value of the kernel function values corresponding to all abnormal historical building data is the +.>The historical building data are abnormal historical building data, the abnormal historical building data and the abnormal historical building data should be separated together, and the minimum value is selected because the minimum value determines the lower limit of the kernel function value between the abnormal historical building data so as to prevent part of abnormal historical buildingThe data is misclassified; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, the maximum value of the kernel function values corresponding to all the normal historical building data is selected at the moment because the maximum value represents the upper limit of the kernel function value between the abnormal historical building data and the normal historical building data so as to prevent part of the normal historical building data from being misclassified; when->When the history building data is abnormal history building data, the +.>Dividing by 2 to obtain the average value +.>The larger the initial suitable value is, the better the classifying effect on the m-th historical building data is under the target scale parameter.
So far, the initial proper value of the target scale parameter for each historical building data in the training set is obtained, and the average value of the initial proper values of the target scale parameter for all the historical building data in the training set is used as the initial proper value of the target scale parameter.
Similarly, an initial fit value for each scale parameter within the scale interval is obtained.
S003, obtaining the optimal scale parameters according to the initial proper value of each scale parameter.
Performing preliminary screening operation according to the initial proper value of each scale parameter in the scale interval to obtain the preferred scale parameter in the scale interval, wherein the preferred scale parameter is specifically as follows:
after obtaining the initial suitable values corresponding to the scale parameters, a suitable value threshold Sr is preset, in this embodiment, sr=0.5 is described as an example, and the value of Sr may be set by an operator according to the specific implementation situation without specific limitation. When the initial proper value of the scale parameter is smaller than the proper value threshold Sr, the scale parameter is eliminated, and the cross verification is not performed by using the scale parameter. When the initial suitable value of the scale parameter is greater than or equal to the suitable value threshold Sr, the scale parameter is retained as the preferred scale parameter for subsequent cross-validation. After the primary screening of the scale parameters, the redundancy of the scale parameters can be effectively reduced, and the efficiency of obtaining the optimal scale parameters through cross verification is improved.
And performing cross verification on all the preferred scale parameters to obtain the classification accuracy of each preferred scale parameter. It should be noted that, along with the change of the scale parameters, the classification accuracy rate in each verification process changes, and in the existing process of obtaining the optimal scale parameters through cross verification, the scale parameter value corresponding to the highest accuracy rate is selected from all verification results to be used as the optimal scale parameter, but the optimal scale parameter obtained at this time is the optimal in the initial scale parameters set by people, and is not the optimal for actually classifying the historical building data of the target building type.
In this embodiment, if there is a classification accuracy of 100% for the preferred scale parameter, the preferred scale parameter with a classification accuracy of 100% is taken as the optimal scale parameter. If the classification accuracy of the optimal scale parameters is 100%, the conventional cross-validation method cannot obtain the truly optimal scale parameters, so that effective screening of abnormal data in building data cannot be performed, and the subsequently constructed BIM building model is inaccurate.
When the classification accuracy of the optimal scale parameter is 100%, the interval approximation is required to be continuously performed by combining the initial proper value of the optimal scale parameter to obtain the optimal scale parameter as far as possible. It should be noted that, if the initial suitable value of the preferred scale parameter is larger, the classification effect of classifying the historical building data of the target building type by using the gaussian kernel function matrix corresponding to the current preferred scale parameter to construct the hyperplane model is more suitable, if the classification accuracy of the current preferred scale parameter is higher in the cross verification, the classification effect of classifying the historical building data of the target building type by using the gaussian kernel function matrix corresponding to the current preferred scale parameter to construct the hyperplane model is better. Therefore, the initial proper value of each preferred scale parameter and the classification accuracy should have a positive correlation, and in the case of consistent classification accuracy, a preferred scale parameter with a higher initial proper value should be adopted as a preferred scale parameter. FIG. 5 is a diagram showing the variation of the initial suitable values of scale parameters and classification accuracy.
In this embodiment, the obtaining operation of the preferred interval is performed according to the initial suitable values and the classification accuracy of all the preferred scale parameters in the scale interval, and the specific process is as follows:
and acquiring pearson correlation coefficients between the initial proper values of all the preferred scale parameters in the scale interval and the classification accuracy, and obtaining a correlation value Z between the initial proper values and the classification accuracy. After the screening of the proper value threshold Sr, the initial proper value is obtained as an effective value, and the obtained correlation is more reliable according to the optimal scale parameters obtained after the screening. If the correlation value Z between the initial proper value and the classification accuracy is larger, when the left and right nearest scale parameters are selected to carry out secondary interval division by taking the optimal scale parameter corresponding to the maximum classification accuracy as the center, the optimal selection can be carried out more depending on the initial proper value.
And taking the optimal scale parameter corresponding to the maximum classification accuracy in the scale interval as the center parameter of the scale interval, taking the optimal scale parameter which is smaller than the center parameter and has the smallest difference value with the center parameter as the left boundary of the first interval, and taking the center parameter as the right boundary of the first interval to obtain the first interval of the scale interval. And taking the optimal scale parameter which is larger than the central parameter and has the smallest difference value with the central parameter as the right boundary of the second interval, and taking the central parameter as the left boundary of the second interval to obtain the second interval of the scale interval. Taking the average value of the classification accuracy rates corresponding to the left boundary and the right boundary of the first interval as the classification accuracy rate of the first interval, and taking the average value of the classification accuracy rates corresponding to the left boundary and the right boundary of the second interval as the classification accuracy rate of the second interval.
It should be noted that, the selection of the more optimal section is performed in the first section and the second section of the scale section to improve the classification accuracy of the historical building data, and generally, the section corresponding to the maximum value of the classification accuracy is taken as the more optimal section according to the classification accuracy of each section, because the scale parameter with better classification effect on the historical building data of the target building type is more likely to exist in the section corresponding to the maximum value. However, the selection of the more optimal section is performed according to the maximum value of the classification accuracy of the first section and the classification accuracy of the second section, so that the analysis on pure data is too absolute, the classification model of the historical building data of the target building type is not necessarily applicable, and the initial proper value is related to the distribution of the historical building data of the target building type, so that new scale parameters can be acquired in the first section and the second section, and the more optimal section can be selected in combination with the initial proper value of the new scale parameters.
In the present embodiment, the step size is adjustedUpdate is performed so that->. Within the first interval, starting with the left boundary of the first interval,/every interval +.>And acquiring a numerical value, and taking each acquired numerical value as a scale parameter in the first interval. And similarly, acquiring all scale parameters in the second interval.
And (2) acquiring initial proper values of each scale parameter in the first interval and the second interval by using the method in the step (S002), taking the average value of the initial proper values of all the scale parameters in the first interval as the average initial proper value of the first interval, and taking the average value of the initial proper values of all the scale parameters in the second interval as the average initial proper value of the second interval.
Obtaining the preference degree of each interval according to the average initial suitable value and the classification accuracy of each interval:
wherein ,、/>preference degree of the first section and the second section respectively, < >>、/>Average initial fit values of the first section and the second section respectively, +.>For the correlation value between the initial fit value and the classification accuracy>、/>The classification accuracy rates of the first interval and the second interval are respectively.
And taking the section with the highest preference degree of the first section and the second section as a more preferred section. The data distribution in the Gaussian kernel function matrix corresponding to the scale parameter in the more optimal interval is more in accordance with the expected classification effect of the historical building data of the target building type, so that the optimal scale parameter is further searched in the more optimal interval.
Performing a better interval update operation, including: performing preliminary screening operation according to the initial suitable value of each scale parameter in the better interval to obtain the preferred scale parameter in the better interval, performing cross-validation on all the preferred scale parameters in the better interval to obtain the classification accuracy of each preferred scale parameter in the better interval, and performing new acquisition operation of the better interval according to the initial suitable value and the classification accuracy of all the preferred scale parameters in the better interval.
And repeating the updating operation of the more optimal section until the scale parameters with higher classification accuracy than the classification accuracy of the central parameters of the latest more optimal section do not exist in the first section and the second section of the latest more optimal section, and stopping iteration. Taking the center parameter of the latest more optimal interval as the optimal scale parameter.
So far, the optimal scale parameters are obtained.
S004, constructing an optimal hyperplane model according to the optimal scale parameters.
Substituting the optimal scale parameters into the corresponding Gaussian kernel functions to obtain an optimal Gaussian kernel function matrix of the scale parameters, inputting the optimal Gaussian kernel function matrix of the scale parameters into an SVM support vector machine, and performing hyperplane construction to obtain an optimal SVM hyperplane model of the target building type.
Similarly, the optimal scale parameters of each building type are obtained according to the historical building data of each building type, and an optimal SVM hyperplane model of each building type is constructed.
S005, intelligent processing of building data is achieved according to the optimal hyperplane model, and a BIM building model is built.
And inputting the building data of each building type which is currently collected into an optimal SVM hyperplane model of each building type to obtain a corresponding classification result. When the classification result of the building data is abnormal data, the related personnel re-acquire and check the building data of the corresponding building type, and when the classification result of the building data is normal data, the building data is input into software for constructing the BIM building model to perform BIM modeling, so as to obtain the BIM building model.
Through the above steps, intelligent processing of building data for BIM modeling is achieved.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. The intelligent processing method for the acquired data of the BIM model is characterized by comprising the following steps of:
collecting building data of different building types; taking any building type as a target building type, and constructing a data set of an SVM support vector machine according to historical building data of the target building type;
presetting a scale interval and acquiring all scale parameters in the scale interval; acquiring a training set of each scale parameter according to a data set of the SVM support vector machine, and acquiring a Gaussian kernel function matrix corresponding to each scale parameter according to the training set of each scale parameter; acquiring an initial proper value of each scale parameter according to the Gaussian kernel function matrix corresponding to each scale parameter;
acquiring a preferred scale parameter according to the initial proper value of each scale parameter; cross-verifying all the preferred scale parameters to obtain the classification accuracy of each preferred scale parameter;
Acquiring a more optimal section according to the initial proper values and the classification accuracy of all the preferred scale parameters in the scale section, and acquiring the optimal scale parameters of the target building type according to the more optimal section;
and constructing an optimal SVM hyperplane model of the target building type according to the optimal scale parameters, classifying building data according to the optimal SVM hyperplane model, and constructing a BIM building model according to classification processing results.
2. The intelligent processing method for collected data of a BIM model according to claim 1, wherein the constructing a data set of an SVM support vector machine according to the historical building data of the target building type includes the following specific steps:
acquiring a plurality of historical building data of a target building type, marking all the historical building data by adopting a manual marking method, marking abnormal historical building data as-1, marking normal historical building data as 1, and obtaining a label of each historical building data; and taking all the historical building data of the target building type and the corresponding labels as a data set of the SVM support vector machine.
3. The method for intelligently processing the acquired data of the BIM model according to claim 1, wherein the step of acquiring the gaussian kernel function matrix corresponding to each scale parameter according to the training set of each scale parameter includes the following specific steps:
Substituting each scale parameter into a Gaussian kernel function, and taking a training set corresponding to each scale parameter as input of a corresponding Gaussian kernel function to obtain a Gaussian kernel function matrix corresponding to each scale parameter.
4. The method for intelligently processing the acquired data for the BIM model according to claim 1, wherein the step of obtaining the initial suitable value of each scale parameter according to the gaussian kernel function matrix corresponding to each scale parameter includes the following specific steps:
taking any one scale parameter as a target scale parameter, and acquiring a kernel function value sequence of each historical building data in a training set of the target scale parameter; acquiring an initial proper value of the target scale parameter for each historical building data in the training set according to the kernel function value sequence of each historical building data in the training set of the target scale parameter; and taking the average value of the initial proper values of the target scale parameters for all the historical building data in the training set as the initial proper value of the target scale parameters.
5. The intelligent processing method for collected data of BIM model according to claim 4, wherein the step of obtaining the kernel function value sequence of each history building data in the training set of target scale parameters includes the following specific steps:
Acquiring training set of target scale parameters, the firstHistorical building data and division +.>The kernel function value between each of the history building data other than the history building data constitutes +.>A sequence of kernel function values for the historical building data.
6. The method for intelligently processing collected data for a BIM model according to claim 4, wherein the step of obtaining the initial suitable value of the target scale parameter for each of the historical building data in the training set includes the following specific steps:
wherein ,representing the target scale parameter +.>Initial fit values for the individual historical building data; />Training set representing target scale parameter +.>Historical building data; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, and +.>The historical building data are the minimum value of the kernel function values of all normal historical building data with the same specification; />Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, and +.>The historical building data are the maximum value of the kernel function values of all normal historical building data with different specifications; / >Training set representing target scale parameter +.>In the kernel function value sequence of the historical building data, the maximum value of the kernel function values corresponding to all abnormal historical building data; />Training set representing target scale parameter +.>Tags for the historical building data; />Training set representing target scale parameter +.>History of constructionIn the kernel function value sequence of the building data, the minimum value in the kernel function values corresponding to all abnormal historical building data; />Training set representing target scale parameter +.>And in the kernel function value sequence of each history building data, the maximum value in the kernel function values corresponding to all normal history building data.
7. The method for intelligently processing the acquired data for the BIM model according to claim 1, wherein the acquiring the better interval according to the initial suitable values and the classification accuracy of all the preferred scale parameters in the scale interval includes the following specific steps:
the method comprises the steps of obtaining pearson correlation coefficients between initial suitable values and classification accuracy of all preferred scale parameters in a scale interval, and taking the pearson correlation coefficients as correlation values between the initial suitable values and the classification accuracy; taking the preferred scale parameter corresponding to the maximum classification accuracy in the scale interval as the center parameter of the scale interval; acquiring a first interval and a second interval of the scale interval and classification accuracy of the first interval and the second interval according to the central parameter;
Acquiring all scale parameters in a first interval and a second interval, and acquiring initial proper values of each scale parameter in the first interval and the second interval; taking the average value of the initial proper values of all the scale parameters in the first interval as the average initial proper value of the first interval, and taking the average value of the initial proper values of all the scale parameters in the second interval as the average initial proper value of the second interval;
acquiring the preference of the first interval and the second interval according to the average initial fit value and the classification accuracy of the first interval and the second interval; and taking the section with the highest preference degree of the first section and the second section as a more preferred section.
8. The intelligent processing method for the acquired data of the BIM model according to claim 7, wherein the acquiring the first interval and the second interval of the scale interval and the classification accuracy of the first interval and the second interval according to the central parameter includes the specific steps of:
taking the optimal scale parameter which is smaller than the central parameter and has the smallest difference value with the central parameter as the left boundary of the first interval, and taking the central parameter as the right boundary of the first interval to obtain the first interval of the scale interval; taking the optimal scale parameter which is larger than the central parameter and has the smallest difference value with the central parameter as the right boundary of the second interval, and taking the central parameter as the left boundary of the second interval to obtain the second interval of the scale interval; taking the average value of the classification accuracy rates corresponding to the left boundary and the right boundary of the first interval as the classification accuracy rate of the first interval, and taking the average value of the classification accuracy rates corresponding to the left boundary and the right boundary of the second interval as the classification accuracy rate of the second interval.
9. The method for intelligently processing the acquired data for the BIM model according to claim 7, wherein the acquiring the preference of the first section and the second section according to the average initial suitable value and the classification accuracy of the first section and the second section includes the following specific steps:
wherein , and />Preference degree of the first section and the second section, respectively, < >> and />Average initial fit value of the first interval and the second interval, respectively,/respectively>For the correlation value between the initial fit value and the classification accuracy> and />The classification accuracy rates of the first interval and the second interval are respectively.
10. The intelligent processing method for the acquired data of the BIM model according to claim 7, wherein the acquiring the optimal scale parameter of the target building type according to the better interval includes the following specific steps:
performing a better interval update operation, including: acquiring the preferred scale parameters in the preferred interval according to the initial suitable value of each scale parameter in the preferred interval, performing cross verification on all the preferred scale parameters in the preferred interval to obtain the classification accuracy of each preferred scale parameter in the preferred interval, and acquiring a new preferred interval according to the initial suitable value and the classification accuracy of all the preferred scale parameters in the preferred interval;
Repeating the updating operation of the more optimal section until the obtained first section and the obtained second section of the latest more optimal section have no scale parameter with higher classification accuracy than the classification accuracy of the central parameter of the latest more optimal section, and stopping iteration; taking the center parameter of the latest more optimal interval as the optimal scale parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310996803.6A CN116738551B (en) | 2023-08-09 | 2023-08-09 | Intelligent processing method for acquired data of BIM model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310996803.6A CN116738551B (en) | 2023-08-09 | 2023-08-09 | Intelligent processing method for acquired data of BIM model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116738551A true CN116738551A (en) | 2023-09-12 |
| CN116738551B CN116738551B (en) | 2023-10-17 |
Family
ID=87917153
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310996803.6A Active CN116738551B (en) | 2023-08-09 | 2023-08-09 | Intelligent processing method for acquired data of BIM model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116738551B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117743955A (en) * | 2023-12-21 | 2024-03-22 | 广东人信工程咨询有限公司 | BIM (building information modeling) acquired data processing method, system, electronic equipment and storage medium |
| CN118298075A (en) * | 2024-04-16 | 2024-07-05 | 内蒙古尼采时代智能科技有限公司 | Building appearance effect graph generation method, equipment and medium based on text description |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106651865A (en) * | 2016-12-23 | 2017-05-10 | 湖北工业大学 | Novel automatic selection method of optimal segmentation scale of high-resolution remote sensing image |
| KR101741761B1 (en) * | 2015-12-04 | 2017-05-30 | 광운대학교 산학협력단 | A classification method of feature points required for multi-frame based building recognition |
| CN107563324A (en) * | 2017-08-30 | 2018-01-09 | 广东工业大学 | A kind of hyperspectral image classification method and device of the learning machine that transfinited based on core basis |
| CN108985327A (en) * | 2018-06-05 | 2018-12-11 | 东南大学 | A kind of Terrain Matching self-organizing Optimum Classification method based on factorial analysis |
| KR20220151777A (en) * | 2021-05-07 | 2022-11-15 | 연세대학교 산학협력단 | Method and device for classifying building defect using multi task channel attention |
| CN115655717A (en) * | 2022-10-21 | 2023-01-31 | 厦门大学 | Bearing fault diagnosis method based on depth domain generalization network |
-
2023
- 2023-08-09 CN CN202310996803.6A patent/CN116738551B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101741761B1 (en) * | 2015-12-04 | 2017-05-30 | 광운대학교 산학협력단 | A classification method of feature points required for multi-frame based building recognition |
| CN106651865A (en) * | 2016-12-23 | 2017-05-10 | 湖北工业大学 | Novel automatic selection method of optimal segmentation scale of high-resolution remote sensing image |
| CN107563324A (en) * | 2017-08-30 | 2018-01-09 | 广东工业大学 | A kind of hyperspectral image classification method and device of the learning machine that transfinited based on core basis |
| CN108985327A (en) * | 2018-06-05 | 2018-12-11 | 东南大学 | A kind of Terrain Matching self-organizing Optimum Classification method based on factorial analysis |
| KR20220151777A (en) * | 2021-05-07 | 2022-11-15 | 연세대학교 산학협력단 | Method and device for classifying building defect using multi task channel attention |
| CN115655717A (en) * | 2022-10-21 | 2023-01-31 | 厦门大学 | Bearing fault diagnosis method based on depth domain generalization network |
Non-Patent Citations (2)
| Title |
|---|
| GUOCHENG NIU ET AL: "Enhancing Local Feature Extraction with Global Representation for Neural Text Classification", 《PROCEEDINGS OF THE 2019 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING》 * |
| 金超逸，徐旻洋，周向东: "基于建筑信息模型(BIM)的建筑空间功能分类方法", 《计算机应用与软件》, pages 220 - 225 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117743955A (en) * | 2023-12-21 | 2024-03-22 | 广东人信工程咨询有限公司 | BIM (building information modeling) acquired data processing method, system, electronic equipment and storage medium |
| CN118298075A (en) * | 2024-04-16 | 2024-07-05 | 内蒙古尼采时代智能科技有限公司 | Building appearance effect graph generation method, equipment and medium based on text description |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116738551B (en) | 2023-10-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116738551B (en) | Intelligent processing method for acquired data of BIM model | |
| CN111882446B (en) | Abnormal account detection method based on graph convolution network | |
| CN109523518B (en) | Tire X-ray defect detection method | |
| CN111461068B (en) | Chromosome metaphase map identification and segmentation method | |
| CN113222149B (en) | Model training method, device, equipment and storage medium | |
| CN112037222B (en) | Automatic updating method and system of neural network model | |
| CN113591866B (en) | Special operation certificate detection method and system based on DB and CRNN | |
| CN111860596A (en) | Unsupervised pavement crack classification method based on deep learning and model establishment method | |
| CN111796957A (en) | Transaction abnormal root cause analysis method and system based on application log | |
| CN109656808A (en) | A kind of Software Defects Predict Methods based on hybrid active learning strategies | |
| CN113673534B (en) | RGB-D image fruit detection method based on FASTER RCNN | |
| CN108428234B (en) | Interactive segmentation performance optimization method based on image segmentation result evaluation | |
| CN117015812A (en) | System for clustering data points | |
| CN103093239B (en) | A kind of merged point to neighborhood information build drawing method | |
| CN107729942A (en) | A kind of sorting technique of structured view missing data | |
| CN117078007A (en) | Multi-scale wind control system integrating scale labels and method thereof | |
| CN115296851B (en) | Network intrusion detection method based on mutual information and wolf lifting algorithm | |
| CN115809687A (en) | Training method and device for image processing network | |
| CN113177602A (en) | Image classification method and device, electronic equipment and storage medium | |
| CN109948421B (en) | Hyperspectral image classification method based on PCA and attribute configuration file | |
| CN113313207A (en) | Sample unbalanced target detection method of class truth value constraint classification function | |
| CN112801028A (en) | Spectrum and space map hyperspectral image classification method based on inductive representation learning network | |
| CN117312395B (en) | Query system optimization method, device and equipment based on big data big model | |
| CN114708457B (en) | Hyperspectral deep learning identification method for anti-purple fringing identification | |
| CN114708457A (en) | Hyperspectral deep learning identification method for purple fringing resistance identification |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |