CN109559045A - A kind of method and system of personnel's intelligence control - Google Patents
A kind of method and system of personnel's intelligence control Download PDFInfo
- Publication number
- CN109559045A CN109559045A CN201811452187.3A CN201811452187A CN109559045A CN 109559045 A CN109559045 A CN 109559045A CN 201811452187 A CN201811452187 A CN 201811452187A CN 109559045 A CN109559045 A CN 109559045A
- Authority
- CN
- China
- Prior art keywords
- sample
- target object
- cluster
- model
- cluster heart
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Abstract
The invention discloses a kind of method and system of personnel's intelligence control, its method assesses the real-time behavior of target object using dynamic integral model the following steps are included: the dynamic integral model that the truthful data building based on the target object obtained from multiple operation systems is studied and judged and predicted for the behavior to target object；Individual information based on target object assesses target object using classical integral model；The assessed value of assessed value and the target object obtained using classical integral model to the target object behavior obtained using dynamic integral model is weighted summation to obtain the comprehensive assessment value of the real-time behavior of target object；When comprehensive assessment value is more than early warning threshold value, disposition instruction is generated, to carry out control disposition to target object.Accurate, effective and timely control to target object has can be achieved in the present invention.
Description
Technical field
The invention belongs to human behavior pattern analysis technical field, in particular to the method for a kind of personnel's intelligence control and
System.
Background technique
Emphasis personnel control is the important content of information in public security organs work, to relate to probably, relate to the steady and eight major class emphasis people such as be involved in drug traffic
Member carries out scientific control, diminishes crime in prevention, hits illegal activities and maintain social stability and play a significant role.But with society
Meeting expanding economy, movement of population quantity and range increased dramatically, and " registered permanent residence sky extension " " separation of families and registered permanent residence " phenomenon is increasing, tradition
Emphasis personnel's pipe diameter design be difficult to adapt to real needs.And the fast development of the technologies such as internet, big data and artificial intelligence,
Opportunity is brought for the change and innovation of emphasis personnel control.
For this purpose, being instructed under objective in " for information prior to action, information is higher than action, information guidance action ", public security organ
Actively push forward police service practical strategy implementation, the big data of each level has been built in various regions, each department's fitting actual needs in succession
Center and seriation information platform are carried out big data based on " personnel's integral model " and studied and judged and intelligent tool application, right
The control ability and helping effect of emphasis personnel has larger promotion.But existing emphasis personnel managing and control system is not from basic
Problem and control problem are studied and judged present in upper solution real work, so that " column are but regardless of, pipe without controlling " phenomenon still protrudes.
Firstly, existing emphasis personnel managing and control system is mostly from data mapping or single means to the rail of emphasis personnel
Mark carries out analysis and studies and judges, and is difficult accurately to grasp the whereabouts of emphasis personnel, and find its abnormal behaviour in time.Furthermore due to lacking
Feasible data model and data tool, even if having no way in the case where having accumulated massive information relevant to emphasis personnel
Multidimensional is carried out using above-mentioned data to see clearly and association analysis, is studied and judged ineffective.Particular, it is important that being managed as emphasis personnel
Core, existing early warning rule set based on traditional integral model, depends particularly on staff and has business warp
Test, cannot because ground, because when quickly carry out dynamic adjustment.
Summary of the invention
The first technical problem to be solved by the present invention is a kind of method for proposing personnel's intelligence control, based on dynamic product
Sub-model is studied and judged, and to make full use of existing business experience, solves business experience problem of aging, promotion, which is sentenced, grinds effect.
In order to solve the above-mentioned technical problem, embodiments herein provides firstly a kind of side of personnel's intelligence control
Method, comprising the following steps:
Based on the target object obtained from multiple operation systems truthful data building for the behavior to target object into
The dynamic integral model that row is studied and judged and predicted, assesses the real-time behavior of target object using the dynamic integral model；
Individual information based on target object assesses target object using classical integral model；
To the assessed value and utilization classical integral mould of the real-time behavior of target object that the utilization dynamic integral model obtains
The assessed value for the target object that type obtains is weighted summation, obtains the comprehensive assessment value of the real-time behavior of the target object；
When the comprehensive assessment value is more than early warning threshold value, disposition instruction is generated based on pre-defined rule, to target object
Carry out control disposition.
Preferably, the weighting coefficient of the weighted sum is configured according to the maturity of the dynamic integral model.
Preferably, the truthful data based on the target object obtained from multiple operation systems is constructed for target pair
The dynamic integral model that the behavior of elephant is studied and judged and predicted, comprising the following steps:
Arrangement is collected to the truthful data of the target object, obtains training sample set；
Similarity calculation and clustering processing are carried out to the sample data that the training sample is concentrated, obtain cluster heart sample；
Dimension-reduction treatment is carried out to the cluster heart sample, obtains the orthogonalization characteristic quantity and corresponding weighted value of cluster heart sample；
According to the orthogonalization characteristic quantity and corresponding weighted value of cluster heart sample, initial model is constructed；
The initial model is trained using the orthogonalization characteristic quantity of cluster heart sample, optimizes weighted value, thereby determines that
The dynamic integral model.
Preferably, using sigmoid function as kernel function, the initial model F (t) is constructed:
F (t)=a1y1(t)+a2y2(t)+...+amym(t)+a0
Wherein, y1(t) ..., ymIt (t) is orthogonalization characteristic quantity, a0, a1..., amFor corresponding weighted value.
Preferably, the orthogonalization characteristic quantity using cluster heart sample is trained the initial model, optimizes weight
Value, thereby determines that the dynamic integral model, specifically,
The data of either cluster heart sample in the cluster heart sample are inputted the initial model to calculate, the meter that will be obtained
Calculation value actual value corresponding with the data of the cluster heart sample is compared, and obtains the corresponding error of cluster heart sample；
All cluster heart samples are traversed, the corresponding error of each cluster heart sample is obtained；
It is for statistical analysis to the error of all cluster heart samples, weight is carried out more to the initial model based on analysis result
Newly, to obtain the dynamic integral model.
Preferably, the sample data concentrated to the training sample carries out similarity calculation and clustering processing, obtains
Cluster heart sample, specially
Step 1 concentrates one sample of any selection labeled as the cluster heart of cluster in the training sample；
Step 2 is concentrated from the training sample and chooses a sample in remaining sample as new addition sample；
Step 3 uses the range formula based on cosine angle to calculate the new addition sample and each label as the sample of the heart
This similarity,
If each similarity being calculated is both less than predetermined threshold, new cluster is established centered on the new addition sample；
Otherwise by the new addition sample be added to the highest cluster of its similarity, and recalculate the cluster heart of the cluster；
It repeats step 2 and to obtain several clusters, and is chosen each to step 3 until having handled the training sample concentrates sample
Sample corresponding to the cluster heart of cluster is as the cluster heart sample.
Preferably, dimension-reduction treatment is carried out to the cluster heart sample using principal component analytical method.
Preferably, further include, using the real-time behavior of the target object and corresponding disposition result as target object
New sample data is added training sample and concentrates, for advanced optimizing the dynamic integral model.
Preferably, the operation system is detectd platform, PGIS platform, informational intelligence summary and is put down including alert comprehensive platform, the comprehensive platform of net, skill
At least one of platform platform.
Embodiments herein additionally provides a kind of system of personnel's intelligence control, including computer-readable storage medium
Matter, wherein being stored with program, described program realizes such as any of the above-described personnel as described in the examples when being executed by processor
The method of intelligence control.
Compared with prior art, one or more embodiments in above scheme can have following advantage or beneficial to effect
Fruit:
The present invention, using mechanism such as intelligent monitoring, timely early warning and linkage disposition, is realized using dynamic integral model as core
Accurate, effective and timely control to target object avoids in existing emphasis personnel control " column but regardless of, pipe without controlling "
The phenomenon that, and then reduce the generation of related cases, effectively maintain the stable of society.In addition, may be used also using the present invention
To promote public security organ to the control working efficiency of emphasis personnel, the human input of related work is reduced.
Other advantages, target and feature of the invention will be illustrated in the following description to a certain extent, and
And to a certain extent, based on will be apparent to those skilled in the art to investigating hereafter, Huo Zheke
To be instructed from the practice of the present invention.Target and other advantages of the invention can be wanted by following specification, right
Specifically noted structure is sought in book and attached drawing to be achieved and obtained.
Detailed description of the invention
Attached drawing is used to provide to the technical solution of the application or further understanding for the prior art, and constitutes specification
A part.Wherein, the attached drawing for expressing the embodiment of the present application is used to explain the technical side of the application together with embodiments herein
Case, but do not constitute the limitation to technical scheme.
Fig. 1 is the control flow diagram of carry out personnel intelligence control according to an embodiment of the invention；
Fig. 2 is the foundation of dynamic integral model according to an embodiment of the invention and with process schematic.
Specific embodiment
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and examples, how to apply to the present invention whereby
Technological means solves technical problem, and the realization process for reaching relevant art effect can fully understand and implement.This Shen
Please each feature in embodiment and embodiment, can be combined with each other under the premise of not colliding, be formed by technical solution
It is within the scope of the present invention.
To realize accurate, quick, the efficient control to emphasis personnel, need to reform control theory, more rails obtain related letter
Breath, carries forward vigorously resource-sharing with a line beading；On this basis, emphasis of the present invention is managed actual combatization around emphasis personnel and is needed,
Intelligent control process of the design based on big data, completely covers the key links such as data convergence, integration, analysis and application；Most
It is essential that the present invention, which constructs an actual combatization, studies and judges model, existing business experience is made full use of, and pass through data training
The optimization of automatic study correlation criterion implementation model and parameter solves business experience problem of aging.
With a specific embodiment, the invention will be further described below.
The control process of personnel's intelligence involved in the present invention is summarized first.As shown in Figure 1, being people
The flow diagram of the intelligent control of member.
The control process is needed in conjunction with current public business, is driving with data, using multidimensional data convergence analysis as core
It is designed, specifically includes data convergence, data preparation, data are seen clearly, early warning is disposed and dispose the multiple links of feedback.
Wherein, information convergence is the premise that system data is studied and judged and applied.Information converges in link, realizes comprehensive flat with police
Platform, the comprehensive platform of net, skill are detectd platform, PGIS platform, all kinds of business of informational intelligence summary platform and video image Ku Deng public security organ and are put down
The docking of number of units evidence (public security internal data), and realized by social public information shared switching plane (social external data)
The docking of the units operation systems such as operator, civil aviaton, railway, hotel, the industrial and commercial tax, and website, forum and blog are acquired in real time
Etc. the network informations.
Data preparation link, be responsible for people's (object) of remittance, vehicle (), electric (fence), net (network), as (figure) etc. respectively
Class data are classified, extracted, cleaned and quality analysis.Uniform data mesh is constructed according to dimensions such as carrier attribute, service attributes
Record constructs professional subject data base, and having data, class is orderly, convenient for analysis and application.
It is the system core that data, which see clearly link, according to the more of the target object (including often controlling list and facing control list) inputted
Dimension data constructs complete archives, on the basis of classification analysis, clustering, association analysis and statistical analysis, according to early warning mould
Type carries out comprehensive analysis and prediction with the behavior to target object.
Specifically, in the present embodiment, the truthful data based on the target object obtained from multiple operation systems, which constructs, to be used
In the dynamic integral model that the behavior to target object is studied and judged and predicted, using the dynamic integral model to target object reality
Shi Hangwei is assessed, and corresponding assessed value is obtained.Individual information based on target object, using classical integral model to mesh
Mark object is assessed, to obtain corresponding assessed value.
The assessed value and utilization classical integral mould for the real-time behavior of target object that utilization dynamic integral model is obtained later
The assessed value for the target object that type obtains is weighted summation, the comprehensive assessment value of the real-time behavior of target object is obtained, to mesh
The behavior for marking object carries out comprehensive analysis and prediction.Here comprehensive assessment value Ftotal(t) it can be indicated with following expression:
Ftotal(t)=(1-m) Fclassic(t)+mFupdate(t) (1)
In expression formula (1), Fclassic(t) represent using classical integral model obtain target object assessed value (or
For static integration value), individual information and classical integral model based on target object embody business experience (expertise)
Accumulation.The relevant information of classical integral model can be looked into existing disclosed technical literature, just not detail in the application.
In expression formula (1), m indicates weighting coefficient, value [0,1], by business personnel according to the dynamic model of building at
Setting, that is, the weighting coefficient for being weighted summation are configured ripe degree according to the maturity of dynamic integral model on demand.For example,
Dynamic integral model running early period, in the case that training data lacks, value can be reduced suitably；With the increase of training data,
The maturity of model dynamic part increases, and value increases accordingly.
In expression formula (1), Fupdate(t) assessment of the real-time behavior of target object obtained using dynamic integral model is represented
Value (or being dynamic integral value), is used to characterize very fugacious information contain in behavioral data and artificial, phase is inside the Pass
Appearance is described in detail later；
And comprehensive analysis and the detailed process of prediction are in the present embodiment, as comprehensive assessment value FtotalIt (t) is more than early warning door
When limit value, periphery police strength deployment scenario is considered as a whole according to the position of target object and trend, disposition is generated based on pre-defined rule
Instruction, indicates that neighbouring police strength unit carries out corresponding control processing, carries out control disposition to target object to realize.
In addition, as shown in Figure 1, when police strength unit complete disposition task after, result will be disposed by also needing is put in storage storage automatically, with
For dynamic integral model further to be adjusted and is optimized as sample data.Relative to the existing integral mould of industry
For type, the depth of the stronger adjusting data information of dynamic integral model is excavated and is utilized, and from data, study is related quasi- automatically
Then, it is not completely cured and model and parameter is optimized according to the new data that import.
Below with reference to Fig. 2 to dynamic integral value F in the present inventionupdate(t) acquisition process is specifically described.
In the present invention, the dynamic integral value F of the target object behavior of acquisitionupdate(t) basic ideas are that building is based on
The mathematical model of machine learning carries out this model training using historical sample data, the mathematical model optimized；Further according to weight
The current real-time behavioral data of point personnel, obtains dynamic integral value in line computation in time.
As shown in Fig. 2, being arranged first to the truthful data of the target object obtained from multiple operation systems, instructed
Practice sample set.
Training sample concentrates data packet to include but be not limited to, and in control state (controlled or uncontrolled), verifies the time (when from last time
Between more long possibility out of control it is bigger), leave the permanent residence time, go to place, social interaction, belongings, economic situation etc..
These data are usually directly to be converged and formed by the multiple operation systems of public security and social information in the regular period.When by as with
When constructing model and the sample data being trained, there will be redundancy between sample data and on each attribute of sample data
Even contradiction, leads to the presence of all kinds of " noises ", can slow down the convergence rate of machine learning model, increase trained workload,
As sample is unevenly distributed in classification in vector space, subsequent early warning accuracy is influenced.
For example, following table 1 show sample data example.In table 1, emphasis personnel A is from this section of 15:00~18:30
Time, frequent activity form 4 information.3rd article of information and the 4th article of information time interval are short, mutual information redundance
It is high.And " place finally occur " of every information, " sensitizing range monitoring ", " WiFI trace information " content are similar.Obviously, if
Each attribute is used as an independent dimension to go if participating in modeling, and model will be extremely complex.And emphasis personnel B activity is not frequently
It is numerous, only form 1 information.Obviously, sample is unevenly distributed in classification in vector space.If 4 of emphasis personnel A
Information and 1 information of emphasis personnel B are all used for model training, and will lead to the model trained can be more biased towards in " emphasis personnel
A " class personnel, i.e. over-fitting are in " emphasis personnel A " class.
1 sample data example of table
Therefore, to solve the above problems, the present invention increases sample clean, sample dimensionality reduction in the machine learning process on basis
Two links are respectively used to solve sample distribution problem of non-uniform and sample attribute value redundancy issue in vector space.Specifically,
As in the present invention as shown in Fig. 2,
After obtaining training sample set, the sample data that training sample is concentrated is carried out at similarity calculation and cluster first
Reason carries out sample clean, to obtain cluster heart sample.Specific step is as follows for sample clean:
Step 1, concentrate one sample of any selection labeled as the cluster heart of cluster in training sample；
Step 2, it is concentrated from training sample and chooses a sample in remaining sample as new addition sample；
Step 3, new addition sample and each label are calculated using the range formula (following formula (2)) based on cosine angle
For the similarity of the sample of the cluster heart,
cos(Xi,Xj)=(Xi.Xj)/(|Xi|×|Xj|) (2)
In expression formula (2), XiAnd XjThe corresponding vector of respectively sample i and sample j indicates.
If each similarity being calculated is both less than predetermined threshold, new cluster is established centered on the new addition sample；
Otherwise by the new addition sample be added to the highest cluster of its similarity, and recalculate the cluster heart of the cluster；
It repeats step 2 and to obtain several clusters, and chooses each cluster until having handled the sample of training sample concentration to step 3
The cluster heart corresponding to sample as cluster heart sample.
In the present invention, by above-mentioned sample clean, the biggish sample of similarity in classification is reduced, solves following model
Building is existing to report over-fitting, and then can the accuracy studied and judged of lift scheme.
Later, as shown in Fig. 2, carrying out dimension-reduction treatment to cluster heart sample, the orthogonalization characteristic quantity of cluster heart sample and right is obtained
The weighted value answered.
For example, principal component analytical method (PCA analysis) can be used, dimension-reduction treatment is carried out to cluster heart sample, to obtain cluster
The orthogonalization characteristic quantity y of heart sample1(t) ..., ym(t) and corresponding weighted value a0, a1..., am.It is high in sample attribute redundancy
In the case where, m here that far smaller than sample is original attribute value quantity can significantly reduce subsequent builds mathematical model
Complexity.
In the present invention, above-mentioned sample dimensionality reduction is that there are the situations of redundancy and " noise " on each attribute for sample data
, after sample dimensionality reduction, the complexity of subsequent builds mathematical model is reduced, accelerates model convergence rate.
Continue later as shown in Fig. 2, carrying out model structure according to the orthogonalization characteristic quantity and corresponding weighted value of cluster heart sample
It builds, building obtains initial model.
As a preference, with sigmoid functionFor kernel function, building obtains following initial model F
(t):
F (t)=a1y1(t)+a2y2(t)+...+amym(t)+a0 (3)
In expression formula (3), y1(t) ..., ym(t) it is orthogonalization characteristic quantity, indicates input vector, a0, a1..., amFor correspondence
Weighted value, be initial weight.Here initial model is linear model, can make that subsequent calculating process is faster, model convergence
Faster, more agree with the real-time demand that personnel deploy to ensure effective monitoring and control of illegal activities.
Continue later as shown in Fig. 2, be trained using the orthogonalization characteristic quantity of cluster heart sample to obtained initial model,
By model training, optimizes weighted value, thereby determine that dynamic integral model.In the present embodiment, the detailed process of model training is such as
Under:
The data input initial model of either cluster heart sample in cluster heart sample is calculated, by obtained calculated value and is somebody's turn to do
The corresponding actual value of data of cluster heart sample is compared, and obtains the corresponding error of cluster heart sample；
All cluster heart samples are traversed, the corresponding error of each cluster heart sample is obtained；
It is for statistical analysis to the error of all cluster heart samples, weight update is carried out to initial model based on analysis result,
To obtain dynamic integral model.
Continue later as shown in Fig. 2, using information such as the real-time behaviors of target object as input, based on obtained dynamic product
Sub-model calculates the dynamic integral value of the real-time behavior of target object, and the static integration value and early warning thresholding of combining target object
Carry out comprehensive assessment to the real-time behavior of target object.
When the comprehensive assessment value that assessment obtains triggers early warning thresholding, the corresponding disposition that comprehensive assessment value is mapped as is instructed
Rank notifies that nearby police strength is disposed.
To finally result be disposed and carry out this time and calculates input data (i.e. this real-time behavior etc. of target object assessed
Information) carry out convergence form new sample data, and feed back to database be added training sample set, with for subsequent model it is excellent
Change, such as dynamic integral model is trained again using the sample data, to continue to optimize the integral model, includes as utilized
The training sample set of the sample data re-starts the process of building and training optimization, to obtain corresponding dynamic integral model.
The system that the present invention also proposes personnel's intelligence control, including computer readable storage medium, wherein being stored with journey
Sequence, the method which realizes personnel's intelligence control as described in the embodiment above when being executed by processor.
The present invention is directed to the serial problem of current emphasis personnel control, proposes one kind and is intelligently studied and judged with big data as core
Novel pipe diameter design, cover information convergence, information collation, information collision, early warning disposition and disposition and feed back multiple links.
In the pipe diameter design, the invention proposes novel emphasis personnel's comprehensive analysis integral model, the models
It is not limited to a certain specific information source to be studied and judged, but the people of integration objective object, vehicle, electricity, net, grinds as multidimensional data
Sentence.The comprehensive analysis integral model can support the real-time discovery of target object abnormal behaviour, and provide Warning Service in time.
The comprehensive analysis integral model has merged expertise and big data technology, is model calculating by empirical conversion.And
Dynamic model in the comprehensive analysis integral model can on-line automatic study and optimization, and to carry out model excellent for sustainable utilization data
Change, is constantly examined in actual combat, General Promotion studies and judges accuracy.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Within the technical scope disclosed by the invention, any changes or substitutions that can be easily thought of by any those skilled in the art, should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be subject to the protection scope in claims.
Claims (10)
1. a kind of method of personnel's intelligence control, comprising the following steps:
Truthful data building based on the target object obtained from multiple operation systems is ground for the behavior to target object
The dynamic integral model sentenced and predicted assesses the real-time behavior of target object using the dynamic integral model；
Individual information based on target object assesses target object using classical integral model；
It assessed value to the real-time behavior of target object obtained using the dynamic integral model and is obtained using classical integral model
The assessed value of the target object obtained is weighted summation, obtains the comprehensive assessment value of the real-time behavior of the target object；
When the comprehensive assessment value is more than early warning threshold value, disposition instruction is generated based on pre-defined rule, target object is carried out
Control disposition.
2. the method according to claim 1, wherein the weighting coefficient of the weighted sum is according to dynamic product
The maturity of sub-model is configured.
3. the method according to claim 1, wherein described based on the target object obtained from multiple operation systems
Truthful data construct the dynamic integral model being studied and judged and predicted for the behavior to target object, comprising the following steps:
Arrangement is collected to the truthful data of the target object, obtains training sample set；
Similarity calculation and clustering processing are carried out to the sample data that the training sample is concentrated, obtain cluster heart sample；
Dimension-reduction treatment is carried out to the cluster heart sample, obtains the orthogonalization characteristic quantity and corresponding weighted value of cluster heart sample；
According to the orthogonalization characteristic quantity and corresponding weighted value of cluster heart sample, initial model is constructed；
The initial model is trained using the orthogonalization characteristic quantity of cluster heart sample, optimizes weighted value, is thereby determined that described
Dynamic integral model.
4. according to the method described in claim 3, it is characterized in that, constructing the introductory die using sigmoid function as kernel function
Type F (t):
F (t)=a1y1(t)+a2y2(t)+...+amym(t)+a0
Wherein, y1(t) ..., ymIt (t) is orthogonalization characteristic quantity, a0, a1..., amFor corresponding weighted value.
5. according to the method described in claim 3, it is characterized in that, the orthogonalization characteristic quantity using cluster heart sample is to described
Initial model is trained, and is optimized weighted value, is thereby determined that the dynamic integral model, specifically,
The data of either cluster heart sample in the cluster heart sample are inputted the initial model to calculate, the calculated value that will be obtained
Actual value corresponding with the data of the cluster heart sample is compared, and obtains the corresponding error of cluster heart sample；
All cluster heart samples are traversed, the corresponding error of each cluster heart sample is obtained；
It is for statistical analysis to the error of all cluster heart samples, weight update is carried out to the initial model based on analysis result,
To obtain the dynamic integral model.
6. according to the method described in claim 3, it is characterized in that, the sample data concentrated to the training sample carries out
Similarity calculation and clustering processing obtain cluster heart sample, specially
Step 1 concentrates one sample of any selection labeled as the cluster heart of cluster in the training sample；
Step 2 is concentrated from the training sample and chooses a sample in remaining sample as new addition sample；
Step 3 uses the range formula based on cosine angle to calculate the new addition sample and each label as the sample of the heart
Similarity,
If each similarity being calculated is both less than predetermined threshold, new cluster is established centered on the new addition sample；
Otherwise by the new addition sample be added to the highest cluster of its similarity, and recalculate the cluster heart of the cluster；
It repeats step 2 and to obtain several clusters, and chooses each cluster until having handled the training sample concentrates sample to step 3
Sample corresponding to the cluster heart is as the cluster heart sample.
7. according to the method described in claim 3, it is characterized in that, being carried out using principal component analytical method to the cluster heart sample
Dimension-reduction treatment.
8. according to the method described in claim 3, it is characterized in that, further include, by the real-time behavior of the target object and right
New sample data of the disposition result answered as target object is added training sample and concentrates, with described for advanced optimizing
Dynamic integral model.
9. method according to any one of claim 1 to 8, which is characterized in that the operation system include alert comprehensive platform,
Net comprehensive platform, skill detects platform, at least one of PGIS platform, informational intelligence summary platform platform.
10. a kind of system of personnel's intelligence control, including computer readable storage medium, wherein it is stored with program, the journey
The method that sequence realizes personnel's intelligence control as claimed in any one of claims 1-9 wherein when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811452187.3A CN109559045A (en) | 2018-11-30 | 2018-11-30 | A kind of method and system of personnel's intelligence control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811452187.3A CN109559045A (en) | 2018-11-30 | 2018-11-30 | A kind of method and system of personnel's intelligence control |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109559045A true CN109559045A (en) | 2019-04-02 |
Family
ID=65868187
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811452187.3A Pending CN109559045A (en) | 2018-11-30 | 2018-11-30 | A kind of method and system of personnel's intelligence control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109559045A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110659335A (en) * | 2019-09-26 | 2020-01-07 | 四川科瑞软件有限责任公司 | Analysis method for recessive drug addicts |
| CN110852517A (en) * | 2019-11-15 | 2020-02-28 | 北京明略软件系统有限公司 | Abnormal behavior early warning method and device, data processing equipment and storage medium |
| CN111310100A (en) * | 2020-01-06 | 2020-06-19 | 北京中天锋安全防护技术有限公司 | Intelligent evaluation method for drug addict |
| CN111324656A (en) * | 2020-02-25 | 2020-06-23 | 深圳市天彦通信股份有限公司 | Personnel management method and related equipment |
| CN112132357A (en) * | 2020-09-29 | 2020-12-25 | 佳都新太科技股份有限公司 | Behavior prediction method, behavior prediction device, behavior prediction equipment and storage medium based on big data |
| CN116109041A (en) * | 2023-03-06 | 2023-05-12 | 中建安装集团有限公司 | Engineering information security management system and method for digital project level |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546484A (en) * | 2009-04-30 | 2009-09-30 | 南京航空航天大学 | Flight delay conformance analysis and forecast system based on SOA and operation method thereof |
| CN104299003A (en) * | 2014-11-02 | 2015-01-21 | 天津理工大学 | Gait recognition method based on similar rule Gaussian kernel function classifier |
| CN106295565A (en) * | 2016-08-10 | 2017-01-04 | 中用环保科技有限公司 | Monitor event identifications based on big data and in real time method of crime prediction |
| CN106485635A (en) * | 2016-10-08 | 2017-03-08 | 浪潮软件集团有限公司 | Dynamic integral early warning method in anti-terrorism field |
| CN106780263A (en) * | 2017-01-13 | 2017-05-31 | 中电科新型智慧城市研究院有限公司 | High-risk personnel analysis and recognition methods based on big data platform |
| CN107463587A (en) * | 2016-06-06 | 2017-12-12 | 支录奎 | Method of the police work two-way integration in Police Geographic Information System visualization classification application |
| CN108366374A (en) * | 2018-03-08 | 2018-08-03 | 中国联合网络通信集团有限公司 | A suspect's judgment method and device |
| CN108805202A (en) * | 2018-06-11 | 2018-11-13 | 上海新增鼎数据科技有限公司 | A kind of machine learning method and its application for electrolytic cell fault pre-alarming |
-
2018
- 2018-11-30 CN CN201811452187.3A patent/CN109559045A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546484A (en) * | 2009-04-30 | 2009-09-30 | 南京航空航天大学 | Flight delay conformance analysis and forecast system based on SOA and operation method thereof |
| CN104299003A (en) * | 2014-11-02 | 2015-01-21 | 天津理工大学 | Gait recognition method based on similar rule Gaussian kernel function classifier |
| CN107463587A (en) * | 2016-06-06 | 2017-12-12 | 支录奎 | Method of the police work two-way integration in Police Geographic Information System visualization classification application |
| CN106295565A (en) * | 2016-08-10 | 2017-01-04 | 中用环保科技有限公司 | Monitor event identifications based on big data and in real time method of crime prediction |
| CN106485635A (en) * | 2016-10-08 | 2017-03-08 | 浪潮软件集团有限公司 | Dynamic integral early warning method in anti-terrorism field |
| CN106780263A (en) * | 2017-01-13 | 2017-05-31 | 中电科新型智慧城市研究院有限公司 | High-risk personnel analysis and recognition methods based on big data platform |
| CN108366374A (en) * | 2018-03-08 | 2018-08-03 | 中国联合网络通信集团有限公司 | A suspect's judgment method and device |
| CN108805202A (en) * | 2018-06-11 | 2018-11-13 | 上海新增鼎数据科技有限公司 | A kind of machine learning method and its application for electrolytic cell fault pre-alarming |
Non-Patent Citations (6)
| Title |
|---|
| 冯文刚: ""基于深度学习的民航安检和航班预警研究"", 《数据分析与知识发现》 * |
| 姜桂艳: "《道路交通状态判别技术与应用》", 30 November 2004, 人民交通出版社 * |
| 崔蒿: "《情报攻歼全维战法 战术研判方法模型与实战操作》", 31 January 2013, 中国人民公安大学出版社 * |
| 胡小生: ""一种基于聚类提升的不平衡数据分类算法"", 《集成技术》 * |
| 赵守香: "《大数据分析与应用》", 31 December 2015, 航空工业出版社 * |
| 韩忠明: "《数据分析与R》", 31 August 2014, 北京邮电大学出版社 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110659335A (en) * | 2019-09-26 | 2020-01-07 | 四川科瑞软件有限责任公司 | Analysis method for recessive drug addicts |
| CN110852517A (en) * | 2019-11-15 | 2020-02-28 | 北京明略软件系统有限公司 | Abnormal behavior early warning method and device, data processing equipment and storage medium |
| CN110852517B (en) * | 2019-11-15 | 2023-08-25 | 北京明智和术科技有限公司 | Abnormal behavior early warning method and device, data processing equipment and storage medium |
| CN111310100A (en) * | 2020-01-06 | 2020-06-19 | 北京中天锋安全防护技术有限公司 | Intelligent evaluation method for drug addict |
| CN111324656A (en) * | 2020-02-25 | 2020-06-23 | 深圳市天彦通信股份有限公司 | Personnel management method and related equipment |
| CN111324656B (en) * | 2020-02-25 | 2023-08-08 | 深圳市天彦通信股份有限公司 | Personnel management method and related equipment |
| CN112132357A (en) * | 2020-09-29 | 2020-12-25 | 佳都新太科技股份有限公司 | Behavior prediction method, behavior prediction device, behavior prediction equipment and storage medium based on big data |
| CN116109041A (en) * | 2023-03-06 | 2023-05-12 | 中建安装集团有限公司 | Engineering information security management system and method for digital project level |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109559045A (en) | A kind of method and system of personnel's intelligence control | |
| Zhang et al. | Intelligent computing system based on pattern recognition and data mining algorithms | |
| Khalil et al. | Machine Learning, Deep Learning and Statistical Analysis for forecasting building energy consumption—A systematic review | |
| Janowicz et al. | Know, Know Where, KnowWhereGraph: A densely connected, cross-domain knowledge graph and geo-enrichment service stack for applications in environmental intelligence | |
| Karaboga et al. | A novel clustering approach: Artificial Bee Colony (ABC) algorithm | |
| CN101782976B (en) | Automatic selection method for machine learning in cloud computing environment | |
| Zhang et al. | A novel case adaptation method based on an improved integrated genetic algorithm for power grid wind disaster emergencies | |
| Dewangan et al. | Load forecasting models in smart grid using smart meter information: A review | |
| CN109636055A (en) | A kind of non-coal mine Safety Risk in Production prediction and warning platform | |
| CN104636751A (en) | Crowd abnormity detection and positioning system and method based on time recurrent neural network | |
| CN109583635A (en) | A kind of short-term load forecasting modeling method towards operational reliability | |
| CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
| Zhao et al. | A review of macroscopic carbon emission prediction model based on machine learning | |
| CN111062511B (en) | Aquaculture disease prediction method and system based on decision tree and neural network | |
| Geng | Influence of self-efficacy improvement on online learning participation | |
| CN109635008B (en) | Equipment fault detection method based on machine learning | |
| Verma et al. | ICT and Mobile Technology features predicting the university of Indian and Hungarian student for the real-time | |
| Stutz | Modeling the pre-industrial roots of modern super-exponential population growth | |
| CN113537607B (en) | Power failure prediction method | |
| CN115293249A (en) | Power system typical scene probability prediction method based on dynamic time sequence prediction | |
| Abdelbari et al. | Optimising a constrained echo state network using evolutionary algorithms for learning mental models of complex dynamical systems | |
| Dang | Optimization of the Intelligent Sports Health Management System Based on Big Data Technology and Internet of Things | |
| Gao et al. | Statistics and Analysis of Targeted Poverty Alleviation Information Integrated with Big Data Mining Algorithm | |
| Prasad et al. | A decision support system for agriculture using natural language processing (ADSS) | |
| Hadiyat et al. | Comparing statistical feature and artificial neural networks for control chart pattern recognition: a case study |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190402 |
|
| RJ01 | Rejection of invention patent application after publication |