CN113571157A - Intelligent risk person psychological image recognition system based on FMT characteristics - Google Patents

Abstract

The invention discloses an intelligent risk personnel psychological image recognition system based on FMT characteristics, which comprises the following steps: s1: feature screening, key feature collection and risk feature screening; s2: high risk person psychology portrait sketching and evaluating; s3: compiling a scale, measuring and collecting data; s4: data analysis, determining an image model through Binarylogicic regression; s5: and obtaining the final risk grade of the tester according to a fuzzy comprehensive judgment method. According to the FMT feature-based intelligent risk person psychological image recognition system, a high-risk person psychological image is constructed through the individual psychological feature information of people who have had risk behaviors, such as collector-end psychological persons and wounded persons. According to the fitting condition of the psychological images of the testers and the high-risk personnel, the high-risk personnel and the non-risk personnel are distinguished by adopting a fuzzy comprehensive judgment method, and the real high-risk population is found more comprehensively and specifically in a targeted manner.

Description

Intelligent risk person psychological image recognition system based on FMT characteristics

Technical Field

The invention relates to the field of psychological assessment, in particular to a system for intelligently identifying psychological images of risk personnel based on FMT characteristics.

Background

Psychometric tests are tools that measure a certain psychology or behavior of an individual according to standardized procedures, and the measurement results can be used to determine interpersonal differences. Psychometric measurement is a technique that standardizes human psychology and behavior. China psychologist Penkeping considers: "psychological test is a scientific means of making inferences and quantifications according to certain principles about psychological characteristics throughout human behavioral activities by observing a few representative behaviors of a human.

In recent years, mental health becomes more and more a focus of modern people, and the mental health is in parallel with physical health and becomes one of key indexes for measuring the health level of human beings. Actively preventing and reducing psychological diseases, restraining the occurrence of psychological crisis events, improving the quality of working and living, and promoting the harmonious development of individuals and society have become the problems of common concern and research of people.

At present, the contradiction associated with the mental health needs is mainly reflected in these aspects: the psychological health service resources of hospital outpatients and psychological health centers are very limited, and the evaluation, diagnosis and treatment needs of all doctors cannot be fully met; traditional assessment tools used by psychological consulting and EAP institutions in service lack a complete and effective assessment system; sensitive and quick psychological crisis early warning means for college students need to be built vigorously; the instability of the mental health level and the stress level of enterprise staff and managers greatly affects the talent reservation and the productivity performance of enterprises, and causes serious problems of staff loss, low staff satisfaction degree and the like. Among these problems, the lack of a good mental health assessment system is the basis and key for all the problems.

Currently, mental health assessment tools used domestically include two types, i.e., foreign-developed domestic revision tools and domestic autonomous development tools. The former is the majority in the use of all mental health assessment tools, but various inadaptations can be generated due to cultural differences and translation problems; the latter is mainly used for special research, and from content to simpler structure and less systematization, a great amount of deep exploration and development are needed. The mental health test and evaluation workers have the following: comprehensive assessment tools, mainly including symptom self-rating scale (SCL-90), Minnesota polyphasepersonalitic test series (MMPI; including MMPI-1, MMPI-2 and MMPI-juveniles edition); ② an assessment tool aiming at specific pathological symptoms, which mainly comprises a Hamilton depression scale, a depression self-rating scale, a Beck anxiety questionnaire, a Marks terrorism obsessive-compulsive scale and the like; and the clinical diagnosis system mainly comprises a diagnosis and classification manual (DSM) and the like. Most of domestic psychological health researches directly use western related theories and concepts, and are directly applied to scientific research, psychological counseling and consultation through revising and formulating scales. However, the cultural and social environments of east and west do not define the same mental health content. The cultural psychological connotation contained in the western questionnaire is difficult to be accurately and effectively applied to Chinese people even if revised. Therefore, it is urgent and important to construct a mental health assessment tool that meets the Chinese cultural background and can provide a personalized mental assessment service.

As far as present, in the field of mental health assessment, there are mainly research and development and technical problems:

the tool system is seriously aged and outdated, and the current domestic commonly-used evaluation tool lacks the absorption of the latest results of the current mental health research. Most of evaluation institutions directly use traditional mental health evaluation tools, such as SCL-90, or develop new scales by splitting and recombining dimensions in different traditional scales, and the latest research results are selected without time. The current generation of mental health is not investigated using the latest research results. In addition, the language of the evaluation item is largely expressed by people before many years, and the change of language habits of people caused by times change is not considered, so that ambiguity is easily caused, evaluation is wrong, and meanwhile, testees feel uncomfortable and a conflicting feeling is easily generated. Currently, most psychological health assessment tools used by assessment companies are translated from foreign countries in the early days, and language expression is hard, so that discontent of a person to be assessed is caused, and the assessment result is inaccurate.

The existing scale is mainly used for measuring symptoms basically, tools developed and used in China are mainly used for measuring symptoms, whether an individual suffers from psychological diseases or not can be judged from a negative angle, and the psychological health cannot be comprehensively and integrally evaluated from positive angles such as social activity range, life satisfaction, self-esteem, social support, subjective happiness and the like, so that the real level of the psychological health of the individual is difficult to be fully reflected. In addition, for the measurement mainly based on symptoms, the evaluation result is often well understood by people with a certain psychological knowledge base, and for the ordinary testee or the enterprise manager, the understanding has certain difficulty, and the expert is often required to fully understand the evaluation result after the interpretation of the evaluation result.

Symptom assessment and functional assessment are mixed up, the existing mental health assessment tool does not clearly distinguish the functional assessment from the symptom assessment, and the existing mental health assessment tool generally takes the functional assessment as a certain aspect of the symptom assessment and does not realize the special role of the functional assessment. For the evaluation of common people, the evaluation of social functions has the overall and comprehensive evaluation properties, and is more reasonable and applicable. Just like going to a hospital for a patient, the patient does not need to be subjected to X-ray fluoroscopy by every person, and most of the patients can obtain feedback through medical guidance and clinical inquiry. That is, most of the current mental health assessment tools are to distinguish normal persons from persons with psychological problems, but there is no better screening for those in "mental sub-health", that is, although some persons do not have obvious psychological problems such as depression, anxiety and the like, the psychological problems in physical functions such as characters, coping ways, psycho-elastic and the like are in the edge of the psychological problems.

The etiology analysis framework of the psychological diseases is not complete, and the existing evaluation tools provide rich materials for checking the clinical symptoms of the psychological diseases, but lack contribution to the etiology analysis of the psychological diseases. More and more studies confirm that psychological diseases are related to various internal and external factors such as individual defense mechanism, attaching mode, social support, growth environment and the like, and the results of the studies are not absorbed by the recent development of mental health assessment tools, and cannot provide etiology analysis references for clinical treatment of psychological consultants and psychotherapists.

The native scale has low text quality and suspicious normal mode quality, and the problems of normal mode and text quality measurement and the like can also exist by directly translating and using foreign evaluation tools. Foreign tools must be provided with a domestic standard before use. In addition, the translated tools often have language and character quality problems, the deviation occurs in terms, grammar and expression habits, and all the factors can obviously interfere with the evaluation effect.

The evaluation results are mostly based on the individual conditions of the testee, and currently, most evaluation results given by evaluation institutions are used for evaluating the mental health level of the testee, but the enterprise or colleges where the testee is located often need to be evaluated. The enterprises or colleges are concerned more about which risks are brought to the enterprises and what losses are caused by the psychological health problems of the employees or students.

The evaluation method generally adopts a score conversion method, and the evaluation method mainly comprises the following 3 methods: one is the developmental scale. Developmental scales can be made by comparing the performance of an individual with those of various developmental levels. Commonly used are age scales and grade scales. Second is the quotient. The most famous is the wisdom quotient, which is made by comparing the mental development level of a person with the actual age. In educational tests, quotients are sometimes used to indicate the rate of educational development or achievement, and often educational quotients and achievement quotients are seen. And three is a percentage scale. The percentage rating represents the relative position of the individual in the normative community. And fourthly, standard scores. Is a scale that expresses the distance of the original score from the mean in units of standard deviation. Since its units are standard deviations, it is called a standard score. The conversion of the standard score is in two ways, namely linear conversion, namely conversion is carried out by using a formula of Z ═ X/SD, wherein Z is the standard score, X is the score of the testee, X is the average number of samples, and SD is the standard deviation of the score of the samples. And secondly, converting the original scores into percentage grades, and then searching corresponding standard scores from a normal curve area table. A prerequisite for this conversion is that the score of the measured trait should be, in fact, normally distributed. These score conversion methods usually require explicit numerical definitions, but in fact, in daily life and work, we often encounter many ambiguities that do not have significant numerical limits, such as "beautiful, hot, far", etc., which are not simply expressed in terms of "yes", "no" or "digital". Generally, the evaluation of a thing is not only from a certain aspect, but also involves many indexes, and the psychological health condition of an individual is particularly the same.

The psychological health is determined mainly by comparing with the norm, which compares the tested person's performance with the relative group formed by persons with certain characteristics, and reports his performance according to the relative position of a person in the group. Here, the reference group used for comparison is called a normative group, and the score distribution of the normative group is called a normative. Determination of the normative community: a normative community is a group of people with some common characteristics. When determining the normative community, attention is paid to: one is that the composition of the population must be defined: secondly, when the population is too large, the normative group is a representative sample of the population to be tested; thirdly, the sampling process must be described in detail; fourthly, the sample size needs to be proper; fifthly, attention is paid to the frequent timeliness; sixthly, the common normal mode and the special normal mode are combined.

Therefore, in the existing mode, we can only judge the relative position of the individual in the group where the individual is located, and then we can only obtain the following by the mode of normal reference: when the mental health score of an individual is far from the score of the norm, the mental health condition is poor, so that the mental health of the individual changes along with the change of the reference norm, and a more accurate mental health condition cannot be given.

Disclosure of Invention

The invention aims to provide an intelligent risk person psychological image recognition system based on FMT characteristics, so as to solve the problems in the background technology.

In order to achieve the purpose, the invention provides the following technical scheme:

intelligent risk person psychological image recognition system based on FMT characteristics comprises the following steps:

s1: feature screening, key feature collection and risk feature screening;

s2: high risk personnel psychology portrait model drawing and evaluation;

s3: compiling a scale, measuring and collecting data;

s4: data analysis, determining an image model through Binarylogicic regression;

s5: and obtaining the final risk grade of the tester according to a fuzzy comprehensive judgment method.

As a further scheme of the invention: in the step S1, key features are collected, 10 enterprise high-level managers in different industries are queried in a semi-structured interview mode for 30-45 minutes, staff risk events which have occurred inside an enterprise, such as extreme psychological, wounded, and deliberate destruction of employees, and some risk behaviors which are worried by staff to implement by staff are known, 56 risk staff key features are obtained by arranging, risk feature screening in the step S1 is performed, key features are evaluated in a voting mode, 8-bit qualified HR is invited to vote for each key feature, and items which are selected by more than 60% of qualified HR and are related to the key features of the risk staff are retained. Therefore, 34 risk characteristics are selected, and 6 types of mental risk staff figure images, extreme mental tendency mental images, wounded person tendency mental images, departed person recovery mental images, high energy consumption tendency mental images, negative energy propagation mental images and glass bottle image mental images are preliminarily formed.

As a still further scheme of the invention: the psychological portrait model sketching of the high-risk personnel in S2 mainly adopts a literature analysis method to read and learn related documents at home and abroad on different psychological risk portraits, learns and masters the risk models of the different human portraits, and comprehensively sketches 6 types of psychological risk portrait by combing and summarizing the existing documents and academic achievements through 5 dimensions of external stimulus, individual emotional state, personality, cognition and psychological elasticity received by an individual, and the psychological risk model assessment of the high-risk personnel in S2 comprises three layers of screening of staff social function condition, checking of clinical symptoms and psychological cause analysis, screening of staff social function condition, and covers interpersonal quality, efficiency of doing things, emotional condition, ambition and range of activities, and the interpersonal quality assessment comprises whether interpersonal relationship is stable and harmonious; the performance efficiency evaluation comprises the functional operation efficiency of memory, reaction, learning and the like; the emotional condition assessment includes whether the subject can maintain a stable, mild emotional state; positive sensory evaluation includes whether an individual has a sense of well-being, a sense of achievement, etc.; the activity range investigation comprises whether an individual has enough abundant social interaction and entertainment activities, the clinical symptom checking mainly analyzes the psychological problems of the employee, such as depression, hostility-attack, paranoia and the like, so that the psychological problems are determined according to the clinical symptom checking scale, and the psychological cause analysis mainly comprises stress events, defense mechanisms, social support and the like, so that the psychological causes of the employee are analyzed.

As a still further scheme of the invention: the method comprises the steps of S3, compiling initial tables according to dimensions of different types of Psychological pictures, compiling the initial tables according to a Psychological health symptom self-rating table (SCL-90), an extreme Psychological tendency table in the internal Psychological model, a Barrat impulse table, a pressure perception table (PSS) compiled by doctor Cohen, compiling the initial tables containing 450 subjects, applying measurement and collecting data in S3, and adopting a fixed-throw mode, a random-throw mode and a large-range throw mode, wherein 6 accurate parts are thrown to typical portrait characters in the fixed-throw mode, for example, a part table for measuring the extreme Psychological tendency is thrown to testers who have extreme Psychological behaviors but are not psychologically thrown, a part table for measuring the injurious tendency is thrown to testers who have excessive injurious behaviors, and the like, wherein the random-throw mode is thrown to general normal people, the large-range is mainly thrown to an APP platform, the obtained product is free for measurement of users, invalid questionnaires are removed through data cleaning, and 16805 parts of the recovered valid questionnaires are obtained.

As a still further scheme of the invention: the data analysis in the S4 includes performing reliability and validity analysis, wherein the reliability is an index reflecting the stability and identity of a tester, the stability and identity of the scale are tested by adopting a 'Krenbach alpha coefficient' index, the validity is an index reflecting the authenticity and accuracy of the measurement to what degree the characteristic to be measured is measured by a test, the collected data is subjected to exploratory factor analysis by adopting SPSS20.0, the validity of the programmed scale is tested, and an inappropriate question is deleted through factor analysis so as to obtain good structural validity.

As a still further scheme of the invention: the portrait model is determined in the S4, the data of testers who do scales with different risk types are divided into two types, namely two types of extreme psychology and non-extreme psychology, two types of wounded people and non-wounded people and the like, the independent variables which really cause the individuals to take the behaviors of extreme psychology, wounded people, post-job reporting and the like are found out by adopting a binary logistic regression mode, statistical analysis is carried out by adopting a logarithmic linear model, and when the two classified variables are taken as dependent variables in the logarithmic linear model, the logarithmic linear regression model becomes the Binarylistic regression model. The derivation and solution of the BinaryLogistic model are described in detail below by taking an extreme psychological tendency model as an example,

assuming N, N ∈ N, individual takes extreme psychological behavior y_nDenotes y_n1 represents an individual taking extreme psychological behaviour, whereas y_n0, assuming in theory that there is one continuous reaction variable y^* _nRepresents y_nThe probability of occurrence, which ranges from negative infinity to positive infinity, causes the traveler n to take extreme psychological actions if the value of this variable exceeds a critical point m, for example, m is 0, and then:

when y is^* _n>0 time y_n＝1

When y is^* _nY at ≦ 0_n＝0

If it is assumed that the amount of strain y is in the opposite direction^* _nWith independent variable selection of influencing factor, x_niThere is a linear relationship between (i ═ 1,2 …, m), i.e.

y^* _n＝β₀+β₁x_n1+…+β_mx_nm+ε_j 1.1

Wherein: epsilon_jIs an error term, obeys Logistic distribution; beta is a₀Is the intercept commonly referred to as the constant term; beta is a_i(i ═ 1,2 …, m) is x_niThe conditional probability of the nth individual taking extreme psychological behavior is given by the formula 1.1 as the partial regression coefficient of (i ═ 1,2 …, m):

P(y_n＝1/x_ni)＝P[(β₀+β₁x_n1+…+β_mx_nm+ε_j)>0] 1.2

the calculation results in:

wherein P (y)_n＝1/x_ni) Is a non-linear function composed of explanatory variables (influencing factors), which can be converted into a linear function.

First, the probability that an extreme psychological behaviour is selected by an individual n is defined as:

the ratio of the probability of occurrence of the extreme psychological behaviour being selected to not being selected is then:

this ratio is called the occurrence ratio of events, abbreviated as Odds advantage, and it can be seen from equation 1.5 that the larger p, the larger Odds; the smaller p, the smaller odds. In order to measure the degree of influence of an independent variable influence factor on a dependent variable by adopting extreme psychological behaviors, the oddstrato abbreviation OR is defined, which is called the odds ratio for short, and the following formula is used:

the meaning of the method is that the independent variable x is under the condition that the influence factors of the selection behavior of other independent variables are unchanged_niBy changing one unit, the corresponding OR of the dependent variable changes exp (. beta.)_i). We use OR to select the influencing factors of extreme psychological behaviors for individuals

Since the binaryogic regression is a nonlinear model, the parameters are solved using maximum likelihood estimation. Assuming a population consisting of N individual selection activities, Y1, … YN, from which N cases were randomly drawn as samples, observations are labeled Y1, … YN. Let p_i＝p(y_i＝1|x_i) For a given influencing factor x_iUnder conditions such that result y is obtained_iWith a conditional probability of 1, then y_iThe conditional probability of 0 is p (y)_i＝0|x_i)＝1-p_iThen, the probability of obtaining an observation is:

wherein y is_iSince each observation is independent of each other, their joint distribution can be expressed as the product of the marginal distributions, as shown by the following equation:

equation 1.8 refers to the likelihood function of whether n individuals will be extremely psychological or not, where:

to find the parameter estimate that maximizes the value of this likelihood function, for L^*A logarithmic transformation is performed to obtain the following function:

to estimate ln (L)^*) The overall parameter beta at which the maximum value is reached₀…β_mThe partial derivatives are calculated separately and then made equal to 0, given the following equation:

…

as can be seen from the above formula, there are m +1 independent variables in the equation set, and there are m +1 simultaneous equations to estimate their values, since the above are all nonlinear functions, the solution by manual computation is very difficult, and the solution is performed by using the sps 20.0.

Through a binary logistic regression mode, 6 regression equations are established to judge whether a tester belongs to a certain type of risk portrait.

The method mainly comprises the steps of giving a certain weight to each measurement dimension according to an equation obtained by regression, enabling a tester to automatically add a dimension value related to a certain image in a system background after the tester completes testing, bringing the value of each dimension into equations of different images, calculating risk values corresponding to different types of images, and indicating that people possibly take behaviors represented by the images are more likely to take the risk values close to 100 points.

As a still further scheme of the invention: in the fuzzy comprehensive judgment method in S5, because the human psychology is often uncertain, and the individual psychological risk level often involves multiple factors or multiple indicators, at this time, it is required to make comprehensive evaluation on the object according to the multiple factors, but cannot evaluate the object only from the condition of a certain factor, so the judgment of the comprehensive psychological risk mainly adopts a fuzzy comprehensive judgment method, where the judgment means to compare and judge the quality and quality of the object according to given conditions; comprehensive means that the evaluation condition comprises a plurality of factors or a plurality of indexes; fuzzy is that an operation method of fuzzy mathematics is used for processing, and fuzzy comprehensive evaluation can effectively synthesize different types of data, including data in non-normal distribution, so that the adaptability is stronger, and finally the evaluation on a testee is obtained.

Inherent logic of fuzzy comprehensive evaluation:

the fuzzy comprehensive evaluation is a very effective multi-factor decision method for comprehensively evaluating things influenced by various factors, and is characterized in that the evaluation result is not absolutely positive or negative, but is represented by a fuzzy set.

The degree of membership or membership function is the basis of fuzzy mathematics and fuzzy systems. The concept of "membership" is introduced to describe the intermediate transition of differences, which is an approximation of the ambiguity by the accuracy, mathematically defined as:

for fuzzy set A on domain of discourse U, a mapping from U to [0,1] is specified:

μ _A :U→[0,1]

u→μ _A (u∈[0,1])

wherein u is _A Is thatAMembership function of u _A (u) is a pair of uADegree of membership. When u is _A When (u) is 1, u is ∈AWhen u is _A When (u) is 0, the reaction is carried out,

when u is _A When (u) is 0 or 1, the membership function is a characteristic function degenerated into a common set, and the same applies toADegenerates to a normal set.

Fuzzy sets are completely characterized by membership functions, which are established for this purpose, in fuzzy mathematics, where a number between 0 and 1 is required to reflect the degree to which an element belongs to a fuzzy set. Whether the affiliation function can be correctly determined is the key to whether the fuzzy set theory can be effectively applied. If the fuzzy set is defined in the real number domain, the membership functions of the fuzzy set are called fuzzy distributions. We mainly use assignment membership functions, so called assignment methods, which apply some form of fuzzy distribution based on the nature of the problem and then determine the parameters contained in the distribution based on the measured data. The fuzzy distribution type used by the people is larger, the larger fuzzy distribution is suitable for describing the fuzzy phenomenon of the larger bias direction such as large, hot, old and color, and the like, and the general form of the membership function is

Where a is a constant and f (x) is a non-decreasing function.

With fuzzy subsets on the set domain U ₁A, ₂A,… _nAForming a standard model library if any element u₀Is e.g. U, has

Then consider u₀Relative membership toA _i。

The maximum membership principle is a method of fuzzy model identification. Model identification is used to identify what category a particular object belongs to. There are two basic facets in model identification: a plurality of standard models are known in advance to form a model library; the object to be identified. In the fuzzy comprehensive evaluation, the final result vector is processed, and when the final evaluation is further made, the method mainly used is the maximum membership principle.

First, a set of factors that determine psychological risk is required:

U＝{U₁,U₂,U₃,U₄,U₅,U₆}

U₁: extreme psychological tendency U₂: tendency to hurt person U₃: tendency to high energy consumption

U₄: tendency of leaving work and reporting recovery U₅: negative energy spread U₆: glass bottle

U₁＝{U₁₁,U₁₂,U₁₃…U_1i}i＝[1,14]

U₂＝{U₂₁,U₂₂,U₂₃…U_2j}j＝[1,13]

U₃＝{U₃₁,U₃₂,U₃₃…U_3k}k＝[1,8]

U₄＝{U₄₁,U₄₂,U₄₃…U_4l}l＝[1,12]

U₅＝{U₅₁,U₅₂,U₅₃…U_5m}m＝[1,10]

U₆＝{U₆₁,U₆₂,U₆₃…U_6n}n＝[1,11]

U₁₁: depression U₁₂: desperation U₁₃: self-attack … … U_1i: extreme psychological thoughts

U₂₁: angry U₂₂: enemy U₂₃: impulse … … U_2j: eccentric beam

U₃₁: anxiety U₃₂: buckling tolerance U₃₃: mood control … … U_3k: pressure of

U₄₁: enalousu₄₂: reporting U₄₃: interpersonal sensitivity … … U_4l: self-centering

U₅₁: complaint U₅₂: anxiety U₅₃: due to mode … … U_5l: depression (depression)

U₆₁: vulnerability U₆₂: negative life event U₆₃: mood control … … U_6l: social support

Then, a weight for each mental risk profile needs to be determined, i.e. a set of weights is determined.

And correcting the result by adopting a Delphi method, so that the opinions of experts tend to be consistent.

The essence of the Delphi method is that information which cannot be quantified and has large ambiguity is processed in a consultation mode by using expert knowledge, experience, wisdom and the like, and is gradually corrected.

The basic steps of the Delphi method are as follows. Selecting 10 experts having actual working experience and deeper theoretical maintenance in the field of the specialty. And secondly, relevant data of the psychological risk index system and a unified index weight determination rule are sent to selected experts, and the experts are asked to independently determine each group of weight values. And thirdly, processing the recovery result. And (4) sorting and analyzing the weights fed back by the experts, and counting the average value and standard deviation of each weight to require all the experts to re-determine each weight on a new basis. And fourthly, repeating the third step until the standard deviation is less than or equal to the preset standard epsilon equal to 0.1, and taking the average value of the estimated values of the weights at the time as the weight result.

Next, a set of judgments, i.e., a ranking of psychological risks, needs to be determined.

In order to improve the discrimination of judgment, the system adopts four-level judgment gears, namely:

V＝(V₁,V₂,V₃,V₄)

V₁: difference V₂: middle V₃: good V₄: superior food

After the decision set is determined, a decision matrix can be generated.

Let the ith factor U in the factor set U_iJudging the jth element V in the judgment set V_jDegree of membership of r_ijThe ith factor u_iThe fuzzy set corresponding to the judged result is represented as:

R_i＝{r_i1,r_i2,r_i3…r_in}

combining the membership degrees of the evaluation sets to form an evaluation matrix as follows:

after obtaining the judgment matrix and the weight set, carrying out a fuzzy transformation to obtainB＝AAnd multiplying the multiple by the multiple, selecting the score of the item with the highest psychological risk of the tester according to the maximum membership rule, and evaluating the risk grade.

Compared with the prior art, the invention has the beneficial effects that:

1. the invention constructs the psychological portrait of high-risk personnel by collecting the individual psychological characteristic information of people who have had risk behaviors, such as extreme psychologists, injured people and the like. According to the fitting condition of the psychological images of the testers and the high-risk personnel, the high-risk personnel and the non-risk personnel are distinguished by adopting a fuzzy comprehensive judgment method, and the real high-risk population is found more comprehensively and specifically in a targeted manner.

2. Because the evaluation result of the invention is read and used by HR, managers or teachers in colleges and universities of enterprises, the evaluation result of the testers is classified, and when the testers are closer to a certain mental risk portrait, the characteristics, risk points, countermeasures and management suggestions of the portrait are mainly described. In this way, even a person without the psychological knowledge base can quickly understand the evaluation result, and in addition, the suggestion can be provided more specifically.

3. Different from the traditional psychological health assessment, the invention not only examines the current psychological problems of individuals, such as depression, anxiety and the like, detects people with the current psychological problems and distinguishes the people from normal people. In addition, the invention also inspects the individuals who may be at the edge of the psychological problem due to the problems of the physical function such as characters, coping ways, psychological elasticity and the like, namely screens the individuals who may have the psychological health problem in the future and prevents the individuals from suffering from the psychological health problem.

4. The invention can measure and check the individual psychological problems, and also measure and check the individual character, the corresponding mode, the psycho-elasticity, the defense mechanism, the social support and the like, thereby providing the etiology analysis reference for the clinical treatment of the psychologist and the psychotherapist in the later period, and being capable of comprehensively and specifically analyzing the reason of the individual psychological health problems.

5. Although the evaluation item of the invention refers to the general evaluation tools at home and abroad, the expression of the item is not only modified, but also psychological experts and users who are not in psychological specialties are asked to evaluate the understandability and the readability of the item, thereby ensuring the quality of language and characters and reducing the deviation of words, grammar and expression habits. In addition, the homemade normals are built for the newly compiled scale, foreign normals are not directly used, and the quality of the normals is guaranteed.

6. In view of the greater concern of enterprises or colleges, the mental health problems of their employees or students may bring about risks to the enterprises or the schools, and cause what kind of loss. The system not only reports the mental health condition of the testee, but also provides an explanation of risks brought to the enterprise or the school by the overall mental condition of the testee for the enterprise or the school, so that the system is named as an intelligent risk person mental image identification system based on FMT characteristics.

7. Problems that are not solved by traditional mathematics are explained by the benefit of fuzzy mathematics. Fuzzy mathematics is to accurately depict the fuzzy extension of things, and comprehensive evaluation is to make a general evaluation that reasonably integrates attributes or factors of things with various attributes or things of which the overall quality is influenced by various factors. Therefore, the method for fuzzy comprehensive evaluation is feasible and a better way for evaluating the psychological risk of people.

8. In order to avoid that the quality of individual mental health conditions changes along with the normal change of reference, each type of mental risk is marked with corresponding risk groups, namely, the extreme mental tendency mental risk is not marked with the extreme psychology

Therefore, the invention can give a more accurate mental health condition.

Drawings

FIG. 1 is a flow chart for making a psychological risk screening scale of an intelligent risk person identification psychological image system based on FMT characteristics.

FIG. 2 is a flow chart of fuzzy comprehensive evaluation in an FMT feature based intelligent risk person psychological image recognition system.

FIG. 3 is a UML diagram of a technical system in an intelligent risk person identification psychographic system based on FMT features.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.

Referring to fig. 1 to 3, in the embodiment of the present invention, the system for intelligently identifying a psychological image of a risk person based on FMT features includes the following steps:

s1: feature screening, key feature collection and risk feature screening;

s2: high risk personnel psychology portrait model drawing and evaluation;

s3: compiling a scale, measuring and collecting data;

s4: data analysis, determining an image model through Binarylogicic regression;

s5: and obtaining the final risk grade of the tester according to a fuzzy comprehensive judgment method.

Collecting key features in S1, inquiring 10 enterprise high-level managers in different industries for 30-45 minutes by adopting a semi-structured interview mode, knowing employee risk events which occur inside the enterprise, such as extreme psychology, hurt, deliberate destruction and other extreme events of the employee, and some risk behaviors which the enterprise managers worry about the employees to implement, sorting to obtain 56 risk employee key features, screening the risk features in S1, evaluating the key features by adopting a voting mode, inviting 8-bit-qualified HR to vote for each key feature, and reserving more than 60% of items selected by the qualified HR and related to the key features of the risk employees. Thereby selecting 34 risk features to initially form 6 kinds of mental risk employee character pictures, extreme mental tendency psychological pictures, wounded tendency psychological pictures, and departure reviewer psychological pictures,

s2 high-risk person psychological picture model sketching mainly adopts a document analysis method to read and learn related documents at home and abroad on different psychological risk pictures, learn and master the risk models of different figure pictures, and through the combing and summarizing of the existing documents and academic achievements, 5 dimensions of external stimulus, individual emotional state, personality, cognition and psychological elasticity received by an individual are adopted to comprehensively sketch 6 types of psychological risk figure pictures, the psychological picture model evaluation of the high-risk person in S2 comprises three aspects of staff social function condition screening, clinical symptom checking and psychological cause analysis, staff social function condition screening covers interpersonal quality, action efficiency, emotional condition, positive feeling and activity range, and the interpersonal quality evaluation comprises whether interpersonal relationship is stable and harmonious; the performance efficiency evaluation comprises the functional operation efficiency of memory, reaction, learning and the like; the emotional condition assessment comprises whether the tested person can maintain a stable and mild emotional state; positive sensory evaluation includes whether an individual has a sense of well-being, a sense of achievement, etc.; the activity range investigation comprises whether an individual has enough abundant social interaction and entertainment activities, the clinical symptom checking mainly analyzes the psychological problems of the employee, such as depression, hostility-attack, paranoia and the like, so that the psychological problems are determined through the clinical symptom checking scale, the psychological cause analysis mainly comprises stress events, defense mechanisms, social support and the like, and the psychological causes of the employee are analyzed.

Compiling scales in S3, compiling initial scales according to dimension composition of different types of Psychological pictures, and simultaneously compiling an original scale containing 450 subjects by referring to a mental health symptom self-rating scale (SCL-90), an extreme Psychological tendency scale in the internal Psychological model, a Barratt impulse scale, a pressure perception scale (PSS) compiled by doctor Cohen, and the like, wherein the measuring and collecting data in S3 are carried out in a fixed-casting, random-casting and wide-range casting mode, wherein 6 parts of accurate casting are given to typical portrait people, for example, the part scale for measuring the extreme Psychological tendency is given to testers with extreme Psychological behaviors but not in the extreme Psychological behaviors, the part scale for measuring the tendency of injured people is given to testers with the excessive Psychological behaviors, and the like, wherein the part scale for measuring the tendency of injured people is given to normal people at random, and the wide-range casting is mainly given to an APP system platform, the obtained product is free for measurement of users, and by data cleaning, invalid questionnaires are removed, and the total number of recovered valid questionnaires is 16805.

The data analysis in the S4 includes performing reliability and validity analysis, wherein the reliability is an index reflecting the stability and identity of a tester, the stability and identity of the scale are tested by adopting a 'Krenbach alpha coefficient' index, the validity is an index reflecting the authenticity and accuracy of the measurement to what degree the characteristic to be measured is measured by a test, the collected data is subjected to exploratory factor analysis by adopting SPSS20.0, the validity of the programmed scale is tested, and inappropriate questions are deleted through factor analysis so as to obtain good structural validity.

An image model is determined in S4, data of testers who do scales with different risk types are divided into two types, namely two types of extreme psychology and non-extreme psychology, two types of wounded people and non-wounded people and the like, independent variables which really cause the individuals to take the behaviors of extreme psychology, wounded people, post-escape and the like are found out in a binary logistic regression mode, statistical analysis is carried out by adopting a logarithmic linear model, and when two classification variables of the logarithmic linear model are taken as dependent variables, the logarithmic linear regression model becomes a Binarogicistic regression model. In the following, the derivation and solution of the BinaryLogistic model is described in detail by taking an extreme psychological tendency model as an example,

assuming N, N ∈ N, individual takes extreme psychological behavior y_nDenotes y_n1 represents an individual taking extreme psychological behaviour, whereas y_n0, assuming in theory that there is one continuous reaction variable y^* _nRepresents y_nThe probability of occurrence, which ranges from negative infinity to positive infinity, causes the traveler n to take extreme psychological actions if the value of this variable exceeds a critical point m, for example, m is 0, and then:

when y is^* _n>0 time y_n＝1

When y is^* _nY at ≦ 0_n＝0

If it is assumed that the amount of strain y is in the opposite direction^* _nWith independent variable selection of influencing factor, x_niThere is a linear relationship between (i ═ 1,2 …, m), i.e.

y^* _n＝β₀+β₁x_n1+…+β_mx_nm+ε_j 1.1

Wherein: epsilon_jIs an error term, obeys Logistic distribution; beta is a₀Is the intercept commonly referred to as the constant term; beta is a_i(i ═ 1,2 …, m) is x_niThe conditional probability of the nth individual taking extreme psychological behavior is given by the formula 1.1 as the partial regression coefficient of (i ═ 1,2 …, m):

P(y_n＝1/x_ni)＝P[(β₀+β₁x_n1+…+β_mx_nm+ε_j)>0] 1.2

the calculation results in:

wherein P (y)_n＝1/x_ni) Is a non-linear function composed of explanatory variables (influencing factors), which can be converted into a linear function.

First, the probability that an extreme psychological behaviour is selected by an individual n is defined as:

the ratio of the probability of occurrence of the extreme psychological behaviour being selected to not being selected is then:

this ratio is called the occurrence ratio of events, abbreviated as Odds advantage, and it can be seen from equation 1.5 that the larger p, the larger Odds; the smaller p, the smaller odds. In order to measure the degree of influence of an independent variable influence factor on a dependent variable by adopting extreme psychological behaviors, the oddstrato abbreviation OR is defined, which is called the odds ratio for short, and the following formula is used:

the meaning of the method is that the independent variable x is under the condition that the influence factors of the selection behavior of other independent variables are unchanged_niBy changing one unit, the corresponding OR of the dependent variable changes exp (. beta.)_i). We performed sensitivity analysis on the individual selection of the influencing factors of extreme psychological behaviors using OR.

Since the binaryogic regression is a nonlinear model, the parameters are solved using maximum likelihood estimation. Assuming a population consisting of N individual selection activities, Y1, … YN, from which N cases were randomly drawn as samples, observations are labeled Y1, … YN. Let p_i＝p(y_i＝1|x_i) For a given influencing factor x_iUnder conditions such that result y is obtained_iWith a conditional probability of 1, then y_iThe conditional probability of 0 is p (y)_i＝0|x_i)＝1-p_iThen, the probability of obtaining an observation is:

wherein y is_iSince each observation is independent of each other, their joint distribution can be expressed as the product of the marginal distributions, as shown by the following equation:

equation 1.8 refers to the likelihood function of whether n individuals will be extremely psychological or not, where:

to find the parameter estimate that maximizes the value of this likelihood function, for L^*A logarithmic transformation is performed to obtain the following function:

to estimate ln (L)^*) The overall parameter beta at which the maximum value is reached₀…β_mThe partial derivatives are calculated separately and then made equal to 0, given the following equation:

…

as can be seen from the above formula, there are m +1 independent variables in the equation set, and there are m +1 simultaneous equations to estimate their values, since the above are all nonlinear functions, the solution by manual computation is very difficult, and the solution is performed by using the sps 20.0.

Through a binary logistic regression mode, 6 regression equations are established to judge whether a tester belongs to a certain type of risk portrait.

Giving a certain weight to each measurement dimension mainly according to an equation derived by regression, automatically adding a dimension value related to a certain image by a system background of a tester after the tester finishes testing, bringing the value of each dimension into equations of different images, calculating risk values corresponding to the images of different categories, wherein the closer the risk values are to 100 points, the more possible behaviors of people represented by the images are taken, taking extreme psychological tendency as an example, if the extreme psychological tendency of the tester is

In the fuzzy comprehensive judgment method in S5, because the psychology of people is often uncertain, the individual psychological risk level often involves multiple factors or multiple indexes, at this time, the object is required to be comprehensively evaluated according to the multiple factors, but the object cannot be evaluated only under the condition of a certain factor, therefore, the comprehensive psychological risk is mainly judged in a fuzzy comprehensive judgment mode, wherein the judgment means that the object is judged to be good or bad according to given conditions; comprehensive meaning means that the evaluation condition contains multiple factors or multiple indexes; the fuzzy is to use fuzzy mathematics operation method to process, the fuzzy comprehensive evaluation can effectively synthesize different types of data, including non-normally distributed data, which has stronger adaptability, and finally obtain the evaluation of the testee.

Inherent logic of fuzzy comprehensive evaluation:

the fuzzy comprehensive evaluation is a very effective multi-factor decision method for comprehensively evaluating things influenced by various factors, and is characterized in that the evaluation result is not absolutely positive or negative, but is represented by a fuzzy set.

The degree of membership or membership function is the basis of fuzzy mathematics and fuzzy systems. The concept of "membership" is introduced to describe the intermediate transition of differences, which is an approximation of the ambiguity by the accuracy, mathematically defined as:

for fuzzy set A on domain of discourse U, a mapping from U to [0,1] is specified:

μ _A :U→[0,1]

u→μ _A (u∈[0,1])

wherein u is _A Is thatAMembership function of u _A (u) is a pair of uADegree of membership. When u is _A When (u) is 1, u is ∈AWhen u is _A When (u) is 0, the reaction is carried out,

when u is _A When (u) is 0 or 1, the membership function is a characteristic function degenerated into a common set, and the same applies toADegenerates to a normal set.

Fuzzy sets are completely characterized by membership functions, which are established for this purpose, in fuzzy mathematics, where a number between 0 and 1 is required to reflect the degree to which an element belongs to a fuzzy set. Whether the affiliation function can be correctly determined is the key to whether the fuzzy set theory can be effectively applied. If the fuzzy set is defined in the real number domain, the membership functions of the fuzzy set are called fuzzy distributions. We mainly use assignment membership functions, so called assignment methods, which apply some form of fuzzy distribution based on the nature of the problem and then determine the parameters contained in the distribution based on the measured data. The fuzzy distribution type used by the people is larger, the larger fuzzy distribution is suitable for describing the fuzzy phenomenon of the larger bias direction such as large, hot, old and color, and the like, and the general form of the membership function is

Where a is a constant and f (x) is a non-decreasing function.

With fuzzy subsets on the set domain U ₁A, ₂A,… _nAForming a standard model library if any element u₀Is e.g. U, has

Then consider u₀Relative membership toA _i。

The maximum membership principle is a method of fuzzy model identification. Model identification is used to identify what category a particular object belongs to. There are two basic facets in model identification: a plurality of standard models are known in advance to form a model library; the object to be identified. In the fuzzy comprehensive evaluation, the final result vector is processed, and when the final evaluation is further made, the method mainly used is the maximum membership principle.

First, a set of factors that determine psychological risk is required:

U＝{U₁,U₂,U₃,U₄,U₅,U₆}

U₁: extreme psychological tendency U₂: tendency to hurt person U₃: tendency to high energy consumption

U₄: reversal inclination reporting after leaving workTo U₅: negative energy spread U₆: glass bottle

U₁＝{U₁₁,U₁₂,U₁₃…U_1i}i＝[1,14]

U₂＝{U₂₁,U₂₂,U₂₃…U_2j}j＝[1,13]

U₃＝{U₃₁,U₃₂,U₃₃…U_3k}k＝[1,8]

U₄＝{U₄₁,U₄₂,U₄₃…U_4l}l＝[1,12]

U₅＝{U₅₁,U₅₂,U₅₃…U_5m}m＝[1,10]

U₆＝{U₆₁,U₆₂,U₆₃…U_6n}n＝[1,11]

U₁₁: depression U₁₂: desperation U₁₃: self-attack … … U_1i: extreme psychological thoughts

U₂₁: angry U₂₂: enemy U₂₃: impulse … … U_2j: eccentric beam

U₃₁: anxiety U₃₂: buckling tolerance U₃₃: mood control … … U_3k: pressure of

U₄₁: enalousu₄₂: reporting U₄₃: interpersonal sensitivity … … U_4l: self-centering

U₅₁: complaint U₅₂: anxiety U₅₃: due to mode … … U_5l: depression (depression)

U₆₁: vulnerability U₆₂: negative life event U₆₃: mood control … … U_6l: social support

Then, a weight for each mental risk profile needs to be determined, i.e. a set of weights is determined.

And correcting the result by adopting a Delphi method, so that the opinions of experts tend to be consistent.

The essence of the Delphi method is that information which cannot be quantified and has large ambiguity is processed in a consultation mode by using expert knowledge, experience, wisdom and the like, and is gradually corrected.

The basic steps of the Delphi method are as follows. Selecting 10 experts having actual working experience and deeper theoretical maintenance in the field of the specialty. And secondly, relevant data of the psychological risk index system and a unified index weight determination rule are sent to selected experts, and the experts are asked to independently determine each group of weight values. And thirdly, processing the recovery result. And (4) sorting and analyzing the weights fed back by the experts, and counting the average value and standard deviation of each weight to require all the experts to re-determine each weight on a new basis. And fourthly, repeating the third step until the standard deviation is less than or equal to the preset standard epsilon equal to 0.1, and taking the average value of the estimated values of the weights at the time as the weight result.

Next, a set of judgments, i.e., a ranking of psychological risks, needs to be determined.

In order to improve the discrimination of judgment, the system adopts four-level judgment gears, namely:

V＝(V₁,V₂,V₃,V₄)

V₁: difference V₂: middle V₃: good V₄: superior food

After the decision set is determined, a decision matrix can be generated.

Let the ith factor U in the factor set U_iJudging the jth element V in the judgment set V_jDegree of membership of r_ijThe ith factor u_iThe fuzzy set corresponding to the judged result is represented as:

R_i＝{r_i1,r_i2,r_i3…r_in}

combining the membership degrees of the evaluation sets to form an evaluation matrix as follows:

after obtaining the judgment matrix and the weight set, carrying out a fuzzy transformation to obtainB＝AAnd multiplying the multiple by the multiple, selecting the score of the item with the highest psychological risk of the tester according to the maximum membership rule, and evaluating the risk grade.

Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that various changes in the embodiments and modifications of the invention can be made, and equivalents may be substituted for elements thereof without departing from the scope of the invention.

Claims (7)

1. Based on FMT characteristic intelligence discernment risk personnel psychology image system, its characterized in that: the method comprises the following steps:

s1: feature screening, key feature collection and risk feature screening;

s2: high risk personnel psychology portrait model drawing and evaluation;

s3: compiling a scale, measuring and collecting data;

s4: data analysis, determining an image model through Binarylogicic regression;

s5: and obtaining the final risk grade of the tester according to a fuzzy comprehensive judgment method.

2. The system for intelligently identifying psychographic images of at risk persons based on FMT features of claim 1, wherein: the key features are collected in the S1, 10 enterprise high-level managers in different industries are queried for 30-45 minutes in a semi-structured interview mode, employee risk events which occur inside the enterprises and are considered, such as extreme psychology, hurt, deliberate damage and other extreme events of the employees and some risk behaviors which are worried about to be implemented by the employees are known by enterprise managers, 56 risk employee key features are obtained through sorting, risk feature screening is conducted in the S1, the key features are evaluated in a voting mode, 8-bit qualified HR is invited to vote for each key feature, and entries which are selected by more than 60% of qualified HR and are related to the key features of the employees are reserved. Therefore, 34 risk characteristics are selected, and 6 types of psychological risk staff figure images, an extreme psychological tendency psychological image, a wounded tendency psychological image, a leave recovery psychological image, a high energy consumption tendency psychological image, a negative energy propagation psychological image and a glass bottle image psychological image are preliminarily formed.

3. The system for intelligently identifying psychographic images of at risk persons based on FMT features of claim 1, wherein: the psychological portrait model sketching method for the high-risk personnel in the S2 is characterized in that a literature analysis method is mainly adopted to read and learn related literatures at home and abroad for different psychological risk portraits, the risk models of different character portraits are learned and mastered, 5-dimension external stimulus, individual emotional state, personality, cognition and psychological elasticity are received by people from individuals through combing and summarizing existing literatures and academic achievements, 6-class psychological risk character portraits are sketched comprehensively, the psychological image model assessment for the high-risk personnel in the S2 comprises three levels of staff social function condition screening, clinical symptom checking and psychological cause analysis, staff social function condition screening covers interpersonal quality, working efficiency, emotional condition, positive feeling and activity range, and the interpersonal quality assessment comprises whether interpersonal relationship is stable and harmonious; the performance efficiency evaluation comprises the functional operation efficiency of memory, reaction, learning and the like; the emotional condition assessment includes whether the subject can maintain a stable, mild emotional state; positive sensory evaluation includes whether an individual has a sense of well-being, a sense of achievement, etc.; the activity range investigation comprises whether an individual has enough abundant social interaction and entertainment activities, the clinical symptom checking mainly analyzes the psychological problems of the employee, such as depression, hostility-attack, paranoia and the like, so that the psychological problems are determined according to the clinical symptom checking scale, and the psychological cause analysis mainly comprises stress events, defense mechanisms, social support and the like, so that the psychological causes of the employee are analyzed.

4. The system for intelligently identifying psychographic images of at risk persons based on FMT features of claim 1, wherein: compiling a scale in S3, compiling an initial scale according to dimension composition of different types of Psychological pictures, and simultaneously compiling an original scale containing 450 subjects by referring to a mental health symptom self-rating scale (SCL-90), an extreme Psychological tendency scale in the internal Psychological model, a Barratt impulse scale, a pressure perception scale (PSS) compiled by Cohen doctor and the like, wherein the measuring and collecting of data are carried out in S3 by adopting a fixed-throw mode, a random-throw mode and a large-range throw mode, wherein 6 parts of accurate throw is given to typical portrait characters, for example, the part of measuring the extreme Psychological tendency is given to a tester with extreme Psychological behaviors but not extreme Psychological behaviors, the part of measuring the human injury tendency is given to the tester with excessive human injury behaviors and the like, wherein the random-throw mode is given to general normal people at random, the large-range throw is mainly given to an APP platform, the obtained product is free for measurement of users, and by data cleaning, invalid questionnaires are removed, and the total number of recovered valid questionnaires is 16805.

5. The system for intelligently identifying psychographic images of at risk persons based on FMT features of claim 1, wherein: the data analysis in the S4 comprises the steps of carrying out reliability and validity analysis, wherein the reliability is an index reflecting the stability and identity of a tester, the stability and identity of the scale are detected by adopting a 'Krenbach alpha coefficient' index, the validity is an index reflecting the authenticity and accuracy of the measurement to what degree the characteristic to be measured is measured by one test, the collected data is subjected to exploratory factor analysis by adopting SPSS20.0, the validity of the programmed scale is detected, and an improper question is deleted through factor analysis so as to obtain good structural validity.

6. The system for intelligently identifying psychographic images of at risk persons based on FMT features of claim 1, wherein: the portrait model is determined in the S4, the data of testers who do scales with different risk types are divided into two types, namely two types of extreme psychology and non-extreme psychology, two types of wounded people and non-wounded people and the like, the independent variables which really cause the individuals to take the behaviors of extreme psychology, wounded people, post-quit and the like are found out by adopting a binary logistic regression mode, the statistical analysis is carried out by adopting a logarithmic linear model, and when the two classified variables are taken as dependent variables in the logarithmic linear model, the logarithmic linear regression model becomes the Binarogicistic regression model. The derivation and solution of the BinaryLogistic model are described in detail below by taking an extreme psychological tendency model as an example,

assuming N, N ∈ N, individual takes extreme psychological behavior y_nDenotes y_n1 represents an individual taking extreme psychological behaviour, whereas y_n0, assuming in theory that there is one continuous reaction variable y^* _nRepresents y_nThe probability of occurrence, which ranges from negative infinity to positive infinity, causes the traveler n to take extreme psychological actions if the value of this variable exceeds a critical point m, for example, m is 0, and then:

when y is^* _n>0 time y_n＝1

When y is^* _nY at ≦ 0_n＝0

If it is assumed that the amount of strain y is in the opposite direction^* _nWith independent variable selection of influencing factor, x_niThere is a linear relationship between (i ═ 1,2 …, m), i.e.

y^* _n＝β₀+β₁x_n1+…+β_mx_nm+ε_j 1.1

Wherein: epsilon_jIs an error term, obeys Logistic distribution; beta is a₀Is the intercept commonly referred to as the constant term; beta is a_i(i ═ 1,2 …, m) is x_niThe conditional probability of the nth individual taking extreme psychological behavior is given by the formula 1.1 as the partial regression coefficient of (i ═ 1,2 …, m):

P(y_n＝1/x_ni)＝P[(β₀+β₁x_n1+…+β_mx_nm+ε_j)>0] 1.2

the calculation results in:

wherein P (y)_n＝1/x_ni) Is a function of an explanatory variable (influence factor)Prime) that can be transformed into a linear function.

First, the probability that an extreme psychological behaviour is selected by an individual n is defined as:

the ratio of the probability of occurrence of the extreme psychological behaviour being selected to not being selected is then:

this ratio is called the occurrence ratio of events, abbreviated as Odds advantage, and it can be seen from equation 1.5 that the larger p, the larger Odds; the smaller p, the smaller odds. In order to measure the degree of influence of an independent variable influence factor on a dependent variable by adopting extreme psychological behaviors, the oddstrato abbreviation OR is defined, which is called the odds ratio for short, and the following formula is used:

the meaning of the method is that the independent variable x is under the condition that the influence factors of the selection behavior of other independent variables are unchanged_niBy changing one unit, the corresponding OR of the dependent variable changes exp (. beta.)_i). We used OR to perform sensitivity analysis on the influence factors of individual selection end psychological behaviors.

Since the binaryogic regression is a nonlinear model, the parameters are solved using maximum likelihood estimation. Assuming a population consisting of N individual selection activities, Y1, … YN, from which N cases were randomly drawn as samples, observations are labeled Y1, … YN. Let p_i＝p(y_i＝1|x_i) For a given influencing factor x_iUnder conditions such that result y is obtained_iWith a conditional probability of 1, then y_iThe conditional probability of 0 is p (y)_i＝0|x_i)＝1-p_iThen, the probability of obtaining an observation is:

wherein y is_iSince each observation is independent of each other, their joint distribution can be expressed as the product of the marginal distributions, as shown by the following equation:

equation 1.8 refers to the likelihood function of whether n individuals will be extremely psychological or not, where:

to find the parameter estimate that maximizes the value of this likelihood function, for L^*A logarithmic transformation is performed to obtain the following function:

to estimate ln (L)^*) The overall parameter beta at which the maximum value is reached₀…β_mThe partial derivatives are calculated separately and then made equal to 0, given the following equation:

…

as can be seen from the above formula, there are m +1 independent variables in the equation set, and there are m +1 simultaneous equations to estimate their values, since the above are all non-linear functions, the solution by manual computation is very difficult, and the solution is performed by using the sps 20.0.

Through a binary logistic regression mode, 6 regression equations are established to judge whether a tester belongs to a certain type of risk portrait.

The method mainly comprises the steps of giving a certain weight to each measurement dimension according to an equation obtained by regression, enabling a tester to automatically add a dimension value related to a certain image in a system background after the tester completes testing, bringing the value of each dimension into equations of different images, calculating risk values corresponding to different types of images, and indicating that the more the risk values are close to 100 points, the more likely the behavior of a person represented by the image is to be taken.

7. The system for intelligently identifying psychographic images of at risk persons based on FMT features of claim 1, wherein: in the fuzzy comprehensive judgment method in S5, because the human psychology is often uncertain, and the individual psychological risk level often involves multiple factors or multiple indicators, at this time, it is required to make comprehensive evaluation on the object according to the multiple factors, but cannot evaluate the object only from the condition of a certain factor, so the judgment of the comprehensive psychological risk mainly adopts a fuzzy comprehensive judgment method, where the judgment means to compare and judge the quality and quality of the object according to given conditions; comprehensive means that the evaluation condition comprises a plurality of factors or a plurality of indexes; the fuzzy is to use fuzzy mathematics operation method to process, the fuzzy comprehensive evaluation can effectively synthesize different types of data, including non-normally distributed data, which has stronger adaptability, and finally obtain the evaluation of the testee.

Inherent logic of fuzzy comprehensive evaluation:

the fuzzy comprehensive evaluation is a very effective multi-factor decision method for comprehensively evaluating things influenced by various factors, and is characterized in that the evaluation result is not absolutely positive or negative, but is represented by a fuzzy set.

The degree of membership or membership function is the basis of fuzzy mathematics and fuzzy systems. The concept of "membership" is introduced to describe the intermediate transition of differences, which is an approximation of the ambiguity by the accuracy, mathematically defined as:

for fuzzy set A on domain of discourse U, a mapping from U to [0,1] is specified:

μ _A :U→[0,1]

u→μ _A (u∈[0,1])

wherein u is _A Is thatAMembership function of u _A (u) is a pair of uADegree of membership. When u is _A When (u) is 1, u is ∈AWhen u is _A When (u) is 0, the reaction is carried out,

when u is _A When (u) is 0 or 1, the membership function is a characteristic function degenerated into a common set, and the same applies toAWill degenerate into a normal set.

Fuzzy sets are completely characterized by membership functions, which are established for this purpose, in fuzzy mathematics, where a number between 0 and 1 is required to reflect the degree to which an element belongs to a fuzzy set. Whether the membership function can be correctly determined is the key of whether the fuzzy set theory can be effectively applied. If the fuzzy set is defined in the real number domain, the membership functions of the fuzzy set are called fuzzy distributions. We mainly use assignment membership functions, so-called assignment methods, which apply some form of fuzzy distribution ready for use depending on the nature of the problem and then determine the parameters contained in the distribution from the measurement data. The fuzzy distribution type used by us is a bigger type, the bigger type fuzzy distribution is suitable for describing the fuzzy phenomenon of the bigger direction such as big, hot, old and color 'dark', and the common form of the membership function is

Where a is a constant and f (x) is a non-decreasing function.

With fuzzy subsets on the set domain U ₁A, ₂A,… _nAForming a standard model library if any element u₀Is e.g. U, has

Then consider u₀Relative membership toA _i。

The maximum membership principle is a method of fuzzy model identification. Model identification is used to identify what category a particular object belongs to. There are two basic facets in model identification: a plurality of standard models are known in advance to form a model library; the object to be identified. In the fuzzy comprehensive evaluation, the method mainly used is the maximum membership principle when the final result vector is processed and the final evaluation is further made.

First, a set of factors that determine psychological risk is required:

U＝{U₁,U₂,U₃,U₄,U₅,U₆}

U₁: extreme psychological tendency U₂: tendency to hurt person U₃: tendency to high energy consumption

U₄: tendency of leaving work and reporting recovery U₅: negative energy spread U₆: glass bottle

U₁＝{U₁₁,U₁₂,U₁₃…U_1i}i＝[1,14]

U₂＝{U₂₁,U₂₂,U₂₃…U_2j}j＝[1,13]

U₃＝{U₃₁,U₃₂,U₃₃…U_3k}k＝[1,8]

U₄＝{U₄₁,U₄₂,U₄₃…U_4l}l＝[1,12]

U₅＝{U₅₁,U₅₂,U₅₃…U_5m}m＝[1,10]

U₆＝{U₆₁,U₆₂,U₆₃…U_6n}n＝[1,11]

U₁₁: depression U₁₂: desperation U₁₃: self-attack … … U_1i: extreme psychological thoughts

U₂₁: angry U₂₂: enemy U₂₃: impulse … … U_2j: eccentric beam

U₃₁: anxiety U₃₂: buckling tolerance U₃₃: mood control … … U_3k: pressure of

U₄₁: enalousu₄₂: reporting U₄₃: interpersonal sensitivity … … U_4l: self-centering

U₅₁: complaint U₅₂: anxiety U₅₃: due to mode … … U_5l: depression (depression)

U₆₁: vulnerability U₆₂: negative life event U₆₃: mood control … … U_6l: social support

Then, a weight for each mental risk profile needs to be determined, i.e. a set of weights is determined.

And correcting the result by adopting a Delphi method, so that the opinions of experts tend to be consistent.

The essence of the Delphi method is that information which cannot be quantified and has large ambiguity is processed in a consultation mode by using expert knowledge, experience, wisdom and the like, and is gradually corrected.

The basic steps of the Delphi method are as follows. Selecting 10 experts having actual working experience and deeper theoretical maintenance in the field of the specialty. And secondly, relevant data of the psychological risk index system and a unified index weight determination rule are sent to selected experts, and the experts are asked to independently determine the weight values of all groups. And thirdly, processing the recovery result. And (4) sorting and analyzing the weights fed back by the experts, and counting the average value and standard deviation of each weight to require all the experts to re-determine each weight on a new basis. And fourthly, repeating the third step until the standard deviation is less than or equal to the preset standard epsilon equal to 0.1, and taking the average value of the estimated values of the weights at the time as the weight result.

Next, a set of judgments, i.e., a ranking of psychological risks, needs to be determined.

In order to improve the discrimination of judgment, the system adopts four-level judgment gears, namely:

V＝(V₁,V₂,V₃,V₄)

V₁: difference V₂: middle V₃: good V₄: superior food

After the decision set is determined, a decision matrix can be generated.

Let the ith factor U in the factor set U_iJudging the jth element V in the judgment set V_jDegree of membership of r_ijThe ith factor u_iThe fuzzy set corresponding to the judged result is represented as:

R_i＝{r_i1,r_i2,r_i3…r_in}

combining the membership degrees of the evaluation sets to form an evaluation matrix as follows:

after obtaining the judgment matrix and the weight set, carrying out a fuzzy transformation to obtainB＝AAnd multiplying the multiple by the multiple, selecting the score of the item with the highest psychological risk of the testee according to the maximum membership rule, and evaluating the risk grade.

Images