CN105354622A - Fuzzy comprehensive evaluation based enterprise production management evaluation method - Google Patents

Abstract

The invention discloses a fuzzy comprehensive evaluation based enterprise production management evaluation method. The method comprises the following steps of: S1: establishing single factor evaluation standards and comprehensive evaluation standards of each production link; S2: performing preliminary evaluation on a single production link; S3: performing fuzzy comprehensive evaluation calculation according to a preliminary evaluation result; and S4: analyzing a comprehensive evaluation calculation result according to the comprehensive evaluation standards to obtain a fuzzy comprehensive evaluation result of each evaluation object. According to the method, production process data are calculated and processed through a constructed fuzzy comprehensive evaluation model, current production health conditions of enterprises are evaluated to finally obtain beneficial information of accident pre-warning, fault diagnosis, optimization suggestion and the like, and reference bases are provided for comprehensive management of deciders in aspects of enterprise optimization, device improvement, personnel allocation and the like.

Description

Based on the enterprise production management evaluation method of fuzzy comprehensive evoluation

Technical field

The present invention relates to enterprise production management assessment technique field, particularly one based on the enterprise production management evaluation method of fuzzy comprehensive evoluation.

Background technology

The enterprise of type of production, generally all relate to the details modules such as production of machinery processing, electricity consumption water, emphasis people/property management reason, huge and the production system of complexity of these module compositions one, and the information such as " production machine state data ", " energy consumption data ", " power quality data ", " emphasis people, thing DYNAMIC DISTRIBUTION data ", the operation conditions of this system (enterprise) can be represented in real time, therefore, by framework Internet of Things infosystem, gather the process data (i.e. production process data) of these links, realize the real-time monitoring to enterprise operation state.And each link data are associated, and in conjunction with enterprise's basic data and business models, utilize corresponding data processing method to diagnose out many potential problems or illness in enterprise, provide early warning or Optimizing Suggestions in time.Realize above-mentioned functions and depend on a set of comprehensive infosystem, this system is referred to as again " enterprise's production health control " platform.

But, the production process data of large enterprise is of a great variety, data volume is large, real-time change, each other very difficult analyzes corresponding relation, at present, in the infosystem of most enterprises application, there are some diagnostic functions, generally there are 2 features: 1) be all carry out data correlation by visible, simple logical relation to reach a conclusion, 2) towards object generally for specialized equipment or mini-system, and reliable solution can only be lacked towards large-scale complicated system.

In view of this, the present inventor proposes one by model of fuzzy synthetic evaluation to being calculated by production process data and processing, being aided with enterprise's basic data, business models and setting up criterion of acceptability realizes the excavation of a large amount of production process data, association and analysis, draw the method for enterprise's health indicator diagnostic result.

Summary of the invention

The object of the present invention is to provide one based on the enterprise production management evaluation method of fuzzy comprehensive evoluationproduction process data calculated by the model of fuzzy synthetic evaluation built and processes, the current production health status of enterprise is passed judgment on, finally draw the advantageous information such as accident early warning, fault diagnosis, Optimizing Suggestions, for decision maker in optimization of enterprises, to improve equipment and integrated management in personnel depaly etc. provides reference frame.

To achieve these goals, the solution that the present invention adopts is:

based on the enterprise production management evaluation method of fuzzy comprehensive evoluation, the steps include:

S1: single factor evaluation standard and the comprehensive value model of formulating each production link, this single factor evaluation standard and this comprehensive value model are all according to industry standard or performance of enterprises index or mathematical model formulation;

S2: the preliminary assessment carrying out single production link, its step comprises:

S21: the status data gathering each link in production run, calculates respectively the status data of each link, obtain status data preliminary processing results;

S22: preliminary processing results and single factor evaluation standard are compared, draws the preliminary assessment result of each production link, and this preliminary assessment result represents with Boolean quantity, i.e. 0 (up to standard) or 1 (not up to standard);

S3: the calculating carrying out fuzzy overall evaluation according to preliminary assessment result, its step comprises:

S31: setting evaluation object, determines the factor of effect appraise object, sets up the set of factors U of effect appraise object, wherein U={y ₁, y ₂..., y _n, n is the factor quantity affecting this evaluation object, y ₁-y _ngeneration respectively table Sthe preliminary assessment result of each link in 2;

S32: set up the evaluation collection V be made up of comprehensive evaluation result of calculation B, wherein V={x ₁, x ₂..., x _m.

S33: set up single factor evaluation matrix R, wherein

$R=\left[\begin{array}{cccc}{r}_{11}& r_{12}& ...& r_{1m}\\ r_{21}& r_{22}& ...& r_{2m}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ r_{n1}& r_{n2}& ...& r_{nm}\end{array}\right],$ Each element r in matrix R _ijrepresent factor y respectively _ibe under the jurisdiction of and evaluate x _jdegree;

S34: according to set of factors U, adopts statistic law to set up effect appraise object weight set of factors A, wherein A={a ₁, a ₂..., a _m, m is the weight quantity affecting this evaluation object;

S35: obtain comprehensive evaluation result of calculation B, wherein B=AR;

S4: according to comprehensive value model, the result of calculation B of comprehensive evaluation and evaluation collection V is analyzed, draw the fuzzy overall evaluation conclusion of each evaluation object.

In described step S2, in production run, the status data of each link comprises managing power consumption data, production equipment status data, power quality data and people/thing orientation management data.

In described step S3, evaluation object comprises inputoutput, KPI, power supply quality, energy consumption and fault.

Described step S21 is specially: be unit at a certain time interval, get the real-time dynamic data in a period of time, be weighted accumulation calculating to time series data, obtain the status data result in some cycles.

In described step S34, the statistic law setting up effect appraise object weight set of factors A is optional with following several:

Expert statistics method: establish set of factors U={u ₁, u ₂..., u _n, by k expert separately independentprovide each factor u _i(i=1,2 ..., weight n), gets the mean value of each factor weight as its weight $a_i=\frac{1}{k}\underset{j=1}{\overset{k}{\Σ}}{a}_{ij}(i=1,2,\mathrm{...},n),$ Namely $A=(\frac{1}{k}\underset{j=1}{\overset{k}{\Σ}}{a}_{1j},\frac{1}{k}\underset{j=1}{\overset{k}{\Σ}}{a}_{2j},\mathrm{...},\frac{1}{k}\underset{j=1}{\overset{k}{\Σ}}{a}_{nj});$

Weighted statistical method: as expert number k < 30 people, available weights statistical method calculates weight, according to formula calculate corresponding weight vectors A;

Frequency statistics method: establish set of factors U={u ₁, u ₂..., u _n, when expert number k>=30, according to weight allocation investigation, to each element in set of factors U, everyone independentground proposes oneself to think most suitable weight, according to the weight allocation investigation of reclaiming, to each factor u _i(i=1,2 ..., n) carry out monofactorial weight statistical test, its step is as follows:

S341: to factor u _i(i=1,2 ..., n) at its weight a _ij(j=1,2 ..., k), find out maximal value M _iwith minimum value m _i, namely $M_i=\underset{1\&le;j\&le;k}{\max }\left\{a_{ij}\right\},m_i=\underset{1\&le;j\&le;k}{\min }\left\{a_{ij}\right\};$

S342: suitably choose positive integer p, utilizes formula calculate group distance weight being divided into p group, and weight is divided into p group from small to large;

S343: calculate the frequency and frequency that drop on the interior weight of often group;

S344: according to frequency and frequency distribution situation, the class mean of getting the grouping of maximum frequency place is factor u _iweight a _i(i=1,2 ..., n), obtaining weight is A=(a ₁, a ₂..., a _n).

In described step S35, the operational model of comprehensive evaluation result of calculation B can be selected following several:

Model I: M (∧, ∨)

In Fuzzy comprehensive evaluation, be max-min compose operation, namely use model M (∧, ∨) to calculate B=AR, wherein

Modelⅱ: M (, ∨)

This model adopts two kinds of computings: one is ordinary multiplications computing, with representing; Another kind is maximizing operation, represents with ∨, wherein,

Model III:

This model adopts outside minimizing operation ∧, also adopts ring sum operation also claim bounded and computing, it represents that the upper limit is the summation operation of 1, wherein,

Computing for bounded and, namely

Model IV: M (,+)

This model adopts two kinds of computings: one is ordinary multiplications computing; Another kind be common additive operation+, wherein,

$b_j=\underset{i=1}{\overset{n}{\&Sigma;}}\left(a_i{r}_{ij}\right),(j=1,2,...,m).$

After adopting such scheme, beneficial effect of the present invention is: the core algorithm that the present invention is based on fuzzy overall evaluation, the relevance of each production link data is excavated, the relation that influences each other of the individual each production link made full use of, realize the Comprehensive Evaluation of each evaluation object, comprehensive evaluation result of the present invention directly reflects production health status, for accident early warning, fault diagnosis, Optimizing Suggestions etc. provide reference information, for decision maker in optimization of enterprises, to improve equipment and integrated management in personnel depaly etc. provides reference frame

Below in conjunction with accompanying drawingthe present invention will be further described with embodiment.

Accompanying drawing explanation

fig. 1it is flow process of the present invention letter figure.

Embodiment

as Fig. 1shown in, the present invention discloses based on the enterprise production management evaluation method of fuzzy comprehensive evoluation, the steps include:

S1: single factor evaluation standard and the comprehensive value model of formulating each production link, this single factor evaluation standard and this comprehensive value model are all according to industry standard or performance of enterprises index or mathematical model formulation;

S2: the preliminary assessment carrying out single production link, its step comprises:

S21: the status data gathering each link in production run, calculates respectively the status data of each link, obtain status data preliminary processing results;

S22: preliminary processing results and single factor evaluation standard are compared, draws the preliminary assessment result of each production link, and this preliminary assessment result represents with Boolean quantity, i.e. 0 (up to standard) or 1 (not up to standard);

S3: the calculating carrying out fuzzy overall evaluation according to preliminary assessment result, its step comprises:

S31: setting evaluation object, determines the factor of effect appraise object, sets up the set of factors U of effect appraise object, wherein U={y ₁, y ₂..., y _n, n is the factor quantity affecting this evaluation object, y ₁-y _ngeneration respectively table Sthe preliminary assessment result of each link in 2;

S32: set up the evaluation collection V be made up of comprehensive evaluation result of calculation B, wherein V={x ₁, x ₂..., x _m.

S33: set up single factor evaluation matrix R, wherein

$R=\left[\begin{array}{cccc}{r}_{11}& r_{12}& ...& r_{1m}\\ r_{21}& r_{22}& ...& r_{2m}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ r_{n1}& r_{n2}& ...& r_{nm}\end{array}\right],$ Each element r in matrix R _ijrepresent factor y respectively _ibe under the jurisdiction of and evaluate x _jdegree, R calculates by membership function, or is set by expert analysis mode;

S34: according to set of factors U, adopts statistic law to set up effect appraise object weight set of factors A, wherein A={a ₁, a ₂..., a _m, m is the weight quantity affecting this evaluation object;

S35: obtain comprehensive evaluation result of calculation B, wherein B=AR;

S4: according to comprehensive value model, the result of calculation B of comprehensive evaluation and evaluation collection V is analyzed, draw the fuzzy overall evaluation conclusion of each evaluation object.

In described step S2, in production run, the status data of each link comprises managing power consumption data, production equipment status data, power quality data and people/thing orientation management data.

In described step S3, evaluation object comprises inputoutput, KPI, power supply quality, energy consumption and fault.

Described step S21 is specially: be unit at a certain time interval, get the real-time dynamic data in a period of time, be weighted accumulation calculating to time series data, obtain the status data result in some cycles.

In described step S34, the statistic law setting up effect appraise object weight set of factors A is optional with following several:

Expert statistics method: establish set of factors U={u ₁, u ₂..., u _n, by k expert separately independentprovide each factor u _i(i=1,2 ..., weight n), gets the mean value of each factor weight as its weight $a_i=\frac{1}{k}\underset{j=1}{\overset{k}{\&Sigma;}}{a}_{ij}(i=1,2,\mathrm{...},n),$ Namely $A=(\frac{1}{k}\underset{j=1}{\overset{k}{\&Sigma;}}{a}_{1j},\frac{1}{k}\underset{j=1}{\overset{k}{\&Sigma;}}{a}_{2j},\mathrm{...},\frac{1}{k}\underset{j=1}{\overset{k}{\&Sigma;}}{a}_{nj});$

Weighted statistical method: as expert number k < 30 people, available weights statistical method calculates weight, according to formula calculate corresponding weight vectors A;

Frequency statistics method: establish set of factors U={u ₁, u ₂..., u _n, when expert number k>=30, according to weight allocation investigation, to each element in set of factors U, everyone independentground proposes oneself to think most suitable weight, according to the weight allocation investigation of reclaiming, to each factor u _i(i=1,2 ..., n) carry out monofactorial weight statistical test, its step is as follows:

S341: to factor u _i(i=1,2 ..., n) at its weight a _ij(j=1,2 ..., k), find out maximal value M _iwith minimum value m _i, namely $M_i=\underset{1\&le;j\&le;k}{\max }\left\{a_{ij}\right\},m_i=\underset{1\&le;j\&le;k}{\min }\left\{a_{ij}\right\};$

S342: suitably choose positive integer p, utilizes formula calculate group distance weight being divided into p group, and weight is divided into p group from small to large;

S343: calculate the frequency and frequency that drop on the interior weight of often group;

S344: according to frequency and frequency distribution situation, the class mean of getting the grouping of maximum frequency place is factor u _iweight a _i(i=1,2 ..., n), obtaining weight is A=(a ₁, a ₂..., a _n).

In described step S35, the operational model of comprehensive evaluation result of calculation B can be selected following several:

Model I: M (∧, ∨)

In Fuzzy comprehensive evaluation, be max-min compose operation, namely use model M (∧, ∨) to calculate B=AR, wherein

Thisly to ask method mainly through getting little and getting large two kinds of computings, therefore, claim this kind of model to be M (∧, ∨) model, this method is when factor is many, inevitable very little to the weighted value of each factor, and evaluation result can be caused undesirable;

Modelⅱ: M (, ∨)

This model adopts two kinds of computings: one is ordinary multiplications computing, with representing; Another kind is maximizing operation, represents with ∨, wherein,

Wherein, multiplying a _ir _ijcan not drop-out, and maximizing operation ∨ can drop-out.This model advantage is the significance level reflecting single factor evaluation result preferably;

Model III:

This model adopts outside minimizing operation ∧, also adopts ring sum operation also claim bounded and computing, it represents that the upper limit is the summation operation of 1, wherein,

Computing for bounded and, namely

Because weight allocation meets

0≤a _i≤ 1,0≤r _ij≤ 1, so therefore have

In actual applications, when main factor (factor that weight is maximum) plays a leading role in Comprehensive Evaluation, adopt Model I, modelⅱ, model III can be adopted when Model I loses efficacy;

Model IV: M (,+)

This model adopts two kinds of computings: one is ordinary multiplications computing; Another kind be common additive operation+, wherein,

$b_j=\underset{i=1}{\overset{n}{\&Sigma;}}\left(a_i{r}_{ij}\right),(j=1,2,...,m).$

Model IV considers the impact of all factors, and remains the full detail of single factor evaluation, and this model effect in engineering judgment is good.

As the embodiment of an embody rule, in production run, the status data of each link comprises managing power consumption data, production equipment status data, power quality data and people/thing orientation management data; Evaluation object comprises inputoutput, KPI (KPI Key Performance Indicator), power supply quality, energy consumption and fault.

These are only specific embodiments of the invention, the restriction not to protection scope of the present invention.All equivalent variations done according to the mentality of designing of this case, all fall into the protection domain of this case.

Claims (6)

1., based on the enterprise production management evaluation method of fuzzy comprehensive evoluation, it is characterized in that, step is:

S1: single factor evaluation standard and the comprehensive value model of formulating each production link, this single factor evaluation standard and this comprehensive value model are all according to industry standard or performance of enterprises index or mathematical model formulation;

S2: the preliminary assessment carrying out single production link, its step comprises:

S21: the status data gathering each link in production run, calculates respectively the status data of each link, obtain status data preliminary processing results;

S22: preliminary processing results and single factor evaluation standard are compared, draws the preliminary assessment result of each production link, and this preliminary assessment result represents with Boolean quantity, i.e. 0 (up to standard) or 1 (not up to standard);

S3: the calculating carrying out fuzzy overall evaluation according to preliminary assessment result, its step comprises:

S31: setting evaluation object, determines the factor of effect appraise object, sets up the set of factors U of effect appraise object, wherein U={y ₁, y ₂..., y _n, n is the factor quantity affecting this evaluation object, y ₁-y _ngeneration respectively table Sthe preliminary assessment result of each link in 2;

S32: set up the evaluation collection V be made up of comprehensive evaluation result of calculation B, wherein V={x ₁, x ₂..., x _m.

S33: set up single factor evaluation matrix R, wherein

each element r in matrix R _ijrepresent factor y respectively _ibe under the jurisdiction of and evaluate x _jdegree;

S34: according to set of factors U, adopts statistic law to set up effect appraise object weight set of factors A, wherein A={a ₁, a ₂..., a _m, m is the weight quantity affecting this evaluation object;

S35: obtain comprehensive evaluation result of calculation B, wherein B=AR;

S4: according to comprehensive value model, the result of calculation B of comprehensive evaluation and evaluation collection V is analyzed, draw the fuzzy overall evaluation conclusion of each evaluation object.

2. as claimthe enterprise production management evaluation method based on fuzzy comprehensive evoluation described in 1, it is characterized in that: in described step S2, in production run, the status data of each link comprises managing power consumption data, production equipment status data, power quality data and people/thing orientation management data.

3. as claimthe enterprise production management evaluation method based on fuzzy comprehensive evoluation described in 2, it is characterized in that: in described step S3, evaluation object comprises inputoutput, KPI, power supply quality, energy consumption and fault.

4. as claimthe enterprise production management evaluation method based on fuzzy comprehensive evoluation described in 1, it is characterized in that: described step S21 is specially: be unit at a certain time interval, get the real-time dynamic data in a period of time, accumulation calculating is weighted to time series data, obtains the status data result in some cycles.

5. as claimthe enterprise production management evaluation method based on fuzzy comprehensive evoluation described in 1, is characterized in that: in described step S34, and the statistic law setting up effect appraise object weight set of factors A is optional with following several:

Expert statistics method: establish set of factors U={u ₁, u ₂..., u _n, by k expert separately independentprovide each factor u _i(i=1,2 ..., weight n), gets the mean value of each factor weight as its weight namely

Weighted statistical method: as expert number k < 30 people, available weights statistical method calculates weight, according to formula calculate corresponding weight vectors A;

Frequency statistics method: establish set of factors U={u ₁, u ₂..., u _n, when expert number k>=30, according to weight allocation investigation, to each element in set of factors U, everyone independentground proposes oneself to think most suitable weight, according to the weight allocation investigation of reclaiming, to each factor u _i(i=1,2 ..., n) carry out monofactorial weight statistical test, its step is as follows:

S341: to factor u _i(i=1,2 ..., n) at its weight a _ij(j=1,2 ..., k), find out maximal value M _iwith minimum value m _i, namely

S342: suitably choose positive integer p, utilizes formula calculate group distance weight being divided into p group, and weight is divided into p group from small to large;

S343: calculate the frequency and frequency that drop on the interior weight of often group;

S344: according to frequency and frequency distribution situation, the class mean of getting the grouping of maximum frequency place is factor u _iweight a _i(i=1,2 ..., n), obtaining weight is A=(a ₁, a ₂..., a _n).

6., as claimed in claim 1 based on the enterprise production management evaluation method of fuzzy comprehensive evoluation, it is characterized in that: in described step S35, the operational model of comprehensive evaluation result of calculation B can be selected following several:

Model I: M (∧, ∨)

In Fuzzy comprehensive evaluation, be max-min compose operation, namely use model M (∧, ∨) to calculate B=AR, wherein

Modelⅱ: M (, ∨)

This model adopts two kinds of computings: one is ordinary multiplications computing, with representing; Another kind is maximizing operation, represents with ∨, wherein,

Model III:

This model adopts outside minimizing operation ∧, also adopts ring sum operation also claim bounded and computing, it represents that the upper limit is the summation operation of 1, wherein,

Computing for bounded and, namely

Model IV: M (,+)

This model adopts two kinds of computings: one is ordinary multiplications computing; Another kind be common additive operation+, wherein,