CN105976126A - Machine tool manufacturing energy consumption analysis method based on principal component analysis method - Google Patents

Abstract

The invention discloses a machine tool manufacturing energy consumption analysis method based on a principal component analysis method. At first, a machine tool manufacturing energy consumption evaluation index system is established, on this basis, the principal component analysis method is adopted to transform the original related variables into independent principal components, and furthermore, the independent principal components are evaluated, and the duplicate information that is fed back during the evaluation due to correlation between the indexes is eliminated. The weight of the principal component analysis is mainly information weight, namely, a kind of weight that is determined by how much the resolved information of an object to be evaluated included in the evaluation index, and thus being objective. According to the method, the machine tool enterprise energy consumption condition can be scientifically and reasonably evaluated and analyzed, and by means of the calculation of energy efficiency, the machine tool enterprises can improve the energy efficiency on purpose and push forward green manufacturing.

Description

Machine Manufacture energy consumption analysis method based on PCA

Technical field

The present invention relates to Machine Manufacture process efficiency Optimized-control Technique field, a kind of machine based on PCA Bed manufactures energy consumption analysis method.

Background technology

In place of manufacturing energy-saving and emission-reduction are always the key of national energy-saving emission reduction tasks, the machinery with lathe as canonical system adds Work system has the feature huge with energy expenditure that have a large capacity and a wide range, and statistics shows that China's lathe efficiency utilization rate is universal Less than 30%, there is the biggest energy-saving potential.Reduced energy programme by Legislation and manufacturing industry to be affected, high energy efficiency lathe Product becomes the direction of enterprise development.It addition, machine tool system work process to have energy source more, energy consumption link and energy consumption group Part is complicated, and has the features such as dynamically change is big, Random Effect factor is many, and the problem in science related to is many, therefore studies meaning Justice is the most important.

Machine Manufacture process energy consumption analysis includes the assay of energy expenditure state and energy consuming process, and basis at this On assay to energy efficiency.Set up effective efficiency appraisement system, use the analysis means of feasible science, to lathe Manufacturing shop energy conservation has important realistic meaning, provides the foundation of quantitative analysis and decision for the enterprise effective use energy, Contribute to enterprise and improve production technology and enterprise management mode, and then improve energy utilization rate.The energy consumption in evaluation analysis workshop is The basis of workshop managing power consumption, it is impossible to the Energy Consumption Evaluation result obtaining workshop is one of obstruction implementing energy optimization.

Summary of the invention

It is an object of the invention to for above-mentioned energy consumption analysis problem, propose a kind of Machine Manufacture energy consumption based on PCA Analysis method, multi objective problem is converted into less aggregative indicator by the method, makes the question simplification of complexity, the party simultaneously Method provides objective weight, makes up the shortcoming that in other evaluation methodologys, anthropic factor impact is stronger.

The technical solution adopted in the present invention is, described Machine Manufacture energy consumption analysis method based on PCA, comprises Following steps:

Step one, set up Machine Manufacture energy consumption evaluation indexes system and evaluation indice；

Step 2, index reverse in assessment indicator system is carried out forward process so that it is unification；Forward index refers to commenting Valency result produces the index of actively impact, and reverse index refers to produce evaluation result the index of negative influence, to reverse index Carry out forward process and i.e. take its inverse；

Step 3, application classification scoring carry out quantification process to Machine Manufacture energy consumption system qualitative index data；

Step 4, set up Machine Manufacture energy consumption evaluation indexes matrix of variables；

The matrix of variables of step 4 is standardized processing by step 5, application Z-Score conversion；

Step 6, establish main constituent number and evaluation model according to principal component contributor rate size；

Step 7, on the basis of evaluation model, use the information method of weighting to factor weighted method sue for peace, establish evaluate comprehensive letter Number, obtains sample evaluation result and to sample energy consumption level ranking.

Concrete, the method setting up Machine Manufacture energy consumption evaluation indexes matrix of variables described in step 4 is: set the lathe system chosen The number of samples making process energy consumption is n, and each sample packages contains p evaluation index variable, then can set up energy consumption evaluation indexes Matrix of variables X:

$X={\left(X_{ij}\right)}_{n\×p}==\left[\begin{array}{cccc}{X}_{11}& X_{12}& ...& X_{1p}\\ X_{21}& X_{22}& ...& X_{2p}\\ ...& ...& ...& ...\\ X_{n1}& X_{n2}& ...& X_{np}\end{array}\right]$

Wherein X_ijThe energy consumption evaluation indexes variate-value of expression i-th sample jth item, i=1,2,3 ..., n；J=1,2,3 ..., p.

Matrix of variables X is standardized processing by the conversion of step 5 Z-Score, obtains new matrix Z after conversion_n×p, change public affairs Formula is as follows:

$Z_{ij}=\frac{X_{ij}-{\stackrel{\‾}{X}}_j}{S_j}$

${\stackrel{\‾}{X}}_j=\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}{X}_{ij}$

$S_j^2=\frac{1}{n-1}\underset{i=1}{\overset{n}{\Σ}}{(X_{ij}-{\stackrel{\‾}{X}}_j)}^2$

Wherein Z_ijRepresenting matrix Z_n×pThe value of the i-th row j row.

Step 6 achieves lathe energy consumption evaluation indexes main constituent replace, by originally relevant original lathe energy consumption index Variable, is converted to separate main constituent, then these main constituents is carried out overall merit, eliminate due to phase between index The duplicate message closed and feed back when evaluation analysis, decreases the dimension of overall merit simultaneously；

First the correlation matrix R of normalized data:

$R={\[r_{ij}\]}_{P\×P}=\frac{Z_{n\×p}^{\′}{Z}_{n\×p}}{n-1}$

Wherein Z '_n×pIt is Z_n×pTransposed matrix；

Then the characteristic root of correlation matrix R, characteristic vector and main constituent are asked；The characteristic root λ of R₁≥λ₂≥λ₃...≥λ_pRepresent Each main constituent is the size of role on evaluation object, and to ask for corresponding characteristic vector be L_j；

Main constituent is tried to achieve according to standardized index variable:

$F_k=\underset{j=1}{\overset{p}{\Σ}}{L}_j{X}_j$

(j=1,2 ..., p；K=1,2 ..., p)

Wherein F_kFor kth main constituent, as new comprehensive energy consumption index.

When determining main constituent number k, claimFor the contribution rate of kth main constituent, claimFor front k main constituent Accumulative variance contribution ratio, chooses main constituent number according to accumulative variance contribution ratio α >=85%:

$\mathrm{\α}=\frac{\underset{i=1}{\overset{k}{\Σ}}{\mathrm{\λ}}_i}{\underset{i=1}{\overset{p}{\Σ}}{\mathrm{\λ}}_i}.$

Step 7 is weighted summation to k main constituent, tries to achieve final evaluation function；The weight of principal component analysis is mainly information Weight, a kind of weight that the number i.e. comprising object resolution information to be evaluated from evaluation index determines, information weight determines Principle is: about greatly, then this index resolution information is the most being respectively evaluated the deviation of numerical value between object for an index, and its flexible strategy also should The biggest；Otherwise, deviation is the fewest, and index weight the most just should be the least, and evaluation function is as follows:

$\begin{array}{c}F=\frac{{\mathrm{\λ}}_1}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_1+\frac{{\mathrm{\λ}}_2}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_2\\ +\mathrm{...}+\frac{{\mathrm{\λ}}_k}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_k\end{array}.$

The present invention initially sets up Machine Manufacture energy consumption evaluation indexes system, uses PCA on this basis, by master Componential analysis, originally relevant original variable is transformed to separate main constituent, and then is evaluated these main constituents, Eliminate the duplicate message fed back owing to being correlated with between index when evaluating.The weight of principal component analysis is mainly information weight, A kind of weight that the number i.e. comprising object resolution information to be evaluated from evaluation index determines, has objectivity.This method can With scientific and reasonable evaluation analysis Machine Tool Enterprises energy consumption, by calculating efficiency, Machine Tool Enterprises can carry out energy pointedly Effect is improved, and advances green manufacturing.

Accompanying drawing explanation

Fig. 1 is the energy consumption analysis procedure chart of the present invention.

Fig. 2 is the energy consumption evaluation indexes system of the present invention.

Fig. 3 is energy consumption evaluation indexes computation rule.

Specific embodiments

As it is shown in figure 1, the invention mainly comprises following seven steps: step one, set up Machine Manufacture energy consumption evaluation indexes body System and evaluation indice；Step 2, index reverse in assessment indicator system is carried out forward process so that it is unification；Step Three, application classification scoring carries out quantification process to Machine Manufacture energy consumption system qualitative index data；Step 4, set up machine Bed manufactures energy consumption evaluation indexes matrix of variables；Original variable matrix is standardized processing by step 5, application Z-Score conversion； Step 6, establish main constituent number and evaluation model according to principal component contributor rate size；Step 7, on the basis of evaluation model On, the weighted sum to main constituent, establish further and evaluate comprehensive function, obtain sample evaluation result and to sample energy water consumption Flat raft name.

In step one: evaluation index choose needs follow can survey, reliably, three principles of adequacy, herein from efficiency economy, Product energy consumption, energy efficiency of equipment, flow of task efficiency four levels, choose ten evaluation indexes and set up energy consumption evaluation indexes system, And traditional lathe Energy Consumption Evaluation system have ignored production technology efficiency and resources of production scheduling efficiency.The lathe energy of the present invention Consumption assessment indicator system hierarchical structure is as in figure 2 it is shown, its evaluation indice includes: ten thousand yuan of product energy consumption C1, ten thousand yuan of value addeds Energy consumption C2, unit product comprehensive energy consumption C3, unit product amount of energy saving C4, product can horizontal C5, machine tool efficiencies C6, energy transfer efficiency C7, energy processing conversion efficiency C8, production technology efficiency C9, resources of production scheduling efficiency C10.

Step 2 achieves the unification of lathe energy consumption evaluation indexes data type, due to forward index existing in evaluation index again Having reverse index, forward index to refer to produce evaluation result the index of actively impact, reverse index refers to produce evaluation result The index of raw negative influence, wherein reverse index has ten thousand yuan of product energy consumption C1, ten thousand yuan of value added energy consumptions C2, unit product comprehensive Energy consumption C3, remaining is forward index.Needed before sample is carried out overall merit, the type of evaluation index is made unification Process, otherwise cannot pass judgment on composite evaluation function qualitatively.It is good and bad in order to sample can be passed judgment on according to composite evaluation function value, Reverse for evaluation index index C forward is processed, obtains result C^*, finally make index unification.

$C^{*}=\frac{1}{C}$

Qualitative index is converted into quantitative target by step 3, uses classification scoring, give a score value to every grade in the present invention, When being " excellent, good, in, poor " such as index grade, then score value is respectively " 4,3,2,1 ".

Step 4 sets up Machine Manufacture energy consumption evaluation indexes matrix of variables, chooses the sample of n lathe process energy consumption, each sample Originally comprise p=10 evaluation index variable, then can set up energy consumption evaluation indexes matrix of variables X:

$X={\left(X_{ij}\right)}_{n\×p}==\left[\begin{array}{cccc}{X}_{11}& X_{12}& ...& X_{1p}\\ X_{21}& X_{22}& ...& X_{2p}\\ ...& ...& ...& ...\\ X_{n1}& X_{n2}& ...& X_{np}\end{array}\right]$

Wherein X_ijRepresent the energy consumption evaluation indexes variate-value of i-th sample jth item.

Step 5 achieves the standardization of lathe energy consumption evaluation indexes matrix of variables, each relevant due to during analyzing The dimension of energy consumption index, the difference of magnitude, comparability is poor, therefore carries out original variable matrix X with Z-Score conversion Standardization so that it is have good comparability, obtains new matrix z after conversion_n×10, change formula as follows:

$Z_{ij}=\frac{X_{ij}-{\stackrel{\‾}{X}}_j}{S_j}$

${\stackrel{\‾}{X}}_j=\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}{X}_{ij}$

$S_j^2=\frac{1}{n-1}\underset{i=1}{\overset{n}{\Σ}}{(X_{ij}-{\stackrel{\‾}{X}}_j)}^2$

(i=1,2 ..., n；J=1,2 ..., 10)

Step 6 achieves lathe energy consumption evaluation indexes main constituent replace, by originally relevant original lathe energy consumption index Variable, is converted to separate main constituent, then these main constituents is carried out overall merit, eliminate due to phase between index The duplicate message closed and feed back when evaluation analysis, decreases the dimension of overall merit simultaneously.First normalized data Correlation matrix R:

$R={\[r_{ij}\]}_{10\×10}=\frac{{Z^{\′}}_{n\×10}{Z}_{n\×10}}{n-1}$

Then the characteristic root of correlation matrix R, characteristic vector and main constituent are asked.The characteristic root λ of R₁≥λ₂≥λ₃...≥λ₁₀, table Show each main constituent size of role on evaluation object, and to ask for corresponding characteristic vector be L_j.According to standardization Index variable try to achieve main constituent:

$F_k=\underset{j}{\overset{p}{\Σ}}{L}_j{X}_j$

(j=1,2 ..., 10；K=1,2 ..., 10)

Wherein F_kFor kth main constituent, as new comprehensive energy consumption index.

When determining main constituent number, meet following principle: data variation maximum principle, least square principle and group put similarity Change minimum principle, make principal component analysis information dropout minimum, maximum with former Variable Similarity.

Wherein, claimFor the contribution rate of kth main constituent, claimAccumulative variance contribution ratio for front k main constituent. Size according to accumulative variance contribution ratio determines main constituent number, chooses main constituent number according to contribution rate of accumulative total α >=85% herein:

$\mathrm{\α}=\frac{\underset{i=1}{\overset{k}{\Σ}}{\mathrm{\λ}}_i}{\underset{i=1}{\overset{p}{\Σ}}{\mathrm{\λ}}_i}$

Wherein k is the number of main constituent.

Step 7, on the basis of establishing evaluation model, is weighted summation, tries to achieve final evaluation function k main constituent.Main The weight of component analysis is mainly information weight, and the number i.e. comprising object resolution information to be evaluated from evaluation index determines A kind of weight, compensate for the weak point of subjective weight.The principle that information weight determines is: a certain index is respectively being evaluated object Between the deviation of numerical value about big, then this index resolution information is the most, and its flexible strategy also should be the biggest；Otherwise, deviation is the fewest, and index is weighed Number the most just should be the least, and therefore the advantage obtaining other evaluation methodologys relatively of this weight is exactly objectivity.Evaluation function is as follows Shown in:

$\begin{array}{c}F=\frac{{\mathrm{\λ}}_1}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_1+\frac{{\mathrm{\λ}}_2}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_2\\ +\mathrm{...}+\frac{{\mathrm{\λ}}_k}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_k\end{array}$

Below in conjunction with embodiment, the invention will be further described.

Choose the workshop of a Machine Manufacture enterprise herein as analyzing object, gather the sample index of this workshop different times Data value, uses the index computation rule of Fig. 3, draws appraisement system desired value, and use PCA to carry out energy consumption Analyze.As illustrated in chart 1.

Table 1 sample index data value

The quantification of qualitative index processes.

Product in Energy Consumption Evaluation system can level be qualitative index, needs to process its quantification, product energy level pair The numerical value answered is respectively 3,3,4,3,4.

Index unification processes.

Reverse index in Energy Consumption Evaluation system has: ten thousand yuan of product comprehensive energy consumptions, ten thousand yuan of value added energy consumptions and unit product are comprehensive Energy consumption, uses variable data unification to process formula and processes, the such as table 2 of the result after conversion:

Table 2 unification result

It will be seen that the order of abovementioned steps two and step 3 can be exchanged.

Set up initializaing variable matrix.

According to 5 samples and 10 evaluation index values of each sample, set up original variable matrix X:

$X=\left[\begin{array}{cccccccccc}0.130& 0.195& 0.017& 2.6& 3& 0.38& 0.45& 0.48& 0.46& 0.44\\ 0.144& 0.185& 0.019& 3.5& 3& 0.32& 0.48& 0.52& 0.36& 0.48\\ 0.159& 0.163& 0.020& 3.0& 4& 0.34& 0.44& 0.46& 0.43& 0.40\\ 0.133& 0.215& 0.013& 2.5& 3& 0.41& 0.43& 0.50& 0.50& 0.45\\ 0.139& 0.211& 0.021& 2.9& 4& 0.30& 0.50& 0.47& 0.47& 0.45\end{array}\right]$

Standardization.

To original variable standardization, result is as follows:

$Z=\left[\begin{array}{cccccccccc}1.32& -0.16& 0.21& -0.76& -0.73& 0.67& -0.34& -0.25& 0.30& -0.14\\ -0.17& 0.34& -0.37& 1.52& -0.73& -0.67& 0.69& 1.41& -1.58& 1.25\\ -1.46& 1.54& -0.69& 0.25& 1.10& -0.22& -0.69& -1.08& -0.26& -1.53\\ 0.03& -0.92& 1.65& -1.02& -0.73& 1.34& -1.03& 0.58& 1.05& 0.21\\ 0.29& -0.80& -0.79& 0& 1.10& -1.12& 1.37& -0.66& 0.49& 0.21\end{array}\right]$

Solve main constituent and composite evaluation function.

Try to achieve λ₁=4.361, λ₂=3.232, λ₃=1.866, λ₄=0.541, remaining characteristic root is too small, trends towards 0, may be used herein To ignore.First principal component contributor rate is:

According to evaluation index selection rule, owing to first principal component contribution rate is less than 80%, so needing to contribute according to main constituent The size of rate, to front several main constituent linear weighted function integrated treatments.Try to achieve the first two main constituent and add up variance contribution ratio and be:

$\frac{\underset{i=1}{\overset{2}{\Σ}}{\mathrm{\λ}}_i}{\underset{i=1}{\overset{10}{\Σ}}{\mathrm{\λ}}_i}\×100\%=76\%$

Try to achieve first three main constituent and add up variance contribution ratio and be:

$\frac{\underset{i=1}{\overset{3}{\Σ}}{\mathrm{\λ}}_i}{\underset{i=1}{\overset{10}{\Σ}}{\mathrm{\λ}}_i}\×100\%=94\%\≥85\%$

It is more than 85%, so the main constituent number established in evaluation methodology is three owing to first three main constituent adds up variance contribution ratio Individual, i.e. k=3, and the information that first main constituent comprises is most, effect maximum in Machine Manufacture process energy consumption analysis, Corresponding characteristic vector u₁, u₂As shown in chart 3:

Table 3 characteristic vector

According to formula winner composition expression formula:

F₁=0.264Z₁-0.228Z₂-0.376Z₃+...+0.278Z₉+ 0.039Z₁₀

F₂=-0.304Z₁+0.205Z₂+0.461Z₃+...+0.134Z₉+0.305Z₁₀

F₃=0.450Z₁-0.027Z₂+0.176Z₃+...-0.249Z₉-0.247Z₁₀

According to main constituent information weight, it is weighted summation, obtains composite evaluation function as follows:

$F=\frac{{\mathrm{\λ}}_1}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+{\mathrm{\λ}}_3}{F}_1+\frac{{\mathrm{\λ}}_2}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+{\mathrm{\λ}}_3}{F}_2+\frac{{\mathrm{\λ}}_3}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+{\mathrm{\λ}}_3}{F}_3$

Comprehensive evaluation result.

Sample data is substituted into main constituent expression formula, is tried to achieve the final appraisal results of each sample by Energy Consumption Evaluation function, and To sample evaluation result ranking.Evaluation result is as shown in table 4:

Table 4 main constituent evaluation result

By above-mentioned Machine Manufacture energy consumption analysis method based on PCA, draw following result: sample 1, sample 2 Too low with the Machine Manufacture energy consumption level during this section of sample 3, effective Optimized Measures need to be taked, improve efficiency comprehensive utilization Rate；The closely-related energy consumption evaluation indexes of first principal component have ten thousand yuan of product energy consumptions, ten thousand yuan of value added energy consumptions, with can level and Resources of production scheduling efficiency, has all exceeded 0.5, and the effect played on energy consumption level is bigger, can examine by emphasis in optimization method Consider these energy consumption indexs.

It is contemplated that a kind of method providing efficiency overall merit for Machine Manufacture energy consumption analysis technical field, this area non-general Described technical scheme can be modified on the basis of reading description of the invention by logical technical staff, or to wherein Portion of techniques feature carries out equivalent and these are revised or replace, and does not make the essence of appropriate technical solution depart from the present invention each Implement the spirit and scope of technical scheme.

Claims (6)

1. Machine Manufacture energy consumption analysis method based on PCA, is characterized in that, comprise the following steps:

Step one, set up Machine Manufacture energy consumption evaluation indexes system and evaluation indice；

Step 2, index reverse in assessment indicator system is carried out forward process so that it is unification；Forward index refers to evaluation Result produces the index of actively impact, and reverse index refers to produce evaluation result the index of negative influence, carries out reverse index Forward processes and i.e. takes its inverse；

Step 3, application classification scoring carry out quantification process to Machine Manufacture energy consumption system qualitative index data；

Step 4, set up Machine Manufacture energy consumption evaluation indexes matrix of variables；

The matrix of variables of step 4 is standardized processing by step 5, application Z-Score conversion；

Step 6, establish main constituent number and evaluation model according to principal component contributor rate size；

Step 7, on the basis of evaluation model, use the information method of weighting to factor weighted method sue for peace, establish evaluate comprehensive function, Obtain sample evaluation result and to sample energy consumption level ranking.

2. Machine Manufacture energy consumption analysis method based on PCA as claimed in claim 1, is characterized in that, step 4 Described set up Machine Manufacture energy consumption evaluation indexes matrix of variables method be: set the number of samples of the Machine Manufacture process energy consumption chosen For n, each sample packages contains p evaluation index variable, then can set up energy consumption evaluation indexes matrix of variables X:

$X={\left(X_{ij}\right)}_{n\×p}==\left[\begin{array}{cccc}{X}_{11}& X_{12}& ...& X_{1p}\\ X_{21}& X_{22}& ...& X_{2p}\\ ...& ...& ...& ...\\ X_{n1}& X_{n2}& ...& X_{np}\end{array}\right]$

Wherein X_ijThe energy consumption evaluation indexes variate-value of expression i-th sample jth item, i=1,2,3 ..., n；J=1,2,3 ..., p.

3. Machine Manufacture energy consumption analysis method based on PCA as claimed in claim 2, is characterized in that, step 5 It is standardized processing to matrix of variables X with Z-Score conversion, after conversion, obtains new matrix Z_n×p, change formula as follows:

$Z_{ij}=\frac{X_{ij}-{\stackrel{\‾}{X}}_j}{S_j}$

${\stackrel{\‾}{X}}_j=\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}{X}_{ij}$

$S_j^2=\frac{1}{n-1}\underset{i=1}{\overset{n}{\Σ}}{(X_{ij}-{\stackrel{\‾}{X}}_j)}^2$

Wherein Z_ijRepresenting matrix Z_n×pThe value of the i-th row j row.

4. Machine Manufacture energy consumption analysis method based on PCA as claimed in claim 1, is characterized in that, step 6 In achieve lathe energy consumption evaluation indexes main constituent and replace, by originally relevant original lathe energy consumption index variable, be converted to Then these main constituents are carried out overall merit by separate main constituent, eliminate due to relevant between index and in evaluation analysis Time feedback duplicate message, decrease the dimension of overall merit simultaneously；

First the correlation matrix R of normalized data:

$R={\[r_{ij}\]}_{P\×P}=\frac{Z_{n\×p}^{\′}{Z}_{n\×p}}{n-1}$

Wherein Z '_n×pIt is Z_n×pTransposed matrix；

Then the characteristic root of correlation matrix R, characteristic vector and main constituent are asked；The characteristic root λ of R₁≥λ₂≥λ₃...≥λ_pRepresent each Individual main constituent is the size of role on evaluation object, and to ask for corresponding characteristic vector be L_j；

Main constituent is tried to achieve according to standardized index variable:

$\begin{array}{c}{F}_k=\underset{j=1}{\overset{p}{\Σ}}{L}_j{X}_j\\ (j=1,2,\mathrm{...},p;k=1,2,\mathrm{...},p)\end{array}$

Wherein F_kFor kth main constituent, as new comprehensive energy consumption index.

5. Machine Manufacture energy consumption analysis method based on PCA as claimed in claim 4, is characterized in that, determine master During composition number k, claimFor the contribution rate of kth main constituent, claimAccumulative variance contribution for front k main constituent Rate, chooses main constituent number according to accumulative variance contribution ratio α >=85%:

$\mathrm{\α}=\frac{\underset{i=1}{\overset{k}{\Σ}}{\mathrm{\λ}}_i}{\underset{i=1}{\overset{p}{\Σ}}{\mathrm{\λ}}_i}.$

6. Machine Manufacture energy consumption analysis method based on PCA as claimed in claim 4, is characterized in that, step 7 K main constituent is weighted summation, tries to achieve final evaluation function；The weight of principal component analysis is mainly information weight, i.e. from A kind of weight that the number that evaluation index comprises object resolution information to be evaluated determines, the principle that information weight determines is: one About greatly, then this index resolution information is the most, and its flexible strategy also should be the biggest being respectively evaluated the deviation of numerical value between object for index；Otherwise, Deviation is the fewest, and index weight the most just should be the least, and evaluation function is as follows:

$\begin{array}{c}F=\frac{{\mathrm{\λ}}_1}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_1+\frac{{\mathrm{\λ}}_2}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_2\\ +\mathrm{...}+\frac{{\mathrm{\λ}}_k}{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2+...+{\mathrm{\λ}}_k}{F}_k\end{array}.$