Summary of the invention
In order to make up the drawback of grey trend decision method in embedded computer performance evaluation, the embedded computer method of evaluating performance of the grey SITUATION DECISION based on AHP is proposed.Idiographic flow is as follows:
Step 1: set up embedded computer Performance Measuring Indicators
The present invention adopts Delphi expert consulting method to choose embedded computer Performance Measuring Indicators from numerous factors that affect embedded computer performance, is shown in Table 1.
Table 1 embedded computer Performance Measuring Indicators
Step 2: the weight of determining embedded computer Performance Evaluating Indexes
The present invention adopts analytical hierarchy process to determine the weight of embedded computer Performance Evaluating Indexes, and concrete steps are as follows:
Step 2.1: compare according to two two indexes relative importances, according to 9 fraction system scorings (in table 2), form judgment matrix A:
Table 2 standards of grading and implication thereof
In formula, a
ji=1/a
ij, a
ii=1.
Step 2.2: calculate importance ranking vector, i.e. weight
(λ
maxI-A)w=0
In formula, λ
maxfor the maximum characteristic root of matrix A, w=[w
1..., w
m] be λ
maxnormalization proper vector, component w
ifor index U
iweight.
Step 2.3: carry out consistency check
In formula, CR is judgment matrix random index; RI is mean random coincident indicator, and its index checks in from table 3 according to judgment matrix exponent number.
Table 3 mean random coincidence indicator
In the time of CR≤0.1, judgment matrix A has consistance, illustrates that weight allocation is suitable; Otherwise, need judgment matrix to adjust, to meet conforming requirement.
Step 3: build GREY SITUATION DECISION model
Event a
ioccur, use countermeasure b
jtackle a
iappearance, b
jwith a
ijust form situation s
ij, i.e. s
ij=(a
i, b
j);
Step 4: measure of merit
Make x
(0)=(x
(0)(1), x
(0)(2) ..., x
(0)(n)), make r (k) represent x
(0)(k) measure of merit transformed value.
Work as x
(0)for maximum value polarity chron, have
r=(r(1),r(2),…,r(n)) (1)
Wherein
;
Work as x
(0)for minimal value polarity chron, have
r=(r(1),r(2),…,r(n)) (2)
Wherein
;
Work as x
(0)for moderate polarity chron, have
r=(r(1),r(2),…,r(n)) (3)
Wherein
, x
0for moderate value
Step 5: synthesis effect measure sequence
Order
for situation s
ijmeasure of merit value under p target, if there is p=1,2 ..., m, claims
for s
ijsynthesis effect measure, and if only if
If have
Claim j
*for tackling a
ibe satisfied with countermeasure.
The present invention has set up the evaluation index system of embedded computer combination property, and GREY SITUATION DECISION algorithm is combined with analytical hierarchy process, traditional GREY SITUATION DECISION algorithm is improved, and be applied in embedded computer synthetic performance evaluation.The method has solved the blindness that GREY SITUATION DECISION is evaluated, and has improved correctness and the validity of embedded computer Performance Evaluation.Exemplary application proves that the method is rationally effective, for the evaluation of embedded computer combination property from now on provides valuable reference.
Embodiment
With reference to Fig. 1, process of the present invention comprises: 1) adopt Delphi expert consulting method to set up embedded computer Performance Measuring Indicators (1-1); 2) and adopt analytical hierarchy process to determine the weight coefficient of evaluation index (1-2); 3) build GREY SITUATION DECISION model (1-3); 4) calculate evaluation index measure of merit (1-4); 5) set up synthesis effect measure sequence (1-5); 6) adopt maximization method to evaluate embedded computer performance (1-6).
In order to further illustrate the validity of this invention, take 3 kinds of embedded computers as example, specifically introduce the application of the method below.A type embedded computer is a kind of emerging embedded computer, calculates advanced person, stable performance; Type B embedded computer is the modified of A type embedded computer, has increased storage capacity; C type embedded computer is a kind of old type embedded computer, and each performance is all poor.The algorithm that application the present invention proposes, carries out comprehensive performance evaluation and odds to the embedded computer of these 3 kinds of models, and concrete steps are as follows:
Step 1: set up hierarchy Model
According to embedded computer performance parameter index and the demand that can adapt to FUTURE ENVIRONMENT development, build the assessment indicator system of three layers, as shown in Figure 2.
Step 2: utilize analytical hierarchy process to determine the weight coefficient of each index
Step 3: determine event, countermeasure situation, target
Event a: the combination property of 3 kinds of embedded computers is carried out to quality sequence;
Situation: s
a=(a, b
a) A type embedded computer, s
b=(a, b
b) Type B embedded computer, s
c=(a, b
c) the embedded calculating of C type;
Target: all indexs shown in Fig. 2.
Step 4: determine target polarity, do measure of merit conversion
First analyze the polarity of each index, recycle corresponding measure of merit transformation for mula index is converted.For example, to evaluation index U
11, U
43, U
51do measure of merit conversion.
Target U
11(dominant frequency)
Target U
11belong to maximum value polarity, be respectively 533MHZ, 1.6GHZ and 200MHZ according to the dominant frequency of A, B, tri-kinds of embedded computers of C, the measure of merit that is obtained this target by formula (1) is
Target U
43(mean time between failures)
Mean time between failures belongs to maximum value polarity, is respectively 50000 hours, 60000 hours and 45000 hours according to the mean time between failures of A, B, tri-kinds of embedded computers of C, obtains target U by formula (1)
43measure of merit be
Target U
51(working temperature)
Working temperature belongs to moderate polarity, be respectively-10 ℃~60 ℃ according to A, B, tri-kinds of embedded computer working temperatures of C,-10 ℃~75 ℃ and-10 ℃~50 ℃, be all-10 ℃ because three kinds of embedded computer working temperatures are minimum, therefore this index can be considered as to maximum value polarity, obtain target U by formula (1)
51measure of merit be
。
In like manner, can obtain the measure of merit of other indexs
(ij represents the designator of index), as shown in table 4.
The measure of merit value of each index of table 4
Step 5: find out and be satisfied with situation
By the weight w of evaluation index
ij, i=1,2 ..., 6, j=1,2 ..., n
iwith table 3, obtain the synthesis effect measure of A, B, tri-kinds of embedded computers of C according to formula (4).Wherein, the synthesis effect measure of A type embedded computer is:
In like manner, solving Type B embedded computer synthesis effect measure is
, the synthesis effect measure of C type embedded computer is
.The synthesis effect measure of comprehensive A, B, tri-kinds of embedded computers of C, as shown in table 5.
Table 5 embedded computer synthesis effect measure evaluation result
This shows that, in A, B, tri-kinds of embedded computers of C, the combination property optimum of Type B embedded computer, is secondly A type embedded computer, is finally C type embedded computer.Assessment result conforms to the performance performance in actual use.