www.brydzyk.fora.pl

defgf96tj

Grey Prediction Model of Traffic Accidents

+1) A I-(z \a guitar (_1 (2) + c1 (3))) a 6I ■ l a,[link widoczny dla zalogowanych], ∞ (n) = nf One woman ... -1) + ... ())】 +6 Y Ⅳ = X = L (. (2) z '.' (3) y a】 there a nx + bE = ~ x; E] [:] A [x; E] recorded a B to B [xE] B = (1 (1) + ... (2 )) l Y 『existing one Ba,[link widoczny dla zalogowanych], according to least squares, a d a (BB) By identification parameters can be calculated to be n,[link widoczny dla zalogowanych], b value, prediction model can be determined. In addition, for the prediction calculation convenience,[link widoczny dla zalogowanych], model (1 ) can be written as l; (1 (+1) one. (1) Once) e + Dan (2) \: Accident Gray model test. available residual test and the method of testing in the posterior residual test test model used 2.1 (2) calculation ... (+1) value (i l, 2, ..., n- 1), then the method generates less tired; \Then (+1) and '(g + 1) generations derived the following formula calculated for each year and the actual value of the relative error (points + L). + I (point a l, 2,[link widoczny dla zalogowanych], ..., -1) if e (a 1) greater accuracy of the model is poor. 2.2 posterior posterior deviation of testing in order to test the prediction error e, based on i-EI-year forecast error for the EI-mf an \Observational data for the first i-level; m. For the first i-level forecasts. Posterior difference test are: l according to the size, study bias (forecast error) is small the probability of points, and the forecast error variance with the size of the indicators. Posterior differences are based on test data are: (1) statistics mean a _lm unitary formula: the number of data. (2) Statistical data variance s = 1Σ (Ⅲ a Ge.) 1 (3) the mean prediction error of a formula: n is the number of prediction error data, generally n 0.95, not less than 0.7. According to the size of P and c, the prediction accuracy can be divided into four grades (see Table 1). Table 1 Prediction accuracy class precision grade table top dead C Good resistance to> 095 080 07O <0.50 ≤ 0.70 ≥ 0.653 unqualified example, according to statistics Xi'an City traffic incidents in recent years, the number of (minor incidents are not statistics) in Table 2. Table 21 994 traffic accidents over the years to 1999 in Xi'an, the number of incidents shake a few years l9941997f1998 z'327833903679 the original data as a {2874,3278,3337,3390,3679,3850} can be determined by the method of GM (1,1) prediction model as z \model, its accuracy rating for the level. with the model of equation (3) predicts the number of traffic accidents in 2000 in Xi'an (not including minor accidents) in the 3980 times or so. 4 Conclusion (L) on the basis of accident statistics, prediction model established by the above method, and model accuracy test; (2) if the prediction error is greater or less accuracy class, you must modify the model; (3) that the basis of the residual by the above method EJI build residual GM (1,1), the residuals of the gray model, then the original model equation (2) plus the forecast of the residuals, as revised forecast model can improve prediction accuracy; (4) traffic accident prediction results for the development of effective traffic safety measures to provide a scientific basis.
More articles related to topics：

[link widoczny dla zalogowanych]

[link widoczny dla zalogowanych]

[link widoczny dla zalogowanych]