Delcio,

Basta fazer assim:

durner <- as.formula(theta ~
                         w0 * (thetar + ((thetas - thetar)/(1 + (alpha0 * psi)^n0)^(1 - 1/n0))) +
                         (1 - w0) * (thetar + ((thetas - thetar)/(1 + (alpha1 * psi)^n1)^(1 - 1/n1))))

dur <- nlsLM(durner,
             start = list(w0 = 0.6,
                          thetas = 0.5,
                          thetar = 0.25,
                          alpha0 = 0.1,
                          n0 = 2,
                          alpha1 = 0.1,
                          n1 = 1),
             data = dados,
             algorithm = "LM",
             trace = "F",
             control = list(maxiter = 500))
summary(dur)

À disposição.

Walmes.