We develop two new algorithms for stabilizing time varying queues delay probabilities by expanding the queue length in terms of Hermite polynomials.  The first algorithm uses the mean and variance of the queue length.  The second algorithm uses the mean, variance, and skewness of the queue length.  We show that the skewness helps in stabilizing the queue length at positive target values, especially when the queue is critically loaded.  We illustrate through several simulation studies the added benefit of using higher moments of the queue length distribution.