The invention provides a power harmonic signal analysis method and power harmonic signal analysis equipment based on a Bayes method. Different from the existing harmonic solution method in the frequency domain, a harmonic signal analysis mathematic model is established by using the time domain sparse characteristic of a power system complex harmonic signal and by adopting a Bayes model, so that the amplitude and the frequency of the signal can be acquired through recursive solution on the problem according to the expected maximum algorithm. The power harmonic signal analysis method comprises the following steps: setting a Bayes estimation initial value and a convergence parameter; acquiring a harmonic frequency analysis interval, discretizing the harmonic frequency analysis interval into analysis subintervals with the set number and forming a harmonic measuring matrix; acquiring a sampling signal of a to-be-analyzed signal and establishing a probability density function of the harmonicsignal according to a Bayes formula; performing iterative calculation on the probability density function by the expected maximal algorithm and estimating to acquire the frequency and the amplitude of a harmonic component in the to-be-analyzed signal.