What governs the Earth's orbital period is its orbital angular momentum and the mass of the Sun. Two events have certainly changed the Earth's orbital period (a) whatever collision formed the Moon and (b) the continuous process of mass loss from the Sun. Further possibilities are: (c) Tidal torques from the Sun have increased the angular momentum of the Earth; (d) Radiation pressure from the Sun changes the orbit; (e) drag from the interplanetary medium slows the Earth.
Of these I think only (a) (in the early lifetime of the Earth) and (b) (again, mainly in the first hundred million years or so, but with some effect afterwards) are important.
Early Collisions
Given that (a) probably happened sometime in the first tens of millions of years and likely did not alter the Earth's angular momentum greatly - it depends on the speed, mass and direction of the impactor and the amount of mass lost from the Earth-Moon system - I will ignore it.
Solar Mass Loss
It seems, from observations of younger solar analogues, that the mass loss from the early Sun was much greater than the modest rate at which it loses mass now via the solar wind (or by solar radiation). A review by Guedel (2007) suggests a mass loss rate over the last 4.5 billion years that increases as $t^{-2.3}$ (with considerable uncertainty on the power law index), where $t$ is time since birth, and suggests an initial solar mass between 1% and 7% larger than it is now.
Conservation of angular momentum a
means that $a \propto M^{-1}$ and Kepler's third law leads to $P \propto M^{-2}$. Therefore the Earth's orbital period was 2-14% shorter in the distant past due to solar mass loss, but has been close to its current value for the last few billion years.
If the solar wind power law time dependence is very steep, then most of the mass loss occurred early, but the total mass loss would have been greater. On the other hand, a lower total mass loss implies a shallower mass loss and the earth spending a longer time in a smaller orbit.
Tidal Torques
The tidal torque exerted by the Sun on the Earth-Sun orbit increases the orbital separation, because the Sun's rotation period is shorter than the Earth's orbital period. The Sun's tidal "bulge" induced by the Earth applies a torque that increases the orbital angular momentum, much like the effect of the Earth on the Moon.
Quantifying this is difficult. The tidal torque on a planet from a the Sun is
$$ T = \frac{3}{2} \frac{k_E}{Q} \frac{GM_{\odot}^{2} R_{E}^{5}}{a^6},$$
where $R_E$ is an Earth radius and $k_E/Q$ is the ratio of the tidal Love number and $Q$ a tidal dissipation factor (see Sasaki et al. (2012).
These lecture notes suggest values of $k_E/Q\sim 0.1$ for the Earth and therefore a tidal torque of $4\times 10^{16}$ Nm. Given that the orbital angular momentum of the Earth is $\sim 3\times 10^{40}$ kgm$^2$s$^{-1}$, then the timescale to change the Earth's angular momentum (and therefore $a$ and $P$) is $>10^{16}$ years and thus this effect is negligible.
Edit:
Further effects that can be ignored are a changing radiation pressure from the Sun and drag from the interplanetary medium.
Radiation Pressure
The outward force on the Earth is roughly the solar flux at the Earth multiplied by $\pi R^2$ and divided by the speed of light. However, since the radiation flux is, like gravity, obeying an inverse square law, this simply reduces the effective strength of the solar gravity. Because the Sun is getting more luminous with time, by about 30% since it's early youth, then this should result in a weaker effective gravity and the Earth's orbit behaving like it orbits a correspondingly less massive star.
This is a 1 part in $10^{16}$ effect so can safely be neglected.
Drag from the Interplanetary Medium
The density at the Earth is of order 5 protons per cubic centimetre (though I can't immediately locate a reference for this much quote figure) and thus a mass density $\rho \simeq 10^{-20}$ kg/m$^{3}$. The drag force is about $0.5 \rho v^2 \pi R^2$, where $v$ is the velocity of Earth's orbit and $\pi R^2$ is the cross-sectional area presented by the Earth. The torque is then this multipled by $a$ and if assumed constant over billions of years would decrease the angular momentum of the Earth's orbit by 1 part in a billion - so this can safely be neglected.
