The analysis of three-phase synchronous machines in the natural phase reference frame (abc frame) presents significant mathematical challenges. These challenges arise primarily because the self-inductances and mutual inductances of the stator windings vary sinusoidally with the rotor position. This variation results in a set of time-varying, non-linear differential equations. To overcome this, R.H. Park introduced the dq0 transformation in 1929. This technique converts the set of time-varying differential equations into a set of constant-coefficient differential equations by referring the stator quantities to a reference frame that rotates with the rotor.
Before deriving the equations, we must define the assumptions for an idealized synchronous machine:
In the phase domain (a, b, c), the voltage equations for the stator windings are expressed as:
Where $\mathbf{v}_{abc}$, $\mathbf{i}_{abc}$, and $\boldsymbol{\lambda}_{abc}$ are column vectors of stator voltages, currents, and flux linkages. The flux linkages are determined by the inductance matrix $\mathbf{L}_{abc}$, which is a function of the rotor angle $\theta_r$:
Because $\mathbf{L}_{abc}$ depends on $\theta_r$, the derivative term $d\boldsymbol{\lambda}/dt$ becomes complex involving $d\mathbf{L}/dt \cdot \mathbf{i}$.
The goal is to replace the variables in the stationary abc frame with variables in a rotating dq0 frame attached to the rotor. The transformation matrix $\mathbf{K}$ (or $\mathbf{T}$) is defined as:
Here, the factor $2/3$ is used for power invariance (though factors of $\sqrt{2/3}$ are also common). The transformation equations are:
And its inverse:
Where $\mathbf{f}$ represents voltage, current, or flux linkage. The inverse transformation matrix is:
Starting with the stator voltage equation:
We substitute the inverse transformations $\mathbf{K}^{-1}\mathbf{v}_{dq0}$ for $\mathbf{v}_{abc}$ and $\mathbf{K}^{-1}\mathbf{i}_{dq0}$ for $\mathbf{i}_{abc}$:
To solve for $\mathbf{v}_{dq0}$, we premultiply the entire equation by $\mathbf{K}$:
Now, apply the chain rule to the derivative term:
Since $\mathbf{K} \mathbf{K}^{-1} = \mathbf{I}$, the last term simplifies to $d\boldsymbol{\lambda}_{dq0}/dt$. The remaining complex term involves the derivative of the transformation matrix with respect to time.
The term $\mathbf{K} \frac{d\mathbf{K}^{-1}}{dt}$ captures the rotation of the reference frame. Since $\theta_r = \omega_r t$, the derivative of the inverse matrix with respect to time is:
Evaluating the matrix product $\mathbf{K} \frac{d\mathbf{K}^{-1}}{dt}$ yields a constant matrix independent of $\theta_r$:
This represents the rotational coupling. Multiplying this matrix by the flux linkage vector $\boldsymbol{\lambda}_{dq0}$ introduces the "speed voltage" terms:
Substituting the results from the previous section back into the voltage equation gives the renowned dq0 voltage equations for the stator circuits:
In these equations, the terms $-\omega_r \lambda_q$ and $\omega_r \lambda_d$ are often referred to as the rotational voltages or speed voltages. They account for the electromagnetic energy conversion due to the rotation of the rotor.
The transformation also simplifies the inductance matrix. In the dq0 frame, the self-inductances become constant. The mutual inductance between axes $d$ and $q$ is zero. The flux linkage equations are:
Where $L_d$ and $L_q$ are the direct-axis and quadrature-axis synchronous inductances, which are constants for a cylindrical rotor machine (or distinct but constant constants for a salient pole machine), and $L_{ls}$ is the leakage inductance.
By applying the Park transformation, the analysis of a three-phase synchronous machine is significantly simplified. The system of time-varying differential equations in the abc frame is converted into a system of time-invariant differential equations in the dq0 frame. This allows for straightforward linearization, stability analysis, and control design, such as in field-oriented control (FOC) strategies, by treating the AC machine parameters as if they were DC quantities.
