I really think nidhin's answer is the best way to go. Use the Laplace transform, calculate \$I_2(s)\$ and then convert it back into a differential equation if you're not interested in the solution.
I just did it for you to show you what you're up against if you try to avoid Laplace transforms:
The Laplace transform gives the solution (I used a CAS for this):
\$I_2\cdot (a\cdot s^3 + b\cdot s^2 + c\cdot s + d) = V_1\cdot (e\cdot s^3 + f\cdot s^2)\$
Yielding the differential equation of the form:
\$a\frac{d^3i_2}{dt^3} + b\frac{d^2i_2}{dt^2} + c\frac{di_2}{dt} + d\cdot i_2 = e\frac{d^3v_1}{dt^3} + f\frac{d^2v_1}{dt^2}\$
Where the coefficients are (I hope I copied them correctly):
\$a = C_1C_2L(R_3R_4 + R_2R_4 + R_1R_4 + R_1R_2)\$
\$b = C_1C_2R_3(R_2R_4 + R_1R_4 + R_1R_2) + (C_1+C_2)L(R_2 + R_3) + L(C_2R_4 + C_1R_1)\$
\$c = C_2R_3(R_4 + R_2) + C_1R_3(R_2 + R_1)\$
\$d = R_3\$
\$e = C_1C_2L(R_2+R_3+R_4)\$
\$f = C_1(C_2R_2R_3+L)\$
I would not advise doing this without transforming to Laplace. While it is possible to solve it without, it can be very hard to see how the equations need to be substituted.
Since you've provided your own work (thanks!), here's how I'd approach the problem. You can weed through your own work, looking over mine, for your own errors. I want to leave something for you to do; that, plus some algebra work I'll also later leave for you, below.
I'm going to always start my loops (all four) from the central node of your schematic. And I will also always follow the direction of your own arrows, as I generate each equation below. And, just to be complete, I will always start each equation with an explicit \$0\:\text{V}\$. The indicated circuit ground is simply ignored for the following purposes:
$$\begin{align*}
0\:\text{V}+V_{I_2}-I_A\cdot R_4-\left(I_A+I_D\right)\cdot R_1&=0\:\text{V}\\\\
0\:\text{V}+V_{I_2}-\left(I_B+I_C\right)\cdot R_2&=0\:\text{V}\\\\
0\:\text{V}-\left(I_C+I_D\right)\cdot R_3+V_1-\left(I_C+I_B\right)\cdot R_2&=0\:\text{V}\\\\
0\:\text{V}-\left(I_C+I_D\right)\cdot R_3+V_{I_1}-\left(I_A+I_D\right)\cdot R_1&=0\:\text{V}\\\\
I_1&=I_D\\\\
I_2&=I_A+I_B
\end{align*}$$
You can either use the last two equations to substitute back into the earlier four equations and then solve for four unknown variables in four equations or else you can simply solve the above all six equations for six unknown variables. Either way works fine.
Note that I did not bother to worry about node voltages, but instead used \$V_{I_1}\$ and \$V_{I_2}\$ to represent the voltages across your two current sources.
I'm not going to bother placing the above equations into matrix form for you. It's just algebraic manipulation and I'm sure you could handle that without difficulty.
It's also left to you as an exercise.
So just let Sage provide the direct results:
sage: var('i1 i2 ia ib ic id r1 r2 r3 r4 v1 vi1 vi2')
(i1, i2, ia, ib, ic, id, r1, r2, r3, r4, v1, vi1, vi2)
sage: e1=Eq(vi2-ia*r4-(ia+id)*r1,0)
sage: e2=Eq(vi2-(ib+ic)*r2,0)
sage: e3=Eq(-(ic+id)*r3+v1-(ic+ib)*r2,0)
sage: e4=Eq(-(ic+id)*r3+vi1-(id+ia)*r1,0)
sage: e5=Eq(i2,ia+ib)
sage: e6=Eq(i1,id)
sage: ans=solve([e1,e2,e3,e4,e5,e6],[ia,ib,ic,id,vi1,vi2])
sage: ans[ia].subs({i1:10e-3,i2:30e-3,v1:10,r1:800,r2:100,r3:300,r4:1000})
-0.00213333333333333
sage: ans[ib].subs({i1:10e-3,i2:30e-3,v1:10,r1:800,r2:100,r3:300,r4:1000})
0.0321333333333333
sage: ans[ic].subs({i1:10e-3,i2:30e-3,v1:10,r1:800,r2:100,r3:300,r4:1000})
0.00946666666666666
sage: ans[id].subs({i1:10e-3,i2:30e-3,v1:10,r1:800,r2:100,r3:300,r4:1000})
0.0100000000000000
sage: ans[vi1].subs({i1:10e-3,i2:30e-3,v1:10,r1:800,r2:100,r3:300,r4:1000})
12.1333333333333
sage: ans[vi2].subs({i1:10e-3,i2:30e-3,v1:10,r1:800,r2:100,r3:300,r4:1000})
4.16000000000000
Those results will be correct, I believe. I just ran it in Spice, only after doing the above, and it produced the same figures.
LTSpice netlist:
R1 N003 N002 800
R2 N001 N003 100
R3 N003 0 300
R4 N001 N002 1k
V1 N001 0 10
I1 0 N002 10m
I2 N003 N001 30m
Best Answer
If you assume that D3 is not quite on the verge of conduction then there is zero current passing through it and, for this ideal diode (that can have a 0.7 volt drop across it before conduction) then the current from Va is all flowing through R2 and this current is 6.7 volts / 10 kohm = 0.67 mA. This current also flows through R1 and therefore the voltage on the left node of R1 is 6.7 volts + 3.35 volts = 10.05 volts. And this means that the voltage on the left hand node of D2 must be (assuming the same ideal diode) is 10.75 volts.
Maybe the flaw is in your equation where you wrote "Va - 0.7 + 6" when it should be "Va - (0.7 + 6)"?