Safety in RL #101 started with condensed matrices

examples/scripts/Safeguard_test.jl

@@ -0,0 +1,134 @@
+using ElectricGrid
+
+CM = [ 0. 0. 1.
+        0. 0. 2.
+        -1. -2. 0.]
+
+#R_load, L_load, _, _ = ParallelLoadImpedance(50e3, 0.95, 230)
+
+parameters =
+Dict{Any, Any}(
+    "source" => Any[
+                    Dict{Any, Any}(
+                        "pwr" => 200e3,
+                        "fltr" => "LCL",
+                        "L1" => 70e-6,
+                        "R1" => 1.1e-3,
+                        "C" => 300e-6,
+                        "R_C" => 7e-3,
+                        "control_type" => "classic",
+                        "mode" => "Droop",),
+                    ],
+    "load"   => Any[
+        Dict{Any, Any}(
+            "impedance" => "R",
+            "R" => 100,
+            "v_limit" => 1e4,
+            "i_limit" => 1e4)
+        ],
+    "grid" => Dict{Any, Any}(
+        "phase" => 1,
+        "ramp_end" => 0.04,)
+)
+
+
+
+env = ElectricGridEnv(
+    #CM =  CM,
+    parameters = parameters,
+    t_end = 1,
+    action_delay = 0,
+    verbosity = 0)
+
+
+env.A
+
+#= To exclude the cable C from matrix and conpare result with matlab:
+ - LCL filter, only use LC (not 2nd L) for SG calculation, since we have no cable in matlab
+    - Matrix look to be the same -> A_P10
+ - Caculate condensed system A_N and B_N
+    - Same?
+ - Calcuate G and E,...
+    - Same?
+ - Calcuate Polyhedron
+    - Same?
+ - Use it to ensure safety
+=#
+
+A_P10 = env.A[1:2,1:2]
+B_P10 = env.B[1:2]
+
+W_u = [-1; 1]
+omega_u = [
+    env.nc.parameters["source"][1]["vdc"];
+    env.nc.parameters["source"][1]["vdc"]
+         ]
+
+
+W_x = [
+    -1 0;
+    1 0;
+    0 -1
+    0 1
+    ]
+
+omega_x = [
+    env.nc.parameters["source"][1]["i_limit"];
+    env.nc.parameters["source"][1]["i_limit"];
+    env.nc.parameters["source"][1]["v_limit"];
+    env.nc.parameters["source"][1]["v_limit"]
+]
+
+A_sqare = A_P10^2
+
+N = 4  # maximum number of iteration
+
+m = 1 # number of inputs to the system,
+n = size(B_P10)[1]
+#TODO: get from B -> here has (2,) dim and not (2,1), why?
+# better: (n, m) = size(B_P10)
+
+global A_N = 1* Matrix(I, size(A_P10)[1], size(A_P10)[1])
+global B_N = zeros(n, m*N)
+
+for ii  = 1:N
+
+    global A_N = [A_N; A_P10^ii]
+
+
+
+    B_N_newline = A_P10^(ii-1)*B_P10
+    println("First:")
+    println(B_N_newline)
+    println("")
+
+    for jj in 2:ii
+        println("First:")
+        println(A_P10^(ii-jj)*B_P10)
+        println("")
+        B_N_newline = hcat(B_N_newline, A_P10^(ii-jj)*B_P10)
+
+        println("")
+        println(jj)
+        println("")
+        println(B_N_newline)
+        println("")
+    end
+    B_N_newline = hcat(B_N_newline, zeros(n, m*(N-ii)))
+    println("Last:")
+    println(B_N_newline)
+    println("")
+
+
+
+    global B_N = [B_N; B_N_newline]
+
+
+
+end
+
+println("")
+println(A_N)
+println("")
+println(B_N)
+println("")