b1_1, b1_2, b1_3 are defined as np.array instead of plain number, and calculating z in the next line by adding b.

Maybe the following code is more reasonable:

x = np.array([200, 17])

w1_1 = np.array([1, 2])

b1_1 = -1

z1_1 = np.dot(w1_1, x) + b1_1