确定信号超出预定义限制的时间长度

Question

我有两个大型列表t和y，我希望以高效的方式确定y中的数据超过预定义limit的时间和持续时间，即>=limit。

可以使用以下示例数据说明问题：

t = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16]
y = [8,6,4,2,0,2,4,6,8,6,4,2,0,2,4,6,8]
limit = 4

在此示例中，代码应返回以下列表：

t_exceedance_start = [0,6,14]
t_how_long_above_limit = [2,4,2]

我希望Numpy可以很好地实现这一点，但是没有找到方法。

非常感谢任何建议。

Answer 1

这是一种利用布尔效率提升的矢量化方法 -

# Get array versions if aren't already
y = np.asarray(y)
t = np.asarray(t)

# Get mask of thresholded y with boundaries of False on either sides.
# The intention is to use one-off shifted comparison to catch the
# boundaries of each island of thresholed True values (done in next step).
# Those appended False values act as triggers to catch the start of 
# first island and end of last island.
mask = np.concatenate(( [False], y>=limit, [False] ))
idx = np.flatnonzero(mask[1:] != mask[:-1])

# The starting indices for each island would be the indices at steps of 2.
# The ending indices would be steps of 2 as well starting from first index.
# Thus, get the island lengths by simply differencing between start and ends.
starts = idx[::2]
ends =   idx[1::2] - 1
lens = ends - starts

# Get starts, ends, lengths according to t times
start_times = t[starts]
end_times = t[ends]
len_times = end_times - start_times