一. MovingAverage权值滑动平均更新
1.1 示例代码：
def create_target_q_network(self,state_dim,action_dim,net):
state_input = tf.placeholder(float,[None,state_dim])
action_input = tf.placeholder(float,[None,action_dim])
ema = tf.train.ExponentialMovingAverage(decay=1-TAU)
target_update = ema.apply(net)
