1. 创建一个Lambda函数，用于监听Cloudwatch Event和EC2实例的事件。

2. 在Lambda函数中，将事件分为“实例启动”和“实例关闭”两类。使用Boto3库中的describe_instances方法获取实例的ID，状态和其他信息。

3. 如果实例已启动，则创建一个Cloudwatch报警，向SNS主题发送通知。

4. 如果实例关闭，则取消已存在的Cloudwatch报警。

以下是使用Python和Boto3库实现上述步骤的代码示例：

import boto3
import os

# AWS账户的访问密钥
AWS_ACCESS_KEY_ID = os.environ['AWS_ACCESS_KEY_ID']
AWS_SECRET_ACCESS_KEY = os.environ['AWS_SECRET_ACCESS_KEY']

# 用于接收通知的SNS主题ARN
SNS_TOPIC_ARN = "arn:aws:sns:us-west-2:123456789012:instance-alarm"

# Cloudwatch报警命名和描述
ALARM_NAME = "Instance-Change-Alert"
ALARM_DESCRIPTION = "Alert when an instance ID has changed"

# 获取AWS客户端
cloudwatch = boto3.client('cloudwatch', aws_access_key_id=AWS_ACCESS_KEY_ID, aws_secret_access_key=AWS_SECRET_ACCESS_KEY)
ec2 = boto3.client('ec2', aws_access_key_id=AWS_ACCESS_KEY_ID, aws_secret_access_key=AWS_SECRET_ACCESS_KEY)
sns = boto3.client('sns', aws_access_key_id=AWS_ACCESS_KEY_ID, aws_secret_access_key=AWS_SECRET_ACCESS_KEY)

def lambda_handler(event, context):
    # 只监听EC2实例事件
    try:
        if event['source'] == 'aws.ec2':
            instance_id = event['detail']['EC2InstanceId']
            state = event['detail']['state']

            # 检查实例是否存在
            response = ec2.describe_instances(InstanceIds=[instance_id])
            if len(response['Reservations']) == 0:
                return

            # 获取实例的Tags
            tags = response['Reservations'][0]['Instances'][0].get('Tags')
            if tags: