根据 Flink 官方文档，Async I/O 运算符的每个实例都会有自己的广播状态，不会共享。如果需要多个实例共享广播状态，则需要手动实现 BroadcastState。

以下是示例代码：

// 定义广播状态描述
public class MyBroadcastStateDescriptor extends MapStateDescriptor {
    public MyBroadcastStateDescriptor(String name, TypeInformation keyTypeInfo, TypeInformation valueTypeInfo) {
        super(name, keyTypeInfo, valueTypeInfo);
    }
}

public class MyAsyncFunction extends RichAsyncFunction {
    private transient MapState broadcastState;

    // 初始化广播状态
    @Override
    public void open(Configuration parameters) throws Exception {
        super.open(parameters);
        MyBroadcastStateDescriptor descriptor = new MyBroadcastStateDescriptor("broadcastState", Types.STRING, Types.LONG);
        broadcastState = getRuntimeContext().getMapState(descriptor);
        if (getRuntimeContext().getIndexOfThisSubtask() == 0) {
            // 只在并行度为1时进行广播
            broadcastState.put("key1", 1L);
            broadcastState.put("key2", 2L);
            broadcastState.put("key3", 3L);
        }
    }

    // 异步处理数据，访问广播状态
    @Override
    public void asyncInvoke(IN input, ResultFuture resultFuture) throws Exception {
        Long value = broadcastState.get("key1");
        // ...
    }

    // 更新广播状态
    @Override
    public void processBroadcastElement(String value, Context ctx, Collector out) throws Exception {
        // ...
        broadcastState.put("key1", 10L);
        // ...
    }
}