TwinCAT3中字节数组转LREAL(double)及数组切片索引实现方法咨询

问题1：动态数组区间语法报错的解决 #

TwinCAT 结构化文本不支持使用变量作为上下限的动态数组切片写法[a..b]，切片的上下限必须为编译期常量。不需要对原数组做切片，有两种更高效的实现方式：

• 直接通过DataReceived[i + 固定偏移量 + 字节序号]的方式逐个访问指定位置的字节
• 调用系统函数MEMCPY，从原数组的指定偏移地址复制8个字节到临时缓冲区，无需手动逐个取值

问题2：字节数组转LREAL的实现 #

你之前用BYTEARR_TO_STRING转字符串再转LREAL的方法是完全错误的，这个函数是用来把ASCII编码的字节数组转成字符串，不能处理二进制数值。正确实现方式如下：

1. 首先定义一个共用体类型，匹配8字节LREAL的内存结构：

TYPE BYTE8_LREAL :
UNION
    byte_buf : ARRAY[0..7] OF BYTE; // 8个字节缓冲区
    lreal_val : LREAL; // 对应解析后的双精度浮点数
END_UNION
END_TYPE

2. 处理字节序问题：UR机器人返回的报文是大端序（对应Python代码里struct.unpack('>d')的>标识），而TwinCAT运行的x86/x64平台是小端序，所以需要把读取到的8个字节反转顺序后再填入共用体，才能得到正确的数值。

完整修改后的功能块代码 #

功能块定义部分 #

FUNCTION_BLOCK DataEncoding
VAR_INPUT
    DataReceived: ARRAY[0..1115] OF BYTE;
END_VAR
VAR_OUTPUT
    moment: ARRAY[0..5] OF LREAL := [6(0.0)];
    joint_pos: ARRAY[0..5] OF LREAL := [6(0.0)];
    joint_vel: ARRAY[0..5] OF LREAL := [6(0.0)];
    tcp_pos: ARRAY[0..5] OF LREAL := [6(0.0)];
    tcp_vel: ARRAY[0..5] OF LREAL := [6(0.0)];
    tcp_force: ARRAY[0..5] OF LREAL := [6(0.0)];
    state: LREAL := 0.0;
    time_new: LREAL := 0.0;
END_VAR
VAR
    i: INT;
    conv_union: BYTE8_LREAL; // 转换用共用体
    j: INT;
END_VAR

功能块主体逻辑部分 #

// 循环上限设为1112，避免i+3越界访问数组
FOR i := 0 TO 1112 BY 1 DO
    // 匹配报文头0x00 0x00 0x04 0x5C（十进制92）
    IF DataReceived[i] = 0 AND DataReceived[i+1] = 0 AND DataReceived[i+2] = 4 AND DataReceived[i+3] = 92 THEN
        // 解析moment数组，偏移从204开始，每个值占8字节
        FOR j := 0 TO 5 DO
            // 复制8个字节到共用体缓冲区
            MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 204 + j*8]), 8);
            // 反转字节序适配大端报文
            conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
            moment[j] := conv_union.lreal_val;
        END_FOR;

        // 解析joint_pos数组，偏移从252开始
        FOR j := 0 TO 5 DO
            MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 252 + j*8]), 8);
            conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
            joint_pos[j] := conv_union.lreal_val;
        END_FOR;

        // 解析joint_vel数组，偏移从300开始
        FOR j := 0 TO 5 DO
            MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 300 + j*8]), 8);
            conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
            joint_vel[j] := conv_union.lreal_val;
        END_FOR;

        // 解析tcp_pos数组，偏移从444开始
        FOR j := 0 TO 5 DO
            MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 444 + j*8]), 8);
            conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
            tcp_pos[j] := conv_union.lreal_val;
        END_FOR;

        // 解析tcp_vel数组，偏移从492开始
        FOR j := 0 TO 5 DO
            MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 492 + j*8]), 8);
            conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
            tcp_vel[j] := conv_union.lreal_val;
        END_FOR;

        // 解析tcp_force数组，偏移从540开始
        FOR j := 0 TO 5 DO
            MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 540 + j*8]), 8);
            conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
            tcp_force[j] := conv_union.lreal_val;
        END_FOR;

        // 解析state，偏移1052
        MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 1052]), 8);
        conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
        state := conv_union.lreal_val;

        // 解析time_new，偏移4
        MEMCPY(ADR(conv_union.byte_buf), ADR(DataReceived[i + 4]), 8);
        conv_union.byte_buf := [conv_union.byte_buf[7], conv_union.byte_buf[6], conv_union.byte_buf[5], conv_union.byte_buf[4], conv_union.byte_buf[3], conv_union.byte_buf[2], conv_union.byte_buf[1], conv_union.byte_buf[0]];
        time_new := conv_union.lreal_val;

        // 找到报文头后退出循环，和Python逻辑一致
        EXIT;
    END_IF;
END_FOR;

如果需要优化性能，也可以提前把字节序反转封装成一个单独的函数，避免重复写反转逻辑。

内容的提问来源于stack exchange，提问作者Harry