单精度浮点数累积量异常分析及解决方案

Analysis and Solution of Abnormity in Accumulation of Single Precision Float-Point Numbers

自动控制系统普遍采用单精度浮点数实现算术运算和控制策略。采用单精度浮点数对过程量进行累积时,当运行时间足够长,会出现累积量达到一定程度时无法继续累积的现象。针对这一现象,对单精度浮点数表示的有效值范围和数学运算规则进行了分析,提出了一种基于单精度浮点数的分段累积算法。根据实际应用对累积量的精度要求,提供分段数量及各段参数设置指导。该算法不仅可以通过控制器逻辑组态实现,也可以在监控后台通过脚本实现。考虑到控制系统可能出现的故障停运后恢复运行的情况,在分段累积算法上增加了掉电保持功能。现场项目运行检验证明了该算法的有效性。该算法采用的基本算法模块支持各控制系统,具有较好的实用性。

In automatic control systems,arithmetical calculation and control strategy are usually implemented by float-point numbers with single precision. While accumulating the process quantity with single-precision floating-point numbers,if the running time is long enough,there is a phenomenon that accumulation cannot be continued when the cumulative amount is large enough. Aiming at this phenomenon,the range of valid values represented by single-precision floating-point numbers and the mathematical operation rules are analyzed,and a piece-wise accumulation algorithm based on single-precision floating-point number is proposed. According to the accuracy requirements of the accumulation in practical application,the instructions for setting the number of sections and the parameters in section are provided. The algorithm can be implemented both in the logic configuration of the controller and in the monitoring background via a script. Considering the control system may be shut down because of failure,and need to resume the operation after the outage,the retention on power off function is added to the piecewise accumulation algorithm. The effectiveness of the algorithm is proved by the tests of on-site project. The basic modules adopted in the algorithm support control systems and the practicability of algorithm is verified.