The identification of multi-input-multi-output process is always challenging due to significant loop interactions. This article proposes an identification technique for two-input-two-output processes with fractional order models via Haar wavelet feature. The presented technique can estimate independent four fractional single-pole time delay models without additional steps to decouple processes during identification. The method uses a well-known relay feedback test to produce the input-output data for measurements. Then, the Haar wavelet operational matrix based algebraic approach is utilised to identify the unknown process models with reduced complexity. Numerical analysis on various examples show the efficacy even in presence of noise and without additional filter or preprocessing of measured signals. The comparative study proves advantages of the presented approach.

由于大量的回路相互作用，多输入多输出过程的识别始终具有挑战性。本文提出了一种通过Haar小波特征对具有分数阶模型的二输入二输出过程进行识别的技术。提出的技术可以估计独立的四个分数单极时间延迟模型，而无需在识别过程中采取额外的步骤来解耦过程。该方法使用众所周知的继电器反馈测试来产生用于测量的输入输出数据。然后，利用基于Haar小波运算矩阵的代数方法识别复杂度降低的未知过程模型。对各种示例进行的数值分析表明，即使存在噪声，也无需额外的滤波器或对测量信号进行预处理，该方法的有效性。