The increasing use of Industrial Internet of Things (IIOT) devices changed the industrial setting in terms of connectivity and automation, which helps in productivity growth, but as the network expands, it becomes more vulnerable to various security threats. To protect IIOT networks from these threats, intrusion detection systems (IDS) become essential. IDS plays a vital role in identifying malicious activities and threats. In this study, we proposed an innovative approach to identify intrusion in IIoT networks by integrating deep learning with a hybrid optimization technique; the proposed approach combines the Bidirectional Long Short Term Memory (BiLSTM) model, which harnesses the ability to capture and learn sequential data precisely. Model training is done incrementally, aiming to adapt to the evolving nature of IIOT data, resulting in enhanced accuracy. We have used Spider-Coyote Optimization, which combines Spider Monkey Optimization (SMO) and Coyote Optimization algorithms (COA) to fine-tune network parameters, speeding up the detection process and providing accurate intrusion detection performance. The proposed model is evaluated on the used NSL-KDD dataset, achieving an accuracy of 99.82% for binary classification and 99.76% for multiclass classification. The result analysis suggests that the proposed incremental learning model surpasses the traditional IDS methods.