To construct concrete in recent decades, the reducing freshwater is one of the basic requirements. Feasibility of replacing freshwater with seawater has been confirmed in literature. This study aims to investigate the effects of fresh and salty water on the compressive strength of the concrete samples after 28 days. Then, two hybrid artificial intelligence techniques namely neuro-swarm and neuro-imperialism are proposed to predict the concrete compressive strength. In these two hybrid models, the particle swarm optimization and imperialist competitive algorithm were used to optimize the weights and biases of the artificial neural network to get a higher performance prediction results. For the purpose of this study, a series of experiments were carried out in laboratory based on different values of cement content, magnetic field intensity, water rotation time, and water to cement ratio. According to the results obtained in laboratory, the mentioned parameters had a deep impact on compressive strength of concrete samples. Therefore, these parameters were set as model inputs to estimate the compressive strength of concrete samples. Several hybrid intelligent models with different effective parameters were built to obtain the most accurate model in estimating compressive strength of concrete. The proposed models were assessed using some statistical indices e.g., system error and coefficient of determination (R²). As a result, both hybrid intelligence models were able to provide a high accuracy level for predicting compressive strength of concrete. However, neuro-swarm model received a better results compared to neuro-imperialism model. Considering R² values for train and test phases, values of (0.9811 and 0.9668) and (0.9785 and 0.9706) were obtained for the neuro-swarm and the neuro-imperialism models, respectively which confirmed the better performance of the neuro-swarm intelligence model in predicting compressive strength values of concrete samples.

近几十年来建造混凝土，减少淡水是基本要求之一。用海水代替淡水的可行性已在文献中得到证实。本研究旨在研究淡水和咸水对 28 天后混凝土样品抗压强度的影响。然后，提出了两种混合人工智能技术，即神经群和神经帝国主义来预测混凝土的抗压强度。在这两个混合模型中，粒子群优化和帝国主义竞争算法被用来优化人工神经网络的权重和偏差，以获得更高的性能预测结果。为了本研究的目的，根据水泥含量、磁场强度、水轮换时间和水灰比。根据实验室获得的结果，上述参数对混凝土样品的抗压强度有很大影响。因此，这些参数被设置为模型输入以估计混凝土样品的抗压强度。建立了几种不同有效参数的混合智能模型，以获得估计混凝土抗压强度最准确的模型。使用一些统计指标评估所提出的模型，例如系统误差和决定系数（R 这些参数被设置为模型输入以估计混凝土样品的抗压强度。建立了几种不同有效参数的混合智能模型，以获得估计混凝土抗压强度最准确的模型。使用一些统计指标评估所提出的模型，例如系统误差和决定系数（R 这些参数被设置为模型输入以估计混凝土样品的抗压强度。建立了几种不同有效参数的混合智能模型，以获得估计混凝土抗压强度最准确的模型。使用一些统计指标评估所提出的模型，例如系统误差和决定系数（R²）。因此，两种混合智能模型都能够为预测混凝土的抗压强度提供高准确度。然而，与神经帝国主义模型相比，神经群模型获得了更好的结果。考虑到训练和测试阶段的R ²值，神经群和神经帝国主义模型分别获得了（0.9811 和 0.9668）和（0.9785 和 0.9706）的值，这证实了神经群智能模型的更好性能预测混凝土样品的抗压强度值。