In this paper, we study the impacts of hardware impairment of cooperative non-orthogonal multiple access (NOMA) with an assistant of an energy-harvesting (EH) relay network and an intelligent reflecting surface (IRS) on both energy efficiency (EE) and sum data rate (SDR) metrics. Specifically, in our proposed model, a base station communicates with two NOMA users via the assistance of the selected EH relay employing a power-splitting (PS) architecture. This EH technique is also used in this system to prolong the lifetime of relaying nodes. Moreover, to enhance the performance of our system, an algorithm of swarm intelligence as particle swarm optimization (PSO) is applied to maximize the value of EE and SDR under the constraint conditions of power allocation for each NOMA user. In addition, in this paper, the influence of key parameters of the system is investigated for two cases as IRS and without IRS models in composing of hardware impairment, the number of IRS elements, the number of IRS elements and intermediate nodes, and energy harvesting efficiency on both EE and SDR.