Abstarct: Management of human resources is one of the most important goals of business process management systems (BPMS) which strive to improve the quality of products and services. Workload balancing and workload uniformity of resources improve the performance of business process execution in BPMS. Different techniques and methods are used for modeling of the business process execution such as Markov Decision Processes (MDPs), Queue Network Model (QNM) and Workflow Network (WF-net). In this thesis, a statistical process-aware model is presented for modeling of business process execution that is called Extended Markov Decision Processes (XMDPs). This model is a developed version of MDPs, so that the business process execution can map on the XMDPs model better than the MDPs model. The main idea in the proposed model is to support more events that they map on the proposed model. Thus, the main goal of optimization in mapping business process execution on the proposed model is workload balancing and creating uniformity in the workload of resources. This problem is solved by using reinforcement learning which helps to make better resource allocation in order to achieve the mentioned goals. The proposed method is compared with Reinforcement Learning Resource Allocation Mechanism (RLRAM) baxsed on the MDPs model on the BPI Challenge 2102 data set. The experimental results show that the problem of workload balancing and uniformity of workload can be managed 12.7% better with a proper mapping of the business process execution on the proposed model. Furthermore, by presenting and employing a statistical method on event logs, a kind of dependency of tasks is extracted. By considering the mentioned tasks dependency, some algorithms are presented and suggested to improve resourse allocation performance. The proposed algorithms are also compared with other algorithms in the research background. The observed results show the optimal effect of the concurrent execution of tasks in the resources allocation performance. In continued research in this thesis, the “tuning concurrency of business processes” problem is introduced with the aim of balancing and uniformity of workload. Solving this problem is studied in three different methods. The tuning concurrency of business processes have experimented on real data set. The experimental results show improvement by 21.64% in the balancing and uniformity of resource workloads.