Study On The Control Method Of Organic Rankine Cycle Systems With Non-Gaussian Disturbance

Energy resources are very abundant in our country,but there are some problems such as unreasonable energy consumption structure,lower energy conversion efficiency and wasted energy in the form of residual heat.Therefore,the recycling of waste heat resources to improve the energy efficiency is of great significance.In low temperature waste heat utilization technology,organic Rankine cycle(ORC)has become an effective waste heat utilization technology due to its simple structure,low operating cost and high heat recovery.It is of great significance to study the optimal operation of ORC from the perspective of control.However,in the actual operation of ORC,it is inevitably be interfered by random noise which obeys non-Gaussian distribution,such as mass flow rate of flue gas and inlet temperature of heat source,etc.Because the traditional control method can not fully reflect the non-Gaussian characteristics of the system,the control performance is not ideal.Therefore,based on the non-Gaussian control theory,generalized correntropy(GC)criterion is adopted to study the control problem of ORC system under non-Gaussian disturbances.The main contributions of the dissertation can be summarized as follows:A single neuron adaptive multi-step predictive control method is proposed for ORC waste heat utilization system with non-Gaussian disturbance.Since there are non-Gaussian disturbances in the waste heat utilization system that follow the heavy-tailed distribution,GC is used to construct the performance index function to characterize the randomness of the system.By optimizing the performance index function and updating the parameters of the single neuron controller,the iterative optimal control law is obtained.As a comparison,the performance of the minimum error entropy(MEE)controller and the mean square error(MSE)controller are tested.The simulation results show that the proposed GC based single neuron predictive control can effectively suppress the influence of the random disturbance following the heavy-tailed distribution.Considering the energy conversion efficiency of the ORC waste heat utilization system,the given set value may not be proper or reachable,a two-layer dynamic economic model predictive control(EMPC)strategy is proposed.In the upper layer,the ratio of net efficiency to total heat transfer area is used as the function of economic performance index.The lower layer uses GC to construct the target function of tracking control,so that the system output can track the optimal reference trajectory as much as possible.Finally,the feasibility and effectiveness of the two-layer dynamic EMPC strategy based on GC is verified by simulation.In addition,due to the real-time change of the optimal set value and the large amount of predictive control calculation,a nonlinear explicit model predictive control(NEMPC)strategy is proposed.It uses the offline training of the controller model to replace the online calculation of the optimal control law under the twolayer dynamic EMPC framework.Once the optimal reference trajectory of the upper layer is sent to the lower layer,the lower layer immediately gets the control action according to the trained controller model,which greatly improves the operating efficiency of the system.Finally,the NEMPC method is compared with the traditional two-layer MPC method to verify the effectiveness of NEMPC method.In this paper,the control method of ORC system is studied under nonGaussian disturbance,which not only enriches the theory of non-Gaussian stochastic control theory,but also can be extended to other industrial processes such as chemical industry,metallurgy,etc.,which has guiding significance for its economical and efficient operation.