DETERMINISTIC CHAOS PRODUCED BY LOCAL OPTIMISATION ALGORITHMS

Gas turbine gas path diagnosis under transient operating conditions: A steady state performance model based local optimization approach

One promising method of improving Evolutionary Algorithm (EA) performance is to improve its fine tuning capabilities by using an additional local optimization operator in the evolutionary cycle. This customization on the traditional EA is typically called a Memetic Algorithm (MA). Adding the appropriate local optimization algorithm can increase performance, while a poor choice can decrease performance. Thus, some knowledge is required to select the correct algorithm. In many cases the optimal algorithm selection is not known and it may not be static, but could change during evolution. This investigation combines a method of local optimizer evolution using Push Genetic Programming with a method of automatic, self-configuration called Supportive Coevolution. This combination creates a novel MA that coevolves local optimization operators with target fitness function solution candidates. Implementation methodology is shown and experimentation details with corresponding results are presented. Some additional parameters that were discovered for performance tuning are discussed along with a study of their impact on the algorithm's performance. Discussion of some interesting insights followed by some suggestions for further investigation are also provided. Results show the proposed technique can improve the performance of an EA by providing automatically configured, coevolved local optimization operators to a MA.

On optimization of the RBF shape parameter in a grid-free local scheme for convection dominated problems over non-uniform centers

We describe an oil and gas production optimization field application of a miscible gasflood for a cluster of oil reservoirs in South Oman. A portion of the produced sour gas from different reservoirs will be processed as sales gas and the extracted CO2 and H2S will be blended with the remaining produced gas stream, which will be injected back into selected reservoirs for miscible gas flooding. An integrated production system modeling (IPSM) methodology with a fully compositional (EOS, equation of state) description is used to couple the various reservoir models to a surface facilities model. A new element is a compositional gas blending optimization model in the gas processing facilities. This is the main source of non-convexity and non-linearity in the overall optimization model and requires a choice of local and global optimizers to uncover high quality solutions. What optimizer to use depends on the structure and targets of the gas blending problem. The produced oil and gas rates could be 10–20% below the optimum, especially when using local optimizers like SLP and SQP. Even for solutions where the fields produce the same oil and gas rates, and the blended sales and injection gas rates are the same, multiple (sub-)optimal solutions are still possible in the gas blending model by creating injection and sales gas with different compositions. Clear benefits of our IPSM methodology are that long term production forecasts were achieved while honoring all available facilities constraints such as those related to oil and gas production, the gas sweetening unit H2S and CO2 mole rates, injection compressor molecular weight, gas export specifications and facilities uptime. In addition, various forecast scenarios could quickly be generated to study the impact of different optimization objectives, e.g. high gas export vs. high oil production. Very important is also that all the constraints and objectives are contained in a single input file for running the integrated subsurface-surface model. This makes auditing the various forecast scenarios very transparent. Introduction A large number of gas and oil fields are contaminated by significant amounts of CO2 and H2S. Separation of CO2 and H2S from the hydrocarbon gas stream is expensive and leads to disposal problems. However, one way to use these contaminants is to re-inject them back into the oil reservoirs for enhanced oil recovery (EOR) by miscible gas flooding, which can result in very high recovery factors. Depending on the requirements of individual reservoirs for miscible conditions, the extracted H2S and CO2 can be blended in various ratios with produced sour hydrocarbon gas. Such oil production and gas blending optimization problems are challenging when producing from a portfolio of reservoirs with different compositions, pressures and temperatures. In this paper, we present the integrated oil and gas production optimization of a miscible gasflood in South Oman. The carbonate reservoirs are about 100 m thick and located at a depth of 3.5 to 5.5 km. There are several subsurface uncertainties, such as the presence of (sealing) faults, or "thief zones", which are higher permeable layers in low permeable rocks. These may cause the gasflood not to perform at all, or lead to early gas breakthrough. To minimize the risk, development of the oil reservoirs is done in a phased manner. A miscible gasflood EOR scheme also requires abundant availability of injection gas and this is provided by a nearby gas field. How the gas can be blended is described by Deinum et al.1. The various subsurface reservoirs are modeled with the in-house MoReS reservoir simulator in compositional mode and linked to a fit for purpose compositional surface facilities model in the Hydrocarbon Field Planning Tool (HFPT)2,3 a Shell proprietary tool for integrated subsurface-surface modeling. The simulations are run efficiently and robustly using an adaptive implicit non-linear solver with a fast PVT flash method. The well equations in MoReS, including those of the separator train, are solved fully implicitly, whereas the coupling to the facilities is treated explicitly. As gas blending problems are strongly non-linear and non-convex, HFPT has a choice of local and global optimization algorithms to deal with the gas production and blending optimization to produce the sales and injection gas.

In consequence of advances in wireless communication, digital systems and micro-electronic-mechanical system technologies wireless sensor networks (WSNs) have been developed and applied in lots of different real-world applications such as military, industrial, environmental, health, etc. Due to their limited resources and constraints, WSNs face with several problems such as energy consumption, node deployment, data aggregation and transmission. Node deployment affects other problem domains in a direct or indirect way. Therefore, in this study, we deal with this problem and try to increase the coverage area of WSN system with an organized deployment approach. This problem is an NP-Hard problem, and it is hard to solve with standard mathematical formulations. Also this problem is known as maximum coverage sensor deployment problem (MCSDP) in literature. In this paper, worked on the topology that consists from sensor nodes which have the different sensing ranges - heterogeneous. A novel GA is developed for deploying sensor nodes on 2 dimensional area. Studied problem, proposed genetic algorithm and novel local optimization algorithm are described deeply. The result of applying GA to initial deployment topology, increase of coverage are due to iteration count and results of experiments with local optimization algorithm are represented. Organized and random deployment techniques are described briefly and fitness function of developed GA is applied to test data. According to results of experiments; proposed GA is performed effective and got desired coverage.

In view of the hold phase in front of and behind the uphill or downhill of a track, the optimal switch point with local optimal algorithm is not the one with the least cost. This paper finds its deficiencies, and presents an improved method to be applied in the global optimization to minimize the energy costs.

In an unknown environment, autonomous mobile robots rely on sensors to continually obtain information about the surrounding environment, discern the location of obstacles, make calculations and make decisions independently. The existing navigation algorithms are prone to repetition on the rigid path in the face of complex environment such as U-shape, which leads to the navigation can not continue. To this end, this paper presents a local optimization navigation algorithm based on fuzzy logic, using recognition - memory strategy to process the sensor information. In the path planning to retain the location of the recent path and angle characteristics and other related resources, the formation of memory. When the current planning path forms a dead zone and runs repeatedly, identification is formed and the path and navigation decision are re-planned to avoid obstacles colliding. The simulation experiments under Webots Pro and Matlab show that the mobile robot can effectively avoid and avoid the dead zone under the guidance of fuzzy rules and realize better autonomous navigation..

A new local search algorithm for multiobjective optimization problems is proposed to find the global optima accurately and diversely. This paper models the cooperatively local search as a potential field, which is called neighborhood field model (NFM). Using NFM, a new Multiobjective Neighborhood Field Optimization (MONFO) algorithm is proposed. In MONFO, the neighborhood field can drive each individual moving towards the superior neighbor and away from the inferior neighbor. MONFO is compared with other popular multiobjective algorithms under twelve test functions. Intensive simulations show that MONFO is able to deliver promising results in the respects of accuracy and diversity, especially for multimodal problems.

Accurate short-term electrical load forecasting is crucial for the stable operation of power systems. Given the nonlinear, periodic, and rapidly changing characteristics of short-term power load forecasts, this paper introduces a novel forecasting method employing an Extreme Learning Machine (ELM) enhanced by an improved Dwarf Mongoose Optimization Algorithm (Local escape Dwarf Mongoose Optimization Algorithm, LDMOA). This method addresses the significant prediction errors of conventional ELM models and enhances prediction accuracy. The enhancements to the Dwarf Mongoose Optimization Algorithm include three key modifications: initially, a dynamic backward learning strategy is integrated at the early stages of the algorithm to augment its global search capabilities. Subsequently, a cosine algorithm is employed to locate new food sources, thereby expanding the search scope and avoiding local optima. Lastly, a “madness factor” is added when identifying new sleeping burrows to further widen the search area and effectively circumvent local optima. Comparative analyses using benchmark functions demonstrate the improved algorithm’s superior convergence and stability. In this study, the LDMOA algorithm optimizes the weights and thresholds of the ELM to establish the LDMOA-ELM prediction model. Experimental forecasts utilizing data from China’s 2016 “The Electrician Mathematical Contest in Modeling” demonstrate that the LDMOA-ELM model significantly outperforms the original ELM model in terms of prediction error and accuracy.

Effective reservoir management relies heavily on the accurate prediction of reservoir flow performance over the entire life of the reservoir. Accurate prediction of flow performance is possible only if reservoir flow parameters are known with reasonable accuracy. Consequently, a major challenge in reservoir management is the estimation of reservoir parameters that affect the flow and distribution of reservoir fluids the most. Estimation of distributed reservoir parameters has been done using mainly gradient-based optimization algorithms because these algorithms are relatively faster than global (stochastic) optimization algorithms. However, the gradient-based algorithms are local in nature and thus limited in their search ability. In this paper, we present a local-global optimization method that generates multiple realizations of reservoir parameters at coarse scale. The method involves the use of a local search optimization algorithm to parameterize the model space at a coarse scale followed by a stochastic search for better estimates in the vicinity of the local estimate. At the end of the search, the method produces a distribution of estimates that can be used for uncertainty quantification. To test the effectiveness of the method, the local-global optimization algorithm was applied to a sample reservoir with a known distributed permeability field. Results obtained indicate that the method is able to produce multiple history-matched realizations of the permeability field, some of which are closer to the true reservoir permeability distribution than the estimate obtained from an exhaustive local search.

Impact of the Quality of Spatial 3D City Models on Sensor Networks Placement Optimization

In this section, brief introductions to local, global, and multi-objective optimization algorithms and their application to plasmonic and optical nanoantenna and metasurface design are given. Additionally, a flowchart which discusses how to select the appropriate optimization algorithm for a given inverse-design problem is provided.

A new methods of construction and optimization of diffraction optics elements with square zones topology which allow to greatly reduce a sidelobe level by rotating square zones relative to each other are presented. The method of finding maximal of sidelobe level using multiple runs of local search algorithms and zones rotation angles optimization algorithm based on principles of optimizing multiextremal table functions were used. So these elements, for example in microelectronics and THz waveband, can be produced on standard equipment used in chips manufacture.

Solving the distorted wavefront in wavefront sensorless adaptive optics (WFSL-AO) relies on excellent optimizers. Many local or global optimization algorithms have been applied to WFSL-AO; however, there is still a challenge to balance the effect and speed of correcting aberrations. To overcome this, a novel global optimization algorithm named asymptotic proximal point (APP) method is introduced into WFSL-AO in this Letter. We compare this algorithm with the various existing optimization algorithms in convergence speed and correction capability by performing numerical simulations. The results show that the APP method beats all competitors with a better correction effect and faster speed.

Prestack elastic parameter inversion (PEPI) is very helpful in quantitative estimation of reservoir properties. Sparse spike inversion (SSI) is the most widely used method in PEPI. However, SSI using some local optimization algorithms (e.g., conjugate gradient method, CG) usually suffers from a strongly ill-posed problem. The global optimization algorithms can avoid this problem effectively, in spite of a little extra computation cost. Besides, the usually used SSI separately inverts sparse-spike's time locations and its amplitudes and it is low efficient. In this abstract, we adopt orthogonal matching pursuit (OMP) non-linear algorithm for SSI, and derive a linear matrix equation to invert the sparse-spike's time locations and its amplitudes simultaneously (TLAS). We name the method as TLASSSI. The linear matrix equation is derived from the cost function without any constraint terms. In spite of no constraint terms, TLASSSI can still obtain exact inversion results including sparse-spike's time locations and its amplitudes. Numerical examples indicate that our method can achieve a good performance even with an existence of some noise, while the CG algorithm fails to get a desirable inversion result.