Application of neural network and weighted improved PSO for uncertainty modeling and optimal allocating of renewable energies along with battery energy storage
Introduction
Nowadays, renewable energy management modeling is considered as the plausible strategy for energy accessibility. These methods are mainly consisting of optimal selecting, planning, and initiating the process for deploying newfound energy generation. Therefore, it has to be mentioned that, studies which have a conservative base in accord with reliability assessment can increase the energy proficiency in conventional and future power networks [1], [2]. Also, applying such evaluations in the concept of self-producing power systems are signed as grid upgrading progress [3], [4].
Practical programming of renewable energy providers in isolated characteristic grids optimizes the investigation cost according to the demand satisfaction cost [5], [6]. However, the slipping point of such surveys is the time intervals of programming. Because of that, the remote dispatching of renewable energy must meet the initial cost of renewable system [7].
Regarding the energy availability index, the uncertainty problem can be in the core of every renewable power planning project. Therefore, Ref. [8] has been presented a systematic framework of implementing energy hubs to show the deterministic nature of assumption in the microgrid as well as unit commitment layout. Another application of green energy providers (GEP) can be seen in [9] where a case study at Kabul city, Afghanistan has been conducted for increasing the sufficiency of such a network. However, energy management, as well as dealing with constraints of demand curve have to include involute infrastructure to gain the optimum practicality. In this regard, [10], [11] have demonstrated smart, robust strategies against the failure of microgrid to satisfy demands accompanied by increasing the outage cost of a power system and decreasing the duration of interruption in a distribution system, respectively.
Moreover, from the system manager perspective, multi-energy carrier structures are vastly improved, taking into account of their instantaneous corresponding to load deviation [12]. Hence, the GEP extension planning is a critical responsibility in comparison to their designing phenomenon [13], [14]. In Ref [15], a three-phase programming has been demonstrated based on wind, solar, and battery administration to divide the power generating role.
It is noted that the accuracy of evolutionary techniques has been proven [16], [17]. In [18], whale optimization algorithm has been used to adjust the controller in a simulated power network which includes conventional generating unit, wind, solar, and wave energies. Plus, this work has been investigated under the uncertainty of renewables and electrical demand. In [19], a probabilistic approach has been studied for distributing networks considering wind, solar, micro-turbine, and fuel cell. In the mentioned work, uncertainty of wind, solar, and electrical consumption has been taken into account of study. Moreover, affine arithmetic has been applied to find the best interval uncertainty modeling in the term of optimal power flow. Then, stochastic weighted trade-off particle swarm optimization (PSO) has been inserted into the simulation for minimizing the cost function. In accord with our premise, a recursive view has been given to optimization problem, which is included illuminating the vague side of wind farm dataset via specific locating system disability for extensive electrification [20]. The operatory system of this reference is the two-stages robust programming (RP). Plus, the uncertainty of wind farm and generator forces outage has been given as unrelaxed variables. This operator is based on insufficient data for conducting probability distribution function which aims to mitigate the impact of change in the term of optimal allocation of wind power and to increase the reliability by minimizing the index of expected energy not served. The uncertainty has been fixed in the first stage of RP. Then, decisions have been linearized to recalculate by second-order cone programming of robust technique.
Besides, the moderate cost of renewable energy founding is an encouraging element for utilizing along with operating. Therefore, final appraisement of wind, solar, and battery settlement has been depicted in various case study [21]. Moreover, the wind generator application can efficiently narrate its advantages by allowing human knowledge formed technique which is hybrid adaptive neural fuzzy inference system-genetic algorithm (ANFIS-GA) method to interact with wind energy conversion system [22]. In the well-known work, ANFIS-GA improves the error of steady-state stability and peak overshoot. The elapsed setting time of controller has been minimized to 0.021(s). In this matter, ANFIS tactic enhances the performance of both generator side converter and the grid side inverter by establishing optimal rules of FIS and optimizing interior parameters of FIS by GA. Then, results have been compared with other ANFIS based controllers, proportional–integral controller, and grey wolf optimization-based controller.
There is strong evidence for the notion that conservative decisions have been supposed rationally in the field of wind energy analysis and flexibility encountering natural disturbances. Given all aspects of the precious debate, the fault detection criterion has been expanded to dominate in high marine potentially sites [23]. Furthermore, the systematic discussion of heterogeneous modeling of wave energy converter (WEC), which is considered as nonlinear tractable theory can be seen in [24], [25]. Plus, the conformity with newfound developing generations is the common topic for accelerating the steps of WEC deployment by means of pragmatic technology [26], [27]. Parallel to this point, the energy storage system expanding has been presented in a study concerning the WEC improvements and administrative [28]. In [29], the overall calibration of the wave driver has been prosecuted for increasing the performance of each renewable energy manufacturing unit through the discrete value-based PSO. In the abovementioned work, the optimization has been included two controlling units for mitigating the stalling obstacle of WEC. In this regard, PSO has been applied to adjust the proportional–integral–differential controller, which has a functionality to damp the fluctuation of airflow. Then, back to back converter has been used to decrease the deviation of turbine speed.
Moreover, the comprehensive deployment of updated PSO for designing the photovoltaic system in the form of optimization problem has been carried out in [30]. In the noticed work, applications of PSO based methods have been comprehensively illuminated for sizing and allocating of solar energy. Plus, the traceability of such PSO based solutions for obtaining maximum power point track has been discussed too.
It is noted that the combination of evolutionary algorithms with searching techniques leads to an acceptable solution. Hence, a hybrid approach which is based on advanced swarm optimization (CSO) and Powell’s pattern search (PPS) has been presented for optimal dispatching of combined heat and power system [31]. It has to be mentioned that PSO basically structures CSO. The well-known work has been ignited by CSO to find the best ranks of members of society. After forming the population of the best members, PPS has been applied to update the positions of the best members by searching new pattern direction. Then, the locations of the best members have been regulated by multiplying the random step size with pattern search and adding with the current situation. Plus, the accumulated error (AE) of solutions has been computed through Euclidean distance which shows the difference of obtained results and solutions of boundaries. Moreover, the objective function has been reset as the minimization problem of the penalty factor for AE and primary statistic function. It is noted that sensitivity tests have been verified the operation.
In [32], the PSO technique along with the relaxing strategy of constraints have been applied to solve the multi-objective problem of hydropower reservoir by increasing the speed of convergence. The conducted PSO technique has been divided into multi-objective swarm which has been simultaneously computed the best subpopulation. Then, knowledge-based constraints have been modified by the mutation operator to increase the speed of convergence.
Another application of PSO can be seen in [33] which the aim is to schedule the hydrothermal systems with a cascaded reservoir in a short time intervals. In this study, the PSO has been weighted to avoid converging local optima.
The energy storage system (EES) are well-profited technologies which can reduce the energy not served index in terms of coordination with GEP. This overview can help system protection sector for efficiently responding to the sharp fluctuation of demand. Considering different types of distributed unit and interior constraints, optimal design of EES for uncertainty management have been investigated in distribution networks [34], [35].
To minimize the objected cost of islanded power network, the demand response (DR) scheduling has been discussed for the battery energy compensatory system cooperation characteristics [36]. This study is a two-stage scheduling which the uncertainty of wind, solar, electrical loads, and price of energy have been given. Firstly, scenario-based normal distribution function has been generated to categorize the uncertainties. Then, the backward method has been implemented to adjust feasible scenarios. Secondly, the mixed-integer quadratic form of objective function has been formulated considering the regulations of battery.
Broadly certified by scholars, the unpredictable disposition of newfound energy providers can make harmful transient penalties for system utilizers. According to the preceding notion, a dominant policy has been initiated by introducing storage capacity for installing future power networks, taking into account of energy efficiency establishment [37]. In [38], a sensitivity description of energy storage has been investigated for disburdening the obstacles of PV precipitant. Additionally, the convex symmetrical polyhedral arranging of energy storage enlarging in term of interoperability with regulatory algorithms can be seen in [39], [40].
Many pilot prediction types of research are now widely pursued in the field of power accumulation operatory appearance due to their governing role in automated power delivery alignment [41], [42]. It is worth noting that, GEP generating systems cannot assure end-user side decision-making services for harnessing GEP sectors taking into account of their inherent output power deviation. Therefore, mechanisms which can normalize the unrelaxed properties of GEP in a cognizable mode, as well as system ambiguity clearance, leads to maximum resiliency harvesting. Hence, a hybrid control scheme has been drawn for wind power estimating by extreme learning machine [43]. This definition has superiority among numerical based functional analyzers for artificial neural network (ANN) property.
According to [44], curve fitting difficulty can be handled by inserting neural network (NN) processing as well as regression of support vector machine for linearizing unbalanced wind speed dataset. To portray the issue of renewable energy application, a simulation has been carried out at Goldwind microsystem of Beijing, which benefits from PV composition [45]. The product of the above-mentioned study is that by optimizing the weight vectors of ANN using GA and PSO tactics incorporated with Gaussian approach of solar irradiance predicting, the output power produced can enhance the stability of assumed field study. It is noted that direct interval estimation of wind power can be improved in the presence of recurrent adaptive neural network architecture [46].
In this paper, the aim is to allocate wind, solar, wave energies with battery energy storage in a microgrid located at Hormoz Island, Iran [47]. To do so, the objective function is formulated based on the technical features of renewables considering interest rate, inflation rate and exacerbation rate of generating section. Plus, the loss of power supply probability (LPSP) of end users is assumed to regulate the cost of allocating through planning width. However, the uncertainties of wind and solar energies must be handled. Hence, two ANN are used to predict the wind speed and solar irradiance daily. To minimize the prediction error, three training algorithms are described which have been structured by back propagation (BP) theorem. In this regard, the first training algorithms describes the function of prediction error. Then, other two training algorithms are presented to increase the speed of search and to find the direction of search for seeking optimal weights and biases of ANN, respectively. After the acceptable prediction observed, the output power of wind and solar are relaxed by the combination of error and last historical datasets of wind and solar energies. In the next step, weighted improved PSO (WIPSO) optimizes the eco-statistic objective function. Further, the LPSP is considered to secure the power balance of system with respect to NPC and LCE of proposed hybrid mechanism. Finally, the results of proposed architecture are compared with six predictors and four searching engines. Hence, the hybrid proposed method is constructed by two layer. The uncertainty relaxation process of wind and solar energies are modeled by DANN in the first layer. Then, WIPSO plays the solving engine in the second layer. Hence, the problem descriptions are specified in Section 2. The architecture of DANN relaxation, GEP uncertainty modeling, and WIPSO description are presented in Section 3. The methodology is illuminated the feasible scenarios through the reliability in Section 4. Finally, Section 5 concludes the robustness of performance as well as suggests the future works.
The contribution of this paper is itemized as follows.
- •
Integrated modeling of wind, solar and wave energies.
- •
Battery energy management strategy for compensating load curtailment.
- •
Uncertainty modeling of wind and solar via dynamic artificial neural network.
- •
Using weighted improved PSO for solving optimization problem.
The flow chart of proposed hybrid technique is illustrated as in Fig. 1. According to Fig. 1, proposed allocating strategy is included two layers. In the first layer, information of wind, solar, wave and electrical demand are gathered. Among the historical datasets, wind speed and solar irradiance are inserted into two DANNs as input to be forecasted daily. The relaxation process is ignited using the prediction errors and last data of wind speed and solar irradiance. In the second layer, eco-statistic objective function is organized. Then, WIPSO finds optimal solutions of objective function. The inputs of this paper is the information of Hormoz Island, Iran.
Section snippets
Problem description
The proposed hybrid field study is mapped for generating units which are solar photovoltaic, wind energy, WEC, and BECS. The case study is conducted in a residential grid which is a suitable region for a smart islanded system implementing in accord to the geographical conditions. The objective function formulation of each unit is depicted in the following.
Wind speed and solar irradiance relaxation by DANN
Due to the meteorological conditions of proposed site, the nonlinear dimension of wind speed and solar irradiance leads to uncertainty obstacle in the optimizing process [61], [62], [63]. Moreover, DANN is fabricated by multi-layer perceptron (MLP) structure of NN which is followed by three main steps: training, testing, and validating of heterogeneous datasets. In this road map, the aim is to link unrelaxed sets with historical time series for decreasing the slope of prediction error. This
Robustness and scenario analysis
In this paper, a revolutionary algorithm is deployed for optimal planning of 15 residential demands. As it is noticed, the load growth and relaxation of nonlinear parameters are considered to increase the accuracy. The conducted methodology is applied by Matlab® software. In this regard, five feasible scenarios are brought in the line to generalize the validity of proposed hybrid scheme as follows:
- 1.
Base case scenario: conventional topology without GEP.
- 2.
Scenario 1: the sole perspective of GEP
Conclusion and future works
The proposed approach in this paper is conducted to provide the electrical demand of a microgrid (Hormoz Island, Iran) by DANN and WIPSO for predicting the uncertain elements and computing the optimization, respectively. The microgrid is prone to harvest from wind, solar and wave energies due to its meteorological condition. Plus, the uncertainty of wind and solar energies is a crucial parameter which can cause contingency in the calculations. Also, the BECS is studied to cover the deficiency
Declaration of Competing Interest
No author associated with this paper has disclosed any potential or pertinent conflicts which may be perceived to have impending conflict with this work. For full disclosure statements refer to https://doi.org/10.1016/j.asoc.2019.105979.
References (73)
- et al.
Emergence of floating offshore wind energy: Technology and industry
Renew. Sustain. Energy Rev.
(2019) - et al.
Energy ratio analysis and accounting for renewable and non-renewable electricity generation: A review
Renew. Sustain. Energy Rev.
(2018) - et al.
Solving non-convex/non-smooth economic load dispatch problems via an enhanced particle swarm optimization
Appl. Soft Comput.
(2017) - et al.
A comprehensive study of practical economic dispatch problems by a new hybrid evolutionary algorithm
Appl. Soft Comput.
(2017) - et al.
Hybrid bio-inspired computational intelligence techniques for solving power system optimization problems: A comprehensive survey
Appl. Soft Comput.
(2018) - et al.
Dynamic resource allocation for parking lot electric vehicle recharging using heuristic fuzzy particle swarm optimization algorithm
Appl. Soft Comput.
(2018) - et al.
Reliability and O & M sensitivity analysis as a consequence of site specific characteristics for wave energy converters
Ocean Eng.
(2017) - et al.
Optimization of nonlinear wave energy converters
Ocean Eng.
(2018) - et al.
Review of control strategies for wave energy conversion systems and their validation: the wave-to-wire approach
Renew. Sustain. Energy Rev.
(2018) - et al.
Evaluation of performance metrics for the wave energy prize converters tested at 1/20th scale
Renew. Sustain. Energy Rev.
(2018)
A review of particle swarm optimization and its applications in solar photovoltaic system
Appl. Soft Comput.
Combined heat and power economic dispatch using integrated civilized swarm optimization and Powell’s pattern search method
Appl. Soft Comput.
A parallel multi-objective particle swarm optimization for cascade hydropower reservoir operation in southwest China
Appl. Soft Comput.
Short-term combined economic emission scheduling of hydrothermal systems with cascaded reservoirs using particle swarm optimization technique
Appl. Soft Comput.
A new method based on type-2 fuzzy neural network for accurate wind power forecasting under uncertain data
Renew. Energy
Short-term wind power forecasts by a synthetical similar time series data mining method
Renew. Energy
A mathematical approach to minimizing the cost of energy for large utility wind turbines
Appl. Energy
Techno-economical evaluation of wind energy potential and analysis of power generation from wind at Gharo, Sindh Pakistan
Renew. Sustain. Energy Rev.
An experimental comparison between commercial hybrid PV-T and simple PV systems intended for BIPV
Renew. Sustain. Energy Rev.
Feasibility of WEC installations for domestic and public electrical supplies: A case study off the Calabrian coast
Renew. Energy
Wave energy assessement for Northern Spain from a 33-year hindcast
Renew. Energy
Overview of energy storage systems in distribution networks: Placement, sizing, operation, and power quality
Renew. Sustain. Energy Rev.
Optimal sizing of energy storage system and its cost–benefit analysis for power grid planning with intermittent wind generation
Renew. Energy
A levelized cost of energy (LCOE) model for wind farms that include power purchase agreements (PPAs)
Renew. Energy
Applied levelized cost of electricity for energy technologies in a small island developing state: A case study in Mauritius
Renew. Energy
Reliability-based design optimization in offshore renewable energy systems
Renew. Sustain. Energy Rev.
A reliability-based stochastic planning framework for AC-DC hybrid smart distribution systems
Int. J. Electr. Power Energy Syst.
Comparative and comprehensive review of maximum power point tracking methods for PV cells
Renew. Sustain. Energy Rev.
Joint optimization of day-ahead and uncertain near real-time operation of microgrids
Int. J. Electr. Power Energy Syst.
Electricity price driven active distribution network planning considering uncertain wind power and electricity price
Int. J. Electr. Power Energy Syst.
Peak load forecasting using Bayesian regularization, resilient and adaptive backpropagation learning based artificial neural networks
Electr. Power Syst. Res.
Forecasting based on neural network approach of solar potential in Turkey
Renew. Energy
Frequency control in micro-grid power system combined with electrolyzer system and fuzzy PI controller
J. Power Sources
Application-specific modified particle swarm optimization for energy resource scheduling considering vehicle-to-grid
Appl. Soft Comput.
The use of hybrid neural networks, wavelet transform and heuristic algorithm of WIPSO in smart grids to improve short-term prediction of load, solar power, and wind energy
An optimal scheduling controller for virtual power plant and microgrid integration using the binary backtracking search algorithm
IEEE Trans. Ind. Appl.
Cited by (44)
Data driven net load uncertainty quantification for cloud energy storage management in residential microgrid
2024, Electric Power Systems ResearchAn Evolutionary Artificial Neural Network approach for spatio-temporal wave height time series reconstruction
2023, Applied Soft ComputingReview of economic dispatch in multi-area power system: State-of-the-art and future prospective
2023, Electric Power Systems ResearchCitation Excerpt :In searching for the optimal and global solution for ELD problem, various modified and improved algorithms have been considered and implemented over the years for ELD problem. An improved PSO was studied in [137–140], Authors in [60,79,141–143] considered enhanced genetic algorithm, modified cuckoo search [144–146], improved artificial bee colony [147,148], enhanced firefly algorithm [149–151], modified shuffle frog leaping algorithm [152,153], improved differential evolution [154,155], modified harmony search algorithm [24,156,157], chaotic whale optimization algorithm [158,159], improved bacterial foraging [160–162], emended salp swarm optimization [163,164], enhanced moth-flame optimization [165], ameliorated ant lion optimizer [166–168] were considered for solving the highly nonlinear ELD problems. The aforementioned modified techniques are better than their original counterparts, but still suffer slow convergence, especially when they are required to handle several ELD constraints like generation-demand balance, discontinuous prohibited operating zones (POZ), multiple-fuel options (MFO), valve-point loading effect (VPE), and ramp rate constraints independently or simultaneously [25,169].
Scientific mapping of optimisation applied to microgrids integrated with renewable energy systems
2023, International Journal of Electrical Power and Energy SystemsForecasting power load: A hybrid forecasting method with intelligent data processing and optimized artificial intelligence
2022, Technological Forecasting and Social ChangeCitation Excerpt :For example, Zhao et al. (2020) improved the PSO algorithm from both cognitive/social coefficients and additional recovery operators to enhance its search ability. Masoumi et al. (2020) used a weighted, improved PSO algorithm to solve an optimization problem in the renewable energy load forecasting. In this subsection, the framework of the proposed hybrid forecasting model for power load is presented.
Decision-making and optimal design of green energy system based on statistical methods and artificial neural network approaches
2022, Sustainable Cities and Society
- 1
Engineering Faculty, Near East University, 99138 Nicosia, North Cyprus, Mersin 10, Turkey.