Two‐step particle swarm optimization algorithm for effective deconvolution and resolution enhancement of various overlapping peaks

Abstract

Rationale The existing particle swarm optimization (PSO) algorithms are only effective in deconvoluting the overlapping peaks in ion mobility spectra with fewer than four component peaks, which limits the applicability of these algorithms. Methods A high‐performance two‐step particle swarm optimization (TSPSO) algorithm was developed. Compared to the existing PSO algorithms, TSPSO can narrow the search ranges of all coefficients for the overlapping peaks through Gaussian model calculation, and thus can deconvolute various overlapping peaks with high accuracy, even for 30‐component overlapping peaks. In addition, the TSPSO could be further applied to enhance the resolution of the spectra by narrowing the peak widths after the peak deconvolution. Results Simulated overlapping peaks were first used to evaluate the performance of TSPSO as compared to the dynamic inertia weight particle swarm optimization (DIWPSO) algorithm. The results showed that the profiles of the peaks deconvoluted by using TSPSO were more consistent with the original ones. The fitness values and the standard deviations of the fitness values from TSPSO were also at least an order of magnitude less than those from DIWPSO. By applying TSPSO, the overlapping peaks from both mass spectrometry (MS) and field asymmetric waveform ion mobility spectrometry (FAIMS) spectra can also be well deconvoluted. In addition, the resolutions of the MS and FAIMS spectra can be effectively enhanced after peak deconvolution. The enhanced spectra matched excellently with the experimental ones acquired at high‐resolution modes. Conclusions The experiment results convincingly demonstrate that the TSPSO algorithm is capable of both deconvoluting complex overlapping peaks and enhancing the spectrum resolution with high accuracy.

Radial basis function neural network optimization algorithm based on dynamic inertial weight particle swarm optimization for separating overlapping peaks in ion mobility spectrometry

“…To overcome these limitations, researchers have widely used dynamic inertial weight particle swarm optimization (DIWPSO) algorithm. 27,28 This algorithm combines differential iteration and inertia weight concepts to achieve enhanced global search and rapid convergence through adaptive inertia weight adjustments. By using DIWPSO to optimize RBF neural network, researchers have achieved adaptive parameter selection capabilities, demonstrating outstanding performance in addressing complex optimization problems.…”

“…However, traditional PSO algorithms have certain limitations, such as slow convergence speed and susceptibility to local optima. To overcome these limitations, researchers have widely used dynamic inertial weight particle swarm optimization (DIWPSO) algorithm 27,28 . This algorithm combines differential iteration and inertia weight concepts to achieve enhanced global search and rapid convergence through adaptive inertia weight adjustments.…”

Overlapping peak separation algorithm for ion mobility spectra based on multistrategy JAYA

“…Moreover, for different problems, the reasonable setting of these parameters significantly contributes to the discovery of the algorithm's optimal solution. TSPSO incorporates Gaussian model calculations into DIWPSO to narrow the coefficient search range in the algorithm 17 . However, TSPSO essentially relies on DIWPSO for optimization search, and the aforementioned DIWPSO deficiencies persist in this algorithm.…”

“…TSPSO incorporates Gaussian model calculations into DIWPSO to narrow the coefficient search range in the algorithm. 17 However, TSPSO essentially relies on DIWPSO for optimization search, and the aforementioned DIWPSO deficiencies persist in this algorithm.…”