International Journal of Computing Science and Mathematics (7 papers in press)

Regular Issues

• Construction of Course Arrangement System for College Sports Teaching Based on Adaptive Factor Optimised SGA  
  by Chunyu Yin, Li Shi 
  Abstract: The course arrangement method for college sports teaching has problems of low efficiency and low quality of course arrangement. This study proposes an intelligent course arrangement method based on improved dual population genetic algorithm. Firstly, course arrangement problems are described, and mathematical model is built. Then, the lowest running time, the highest teacher utilization rate and the best class distribution balance are taken as optimization objectives. Afterwards, an improved dual population genetic algorithm is used to solve course arrangement problems of physical education teaching. Finally, this algorithm is applied to the course arrangement system of college sports teaching for testing and analysis. Experimental results show that the proposed improved dual population genetic algorithm can solve course arrangement problems accurately and quickly. Furthermore, its average running time, average classroom utilisation rate and average class time distribution balance are 53.59s, 96.42% and 0.9, respectively, which are better than those of intelligent course arrangement methods based on ant colony algorithm, firefly algorithm, particle swarm optimisation algorithm, and grey wolf optimization algorithm.
  Keywords: adaptive factor; dual population genetic algorithm; competition mechanism; college sports teaching; intelligent course arrangement.
  DOI: 10.1504/IJCSM.2025.10070151
   
  
• A synergic deep learning approach for efficient grading of glioma via MRI images  
  by Nirmal Yadav 
  Abstract: Computer-aided diagnosis using deep learning approaches has made tremendous improvements in medical imaging for automatically detecting tumor area, tumor type, and grading of the tumor. These advancements are limited due to the fact that (1) medical images are often less in quantity, leading to overfitting, and (2) significant inter-class similarity and intra-class variation between the images. The main aim of the study is to develop a deep learning base model (Zhang et al., 2018; 2019; Krizhevsky et al., 2012) as a backbone for the automatic grading of glioma tumors. The synergic deep learning (SDL) architecture enables two pre-trained models to learn from each other mutually and allows them to perform better than vanilla pre-trained models. Our study uses T1-weighted sagittal tumor magnetic resonance imaging (MRI) slices from the REMBRANDT (Scarpace et al., 2019) dataset. The proposed architecture achieves an accuracy of 98.36%, showing that the model achieves excellent performance metrics on a small dataset.
  Keywords: glioma tumour grading; SDL; synergic deep learning; transfer learning; AlexNet; REMBRANDT.
  DOI: 10.1504/IJCSM.2024.10068238
   
  
• Visual communication method for multi feature media images based on interactive modelling  
  by Ting Zhang, Shi-shun Wang, Xiang-yu Wei 
  Abstract: In order to solve the problems of poor multi feature media image processing, low peak signal-to-noise ratio, and long visual communication processing time in traditional methods, a visual communication method for multi feature media images based on interactive modelling is proposed. hue-saturation-value (HSV) colour histogram, discrete Cosine transform (DCT) transform and generalised search tree (GIST) descriptor were used to extract colour, texture and scene content features of multi-feature media images, and the features were fused. The multi-feature media image is decomposed into semantically independent components, which are mapped to the surface of the 3D model, so as to realise the interactive modelling of the multi-feature media image. Combined with the modelling results, the image is enhanced and reconstructed to complete the visual transmission of the image. Experimental results show that the image detail features of the proposed method are significant, with high clarity. The average peak signal-to-noise ratio is 55.9 dB, and the processing time remains below 0.4 s.
  Keywords: interactive modelling; multi feature media; visual communication; colour; texture; scene content; semantically independent components; enhanced.
  DOI: 10.1504/IJCSM.2024.10068569
   
  
• A method for capturing English oral pronunciation errors based on speech recognition  
  by Wenna Dou 
  Abstract: To improve the accuracy and timeliness of capturing pronunciation errors, the paper proposes a new method for capturing English oral pronunciation errors based on the speech recognition process. Using a voice production system to collect raw English spoken pronunciation signals and extract the features of the speech signals. Then, after determining the confidence level of the intonation points, hidden Markov model (HMM) classification algorithm is used to classify the intonation points and establish a spoken pronunciation comparison database containing standard state sequences. Finally, the degree component signal detection method is used to determine the spectral features of pronunciation errors. By comparing the spectral features with standard state sequences, incorrect English spoken pronunciation is captured. Experiment shows that the recognition accuracy of this method remains above 97%, and the maximum accuracy of capturing pronunciation errors can reach 98.74%. The capture time remains within 3s, indicating that this method has achieved the design expectations.
  Keywords: speech recognition; English speaking test; capture pronunciation errors; classification of intonation points; standard state sequence.
  DOI: 10.1504/IJCSM.2024.10068571
   
  
• Offshore wind power prediction based on chaotic optimisation PSO-SCN-LSTM model  
  by Xu Li, Lei Kou, Benfa Zhang, Zhen Wang, Jingya Wen, Fangfang Zhang, Jinyan Du, Yan Zhou, Wende Ke 
  Abstract: Offshore wind power, as a clean energy source, is receiving increasing attention worldwide. To enhance the economic and safety performance of offshore wind power, short-term forecasting of wind power is essential. This paper proposes a model based on chaos optimisation integrated with particle swarm optimisation (PSO), stochastic configuration network (SCN), and long short-term memory (LSTM) algorithm. Firstly, leveraging the randomness and ergodicity of the complex logistic chaos system, the collected power data from wind turbines is utilised as the input data source for the PSO, enhancing the randomness of the data. Subsequently, the SCN is employed to optimise the PSO, increasing the variation in the hidden layer during iterations and mitigating the PSO's tendency to fall into local optima, thereby obtaining initial prediction values. Finally, the mechanism model of the LSTM is utilised for secondary prediction, further improving prediction accuracy. Compared with traditional algorithms, the optimised algorithm significantly reduces errors and enhances prediction precision.
  Keywords: chaos theory; particle swarm optimisation algorithm; random configuration network; LSTM prediction model.
  DOI: 10.1504/IJCSM.2024.10068563
   
  
• Multi-objective optimisation of cigarette production planning and inventory management  
  by Wanjiang Wang, Feng Zhao, Mingjun Wang, Qi Sun, Huihui Gao, Wei Jian, Renwang Li, Shulan Luo 
  Abstract: This paper proposes a dynamic model-based optimisation method for cigarette production planning to meet market demand and optimise inventory management. The model contains three objectives: Minimising the difference between annual demand and production output, balancing monthly production and demand, and minimising the average age of inventory. The model sets annual demand constraints, production capacity constraints, and considers seasonal adjustment and contingency inventory. The model is solved by a combination of linear programming and genetic algorithms. The results show that the optimised production plan can effectively reduce total costs, avoiding the risks of under-supply and over-stocking. The analysis shows that the planned monthly production quantity closely matches the actual allocation quantity, the inventory management is effective, and hence the ability to cope with peak demand is enhanced.
  Keywords: dynamic stocking model; cigarette production; linear programming; genetic algorithm; inventory management.
  DOI: 10.1504/IJCSM.2025.10069527
   
  
• Short-term power load prediction based on CNN-LSTM model  
  by Jiawen Chen, Chao Cai, Fangbin Yan, Jinfeng Liu 
  Abstract: The load forecasting of power system is to forecast the load of the system in a future period of time, considering the influence of historical load, economic condition, meteorological condition and social events. Therefore, a CNN-LSTM model is proposed to predict short-term power load fluctuations in the next few days on the basis of the original, in which the convolutional layer and pooling layer in convolutional neural network (CNN) are used to extract features and reduce dimensions, and then the reconstructed data output by CNN is forecasted. The experimental results show that the prediction accuracy and error of CNN-LSTM model are obviously better than that of long short-term memory (LSTM) model, which also shows that CNN-LSTM model is suitable for short-term power load data prediction.
  Keywords: short-term power load; CNN; convolutional neural network; LSTM; long short-term memory network; deep learning.
  DOI: 10.1504/IJCSM.2025.10070450