پیش‌بینی ریسک مالی شرکت‌ها بر اساس الگوریتم ترکیبی یادگیری عمیق FA-PSO-LSTM

میرمعینی, فاطمه السادات; قاسمیه, رحیم; سینایی, حسنعلی

پیش‌بینی ریسک مالی شرکت‌ها بر اساس الگوریتم ترکیبی یادگیری عمیق FA-PSO-LSTM

نوع مقاله : مقاله پژوهشی

نویسندگان

فاطمه السادات میرمعینی ¹

رحیم قاسمیه ²

حسنعلی سینایی ³

¹ کارشناسی ارشد مدیریت مالی، دانشکده اقتصاد و علوم اجتماعی، دانشگاه شهید چمران اهواز. اهواز . ایران

² گروه مدیریت- دانشکده اقتصاد و علوم اجتماعی- دانشگاه شهید چمران اهواز،اهواز، ایران

³ استاد گروه مدیریت، دانشکده اقتصاد و علوم اجتماعی، دانشکاه شهید چمران هواز، اهواز، ایران

چکیده

این پژوهش با توجه به لزوم نظارت مستمر یر داده های مالی و با تمرکز بر الگوریتم های هوش مصنوعی طی یک مطالعه تجربی و میدانی و با استفاده از با استفاده از یک مدل جدید مورد، داده های شرکت های فعال در سه صنعت فلزات اساسی، خودروسازی و فرآورده های نفتی را در یک بازه زمانی ده ساله موردارزیابی دقیق قرار داده است. پس از تعیین عوامل موثر مالی حاصل از تحلیل عاملی، از خطای میانگین مربعات و مقادیر پیش بینی شده شبکه عصبی LSTM جهت بهینه سازی تابع الگوریتم ازدحام ذرات و بهینه سازی نرخ یادگیری و تعداد لایه های پنهان شبکه های عصبی استفاده گردید. مدل یادگیری عمیق FA-PSO-LSTM مورد استفاده یک مدل ابتکاری و نسبتا جدید است که از مزایای شبکه LSTM در پردازش سری های زمانی به طور کامل می تواند بهره مند شده و به تکامل نطریه پردازی در این خصوص منتهی شود. نتایج حاکی از آن است که مدل پیشنهادی در صنعت فرآورده‌های نفتی با دقت بالا ریسک مالی را پیش‌بینی نموده و اهمیت متغیرهایی چون نقدینگی، جریان نقدی و سودآوری را برجسته ساخته است. در صنعت خودروسازی، شاخص‌هایی مانند نقدینگی، ظرفیت عملیاتی و توسعه پایدار اثرگذارتر بودند، در حالی که در صنعت فلزات اساسی بیشترین ثبات داده‌ها مشاهده و بهترین عملکرد مدل ثبت شد.در مجموع، متغیرهای نقدینگی، سودآوری، جریان نقدی، ظرفیت عملیاتی و قابلیت توسعه به‌عنوان عوامل کلیدی مشترک ریسک شناسایی گردیدند همچنین مشخص شد که که مدل پیشنهادی در مقایسه با روش‌های سنتی و سایر الگوریتم‌ها، دقت و کارایی بالاتری دارد.

کلیدواژه‌ها

ریسک مالی

یادگیری عمیق

شرکت‌های سهامی عام

بهینه سازی

20.1001.1.27171809.1404.6.4.1.1

موضوعات

مالی

عنوان مقاله English

Predicting the Financial Risk of Public Companies with a Hybrid Algorithm FA-PSO-LSTM

نویسندگان English

Fatemeh Sadat Mirmoeini ¹

Rahim Ghasemiyeh ²

Hasan Ali Sinaie ³

¹ MS in Financial Management. Faculty of Economics and Social Sciences, Shahid Chamran University of Ahvaz. Ahvaz. Iran

² Department of management, Faculty of Economics and Social Sciences, Ahvaz, Iran

³ Professor, Faculty of Economics and Social Sciences, Department of Management, Shahid Chamran university of Ahvaz, Ahvaz, Iran

چکیده English

This research, considering the need for continuous monitoring of financial data and focusing on artificial intelligence algorithms, has carefully evaluated the data of companies active in the three industries of basic metals, automotive, and petroleum products over a ten-year period using an experimental and field study and a new case model. After determining the effective financial factors from factor analysis, the mean square error and predicted values of the LSTM neural network were used to optimize the particle swarm algorithm function and optimize the learning rate and number of hidden layers of the neural networks. The FA-PSO-LSTM deep learning model used is an innovative and relatively new model that can fully benefit from the advantages of the LSTM network in time series processing and lead to the evolution of the theory in this regard. The results indicate that the proposed model predicts financial risk in the petroleum products industry with high accuracy and highlights the importance of variables such as liquidity, cash flow, and profitability. In the automotive industry, indicators such as liquidity, operational capacity, and sustainable development were more effective, while in the basic metals industry, the most data stability was observed and the best model performance was recorded. Overall, the variables of liquidity, profitability, cash flow, operational capacity, and development capability were identified as key common risk factors. It was also found that the proposed model has higher accuracy and efficiency compared to traditional methods and other algorithms..

کلیدواژه‌ها English

Financial Risk

Deep Learning

Publicly Traded Companies

Optimization

Alieva, B., Saparbayev, A., Zhanibekova, G., Noiyanov, M., & Kim, V. (2020). Methodology for managing financial risks of Kazakhstan enterprises. E3S Web of Conferences, 159, 04018. https://doi.org/10.1051/e3sconf/202015904018

Arefieva, O. V., Miagkyh, I. M., & Yashchuk, A. M. (2019). Challenges and pathways to improve company’s financial management by risk minimization. Bulletin of the Kyiv National University of Technologies and Design. Series: Economic Sciences. https://doi.org/10.30857/2413-0117.2019.1.5

Cai, M. (2025). Using Cuckoo Search Algorithm to Predict Corporate Financial Risks and Alleviate Economic Uncertainty. International Journal of Computational Intelligence Systems, 18(1), 216. https://link.springer.com/article/10.1007/s44196-025-00950-0

Chen, S. (2022). Cryptocurrency financial risk analysis based on deep machine learning. Complexity. https://doi.org/10.1155/2022/2611063

Chen, X., & Long, Z. (2023). E-Commerce enterprises financial risk prediction based on FA-PSO-LSTM neural network deep learning model. Sustainability, 15. https://doi.org/10.3390/su15075882

Chen, W. (2025). Enterprise financial risk prediction and intelligent early warning model based on deep learning. Discover Artificial Intelligence. https://doi.org/10.1007/s44163-025-00497-

Ghasemiyeh, R., Sinaei, H., & Dodange, A. R. (2024). A prediction-based portfolio optimization model using CNN neural network and MSAD criterion in Tehran Stock Exchange. [In Persian] Budget and Finance Strategic Research, 5(3), 11-3 DOR:20.1001.1.27171809.1403.5.3.1.2

Ghasemiyeh, R., Sinaei, H., & Ghalambor Dezfuli, E. (2023). A method based on wavelet denoising and DTW algorithm for stock price pattern recognition in Tehran stock exchange. Quietly Journal of quantitative Economics, 20(4), pp1-27. https://doi.org/10.22055/jqe.2023.42285.2521

Khalid, S., Khan, M. A., Mazliham, M. S., Alam, M. M., Aman, N., Taj, M. T., Zaka, R., & Jehangir, M. (2022). Predicting risk through artificial intelligence based on machine learning algorithms: A case of Pakistani nonfinancial firms. Complexity. https://doi.org/10.1155/2022/6858916

Mirzaei-Najad, M., Radfar, R., & Fathi, K. (2021). Designing a dynamic model of Iranian financial markets using system dynamics [In Persian]. Journal of Management Accounting. https://www.sid.ir/fileserver/jf/4003714005101.pdf

Mohammadi, M. (2024). Investigating the effect of risk-taking behavior and financial crisis on earnings quality and financial performance in companies listed on the Tehran Stock Exchange [In Persian]. Accounting and Auditing Research. https://www.iaaaar.com/asrticle_206115.html

Mosala, M., & Moghaddam, F. (2020). Investigating the effect of financial leverage on the risk and return of shares of companies listed on the Tehran Stock Exchange [In Persian]. Journal of Financial Research. https://www.majournal.ir/index.php/ma/article/view/356

Moghadamnia, E., & Hosseini, Z. S. (2025). Analysis of the organizational performance of National Iranian Oil Company — exploration management through the effectiveness of financial management measures of the organization [In Persian]. Budget and Finance Strategic Research. DOR:20.1001.1.27171809.1404.6.2.6.2

Qin, R., & Chen, M. (2022). The construction of corporate financial management risk model based on XGBoost algorithm. Journal of Mathematics. https://doi.org/10.1155/2022/2043369

Raei, R., Fallahpour, S., & Amery Matin, H. (2023). Financial risk assessment model for LNG projects: Case study — Iran LNG project [In Persian]. Journal of Financial. https://doi.org/10.22059/jfr.2013.51058

Salmanian, M., Vakilifard, H., Hamidian, M., Sarraf, F., & Darabi, R. (2021). Predicting financial constraints of companies listed on the Tehran Stock Exchange using Relief-SVM-Caiid methods [In Persian]. Investment Knowledge. http://www.jik-ifea.ir/article_17096.html?lang=en

Siami-Namini, S., & Namin, A. S. (2018). Forecasting economics and financial time series: ARIMA vs. LSTM. arXiv. https://doi.org/10.48550/arXiv.1803.06386

Sun, X., Shen, H., Bi, W., Xu, B., Wang, Z., & Cheng, X. (2022). Company-as-Tribe: Company financial risk assessment on tribe-style graph with hierarchical graph neural networks. In KDD ’22 Proceedings. https://doi.org/10.1145/3534678.3539129

Tavakoli, S., & Ashtab, A. (2021). Identifying the factors affecting the financial risks of companies using the structural equations approach [In Persian]. Journal of Financial Studies. https://doi.org/10.22108/amf.2021.127996.1647

Tavakoli, S., & Ashtab, A. (2023). Comparison of the effectiveness of machine learning models and statistical models in predicting financial risk [In Persian]. Journal of Financial Management. https://doi.org/10.22108/amf.2023.136720.1784

Wang, C. (2025). Corporate financial distress prediction with multiperiod annual report data: A fusion deep neural network model. PLOS ONE. https://doi.org/10.1371/jousrnal.pone.0333064

Yang, D., Ma, H., Chen, X., Liu, L., & Lang, Y. (2022). Design of financial risk control model based on deep learning neural network. Computational Intelligence and Neuroscience. https://doi.org/10.1155/2022/5842039

Zahir, M., Hamidian, M., & Darabi, R. (2024). Designing a conceptual model of factors affecting on financial health in life cycle of companies by using the Foundation’s data approach [In Persian]. Budget and Finance Strategic Research, 5(2), 45-76.

DOR: 20.1001.1.27171809.1403.5.2.2.1

Zhai, M. (2021). Risk prediction and response strategies in corporate financial management based on optimized BP neural network. Complexity. https://doi.org/10.1155/2021/9973377