[Objective] The price volatility of vegetables has profound implications for both farmers and consumers. Fluctuating prices directly impact farmers' earnings and pose challenges to market stability and consumer purchasing behaviors. These fluctuations are driven by a multitude of complex and interrelated factors, including supply and demand, seasonal cycles, climatic conditions, logistical efficiency, government policies, consumer preferences, and suppliers' trading strategies. As a result, vegetable prices tend to exhibit nonlinear and non-stationary patterns, which significantly complicate efforts to produce accurate price forecasts. Addressing these forecasting challenges holds considerable practical and theoretical value, as improved prediction models can support more stable agricultural markets, secure farmers' incomes, reduce cost-of-living volatility for consumers, and inform more precise and effective government regulatory strategies. [Methods] The study investigated the application of neural network-based time series forecasting models for the prediction of vegetable prices. In particular, a selection of state-of-the-art neural network architectures was evaluated for their effectiveness in modeling the complex dynamics of vegetable pricing. The selected models for the research included PatchTST and iTransformer, both of which were built upon the Transformer architecture, as well as SOFTS and TiDE, which leveraged multi-layer perceptron (MLP) structures. In addition, Time-LLM, a model based on a large language model architecture, was incorporated to assess its adaptability to temporal data characterized by irregularity and noise. To enhance the predictive performance and robustness of these models, an automatic hyperparameter optimization algorithm was employed. This algorithm systematically adjusted key hyperparameters such as learning rate, batch size, early stopping, and random seed. It utilized probabilistic modeling techniques to construct performance-informed distributions for guiding the selection of more effective hyperparameter configurations. Through iterative updates informed by prior evaluation data, the optimization algorithm increased the search efficiency in high-dimensional parameter spaces, while simultaneously minimizing computational costs. The training and validation process allocated 80% of the data to the training set and 20% to the validation set, and employed the mean absolute error (MAE) as the primary loss function. In addition to the neural network models, the study incorporated a traditional statistical model, the autoregressive integrated moving average (ARIMA), as a baseline model for performance comparison. The predictive accuracy of all models was assessed using three widely recognized error metrics: MAE, mean absolute percentage error (MAPE), and mean squared error (MSE). The model that achieved the most favorable performance across these metrics was selected for final vegetable price forecasting. [Results and Discussions] The experimental design of the study focused on four high-demand, commonly consumed vegetables: carrots, white radishes, eggplants, and iceberg lettuce. Both daily and weekly price forecasting tasks were conducted for each type of vegetable. The empirical results demonstrated that the neural network-based time series models provided strong fitting capabilities and produced accurate forecasts for vegetable prices. The integration of automatic hyperparameter tuning significantly improved the performance of these models. In particular, after tuning, the MSE for daily price prediction decreased by at least 76.3% for carrots, 94.7% for white radishes, and 74.8% for eggplants. Similarly, for weekly price predictions, the MSE reductions were at least 85.6%, 93.6%, and 64.0%, respectively, for the same three vegetables. These findings confirm the substantial contribution of the hyperparameter optimization process to enhancing model effectiveness. Further analysis revealed that neural network models performed better on vegetables with relatively stable price trends, indicating that the underlying consistency in data patterns benefited predictive modeling. On the other hand, Time-LLM exhibited stronger performance in weekly price forecasts involving more erratic and volatile price movements. Its robustness in handling time series data with high degrees of randomness suggests that model architecture selection should be closely aligned with the specific characteristics of the target data. Ultimately, the study identified the best-performing model for each vegetable and each prediction frequency. The results demonstrated the generalizability of the proposed approach, as well as its effectiveness across diverse datasets. By aligning model architecture with data attributes and integrating targeted hyperparameter optimization, the research achieved reliable and accurate forecasts. [Conclusions] The study verified the utility of neural network-based time series models for forecasting vegetable prices. The integration of automatic hyperparameter optimization techniques notably improved predictive accuracy, thereby enhancing the practical utility of these models in real-world agricultural settings. The findings provide technical support for intelligent agricultural price forecasting and serve as a methodological reference for predicting prices of other agricultural commodities. Future research may further improve model performance by integrating multi-source heterogeneous data. In addition, the application potential of more advanced deep learning models can be further explored in the field of price prediction.
[Objective] In the advancement of intensive agriculture, the contradiction between soil degradation and the demand for large-scale production has become increasingly pronounced, particularly in the core region of black soil in Northeast China. Long-term single-cropping patterns have caused soil structure damage and nutrient imbalance, severely threatening agricultural sustainability. Intensive rice cultivation has led to significant soil degradation, while the city must also balance national soybean planting mandates with large-scale production efficiency. However, existing planting planning methods predominantly focus on area optimization at the regional scale, lacking fine-grained characterization of plot-level spatial distribution, which easily results in fragmented layouts. Against this backdrop, a plot-scale multi-objective planting planning approach is developed to synergistically optimize contiguous crop distribution, soil restoration, practical production, and economic benefits, while ensuring national soybean planting tasks. This approach bridges macro-policy guidance and micro-production practices, providing scientific decision support for planting structure optimization and high-standard farmland construction in major grain-producing areas of Northeast China. [Methods] The multi-objective optimization model was established within a genetic algorithm framework, integrating connected component analysis to address plot-level spatial layout challenges. The model incorporated five indicators: economic benefit, soybean planting area, contiguous planting, crop rotation benefits, and the number of paddy-dryland conversions. The economic benefit objective was quantified by calculating the total income of crop combinations across all plots. A rigid threshold for soybean planting area was set to fulfill national mandates. The contiguous planting was evaluated using a connected-component-based method. The crop rotation benefits were scored according to predefined rotation rules. The paddy-dryland conversions were determined by counting changes in plot attributes. The model employed linear weighted summation to transform multi-objectives into a single objective for solution, generated high-quality initial populations via Latin Hypercube Sampling, and enhanced algorithm performance through connected-component-based crossover strategies and hybrid mutation strategies. Specifically, the crossover strategy was constructed based on connected component analysis: Adjacent plots with the same crop were divided into connected regions, and partial regions were randomly selected for crop gene exchange between parent generations, ensuring that the offspring inherited spatial coherence from parents, avoiding layout fragmentation caused by traditional crossover, and improving the rationality of contiguous planting. The mutation strategies included three types: Soybean threshold guarantee, plot-based crop rotation rule adaptation, and connected components-based crop rotation rule adaptation, which synergistically ensured mutation diversity and policy objective adaptability. Taking the Fujin city, Heilongjiang province—a crucial national commercial grain base—as an example, optimization was implemented using the distributed evolutionary algorithms in python (DEAP) library and validated through the simulation results of the four-year planting plan from 2020 to 2023. [Results and Discussions] Four years of simulation results demonstrated significant multi-objective balance in the optimized scheme. The contiguity index increased sharply from 0.477 in 2019 to 0.896 in 2020 and stabilized above 0.9 in subsequent years, effectively alleviating plot fragmentation and enhancing the feasibility of large-scale production. The economic benefits remained dynamically stable without significant decline, verifying the model's effectiveness in safeguarding economic efficiency. The soybean planting area stably met national thresholds while achieving strategic expansion, strengthening food security. The simulation results of crop rotation benefits reached 0.998 in 2023, indicating effective promotion of scientific rotation patterns and enhanced soil health and sustainable production capacity. The optimization objective of minimizing paddy-dryland conversions took practical production factors into account, achieving a good balance with crop rotation benefits and reflecting effective consideration of real-world production constraints. The evolutionary convergence curve showed the algorithm converged near the optimal solution, validating its convergence stability for this problem. In comparative experiments, this method outperformed traditional plot-based strategies in all optimization indicators except soybean planting area. Compared with the nondominated sorting genetic algorithm-Ⅱ (NSGA-II) multi-objective algorithm, it showed significant advantages in contiguous planting and crop rotation benefits. Although minor gaps existed in economic benefits and paddy-dryland conversions compared to NSGA-II, the planting layout was more regular and less fragmented. [Conclusions] The multi-objective planting planning method based on connected components and genetic algorithms proposed in this study bridges macro policies and micro layouts, effectively balancing black soil protection and production benefits through intelligent algorithms. By embedding spatial topology constraints into genetic operations, it solves the fragmentation problem in traditional methods while adapting to policy-driven planting scenarios via single-objective weighting strategies. Four years of simulations and comparative experiments show that this method significantly improves contiguous planting, ensures soybean production, stabilizes economic benefits, optimizes rotation patterns, and reduces paddy-dryland conversions, providing a scientific and feasible planning scheme for agricultural production. Future research can be expanded in three directions. First, further optimizing genetic algorithm parameters and introducing technologies such as deep reinforcement learning to enhance algorithm performance. Second, integrating multi-source heterogeneous data to build dynamic parameter systems and strengthen model generalization. Third, extending the method to more agricultural regions such as southern hilly areas, adjusting constraints according to local topography and crop characteristics to achieve broader application value. The research findings can provide decision support for planting structure optimization and high-standard farmland construction in major grain-producing areas of Northeast China.
