煤炭企业进行订单交易与生产调度时存在供需双方信息不完全、交易效率低、决策精准性低等问题，这些问题的出现给煤炭企业的高质量发展带来了困难。本文从煤炭企业的实际需求出发，分别建立煤价预测模型、购煤量预测模型、煤炭供需双侧博弈模型以及订单接受与生产调度决策模型，系统地研究不完全信息下煤炭订单交易与生产调度决策问题，为实现智能煤矿的订单科学交易与高效生产提供一种有效途径。具体研究内容如下：

不完全信息下煤炭供需双侧煤价与购煤量预测。在对传统煤炭供需双侧订单交易流程进行分析的基础上，将博弈技术与预测方法引入到煤炭交易过程，构建不完全信息下煤炭供需双侧订单交易决策机制。将决策方法应用到煤炭订单接受与生产调度过程，建立不完全信息下煤炭供需双侧订单交易与生产调度决策流程。运用ARIMA-BP和BiLSTM预测模型，分别实现煤价及购煤量的预测，为煤炭供需双方交易决策提供依据。

不完全信息下煤炭供需双侧订单交易决策研究。对不完全信息下煤炭供需双侧博弈问题进行分析，构建博弈机制，建立以煤炭企业与客户收益最大化为目标的不完全信息下煤炭供需双侧博弈模型。为求解此博弈决策问题，设计了基于遗传算法的博弈求解策略，以决策最优煤价与最优购煤量，形成客户订单。

煤炭企业订单接受与生产调度决策方法研究。对煤炭企业订单接受与生产调度决策问题进行深入分析，构建决策机制。以煤炭企业订单利润最大化为目标，综合考虑煤炭产能与延期惩罚成本，建立煤炭企业订单接受与生产调度决策模型，设计基于变邻域遗传算法的决策算法，通过改进初始种群、染色体编码、适应度函数以及选择、交叉和变异等操作，防止最优解的丢失，提高算法求解效率。

以煤炭企业S为实例对象，对煤炭交易与订单生产调度进行决策优化。以煤炭企业S及电厂SX历史数据为基础，分别运用ARIMA-BP和BiLSTM神经网络模型进行煤价及购煤量预测。建立煤炭企业S与电厂SX博弈模型，运用基于遗传算法的博弈求解策略进行求解最优煤价及购煤量，形成客户订单。最后建立订单接受与生产调度决策模型，利用改进的变邻域遗传算法对模型进行求解，得到的订单接受与生产调度方案可降低订单延期惩罚成本87.30%，可增加订单收益54.26%。

There are issues such as incomplete information between supply and demand, low transaction efficiency, and inaccurate decision-making when coal enterprises engage in order transactions and production scheduling. The emergence of these problems has brought challenges to the high-quality development of coal enterprises. Starting from the actual demand of coal enterprises, this paper establishes a coal price prediction model, a coal purchase quantity prediction model, and a coal supply and demand bilateral game model. An order acceptance and production scheduling decision-making model is proposed to systematically study the decision-making problems related to coal order transactions and production scheduling under incomplete information. To facilitate the efficient exchange of scientific information and streamline order production in smart coal mines. The specific research content is outlined as follows:

(1) Prediction of coal prices and coal purchase quantity under incomplete information. Based on an analysis of the traditional coal supply and demand bilateral order transaction process. Game technology and prediction methods are integrated into the coal trading process to establish a decision-making mechanism for coal supply and demand under incomplete information. The decision-making method is applied to the process of coal order acceptance and production scheduling to establish a decision-making process for order trading and production scheduling for both coal supply and demand sides under incomplete information. The ARIMA-BP and BiLSTM prediction models are utilized to forecast coal prices and coal purchase quantities, providing a basis for decision-making regarding coal supply and demand.

(2) Research on order transaction decision-making of coal supply and demand under incomplete information. This paper analyzes the bilateral game of coal supply and demand under incomplete information and constructs a game mechanism. The study establishes a bilateral game model of coal supply and demand under incomplete information to maximize the income of coal enterprises and customers. To solve this game decision-making problem, a game-solving strategy based on a genetic algorithm is designed to determine the optimal coal price and the optimal coal purchase quantity for forming customer orders.

(3) Research on decision-making methods for order acceptance and production scheduling in coal enterprises. In-depth analysis of order acceptance and production scheduling decision-making problems in coal enterprises and construction of a decision-making mechanism. To maximize the profit of coal enterprises, it is essential to consider both the coal production capacity and the cost of delay penalties. The decision-making model for order acceptance and production scheduling in coal enterprises has been established. A decision-making algorithm based on a variable neighborhood genetic algorithm has been designed. To prevent the loss of the optimal solution and improve the efficiency of the algorithm, enhancements can be made to the initial population, chromosome coding, fitness function, selection process, crossover operation, and mutation phase.

(4) Taking coal enterprise S as an example, the optimization of decision-making in coal trading and order production scheduling is being conducted. Based on the historical data of coal enterprise S and power plant SX, ARIMA-BP and BiLSTM neural network models are used to predict coal prices and coal purchase quantity, respectively. The game model of the coal enterprise S and power plant SX has been established. The game-solving strategy based on a genetic algorithm is applied to determine the optimal coal price and quantity for purchasing coal to fulfill customer orders. Finally, the decision-making model for order acceptance and production scheduling has been established, and the improved variable neighborhood genetic algorithm is employed to solve the model. The developed order acceptance and production scheduling scheme can reduce the penalty cost of order delays by 87.30% and increase order income by 54.26%.