Coal enterprise-based coal type sales planning and sales time collaborative control method and system

Abstract

The application discloses a coal type sales planning and sales time collaborative control method and system based on a coal enterprise, and relates to the technical field of coal sales management.The application integrates multi-source data through a data fusion algorithm and other technical means, solves the problems of data dispersion and different formats, provides high-quality data support for subsequent analysis, accurately predicts sales trends and price fluctuations by using a long short-term memory neural network and a regression analysis model, helps enterprises optimize sales strategies, reduces inventory backlog risks, dynamically adjusts sales tasks by means of linear programming and reinforcement learning algorithms in combination with real-time data, improves market response speed and sales efficiency, and effectively improves the scientific nature and flexibility of coal enterprise sales management and enhances market competitiveness.

Description

A Method and System for Collaborative Management of Coal Type Sales Planning and Sales Time in Coal Enterprises Technical Field

[0001] This invention relates to the field of coal sales management technology, specifically to a method and system for the coordinated management of coal type sales planning and sales time in coal enterprises. Background Technology

[0002] Coal sales management is a crucial area in the energy industry, directly impacting a company's economic benefits and market competitiveness. With rapidly changing market demand and the diversification of coal types, scientifically formulating and efficiently executing sales plans has become key to improving profitability. Traditional sales management methods, relying primarily on manual experience and simple data statistics, are ill-suited to the complex and ever-changing market environment. These methods suffer from significant deficiencies in data processing capabilities, forecasting accuracy, and dynamic adjustments, resulting in imprecise and unrealistic sales strategies that often lead to missed market opportunities or inventory buildup.

[0003] Against this backdrop, the core challenge of coal sales management lies in how to effectively integrate and accurately predict multi-dimensional data. Coal enterprises need to collect data on coal type inventory, quality indicators, and market demand in real time. However, these data sources are scattered and have different formats, making integration difficult. Insufficient data integration directly affects the accuracy of sales trend and price fluctuation predictions. Without accurate prediction models, enterprises cannot formulate scientific sales plans, which further leads to a lack of basis for the decomposition and execution of sales tasks in the time dimension. During the execution process, the lack of real-time monitoring and dynamic adjustment mechanisms for market feedback and sales progress makes it impossible to optimize sales strategies in a timely manner, affecting overall sales efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for the coordinated management of coal type sales planning and sales time in coal enterprises. Through data fusion, intelligent prediction and dynamic adjustment, it solves the problems of difficult data integration, inaccurate prediction and lack of dynamic adjustment mechanism in coal sales management, and significantly improves the scientific nature of sales strategy and market responsiveness.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] This application provides a method for the coordinated management of coal type sales planning and sales time for coal enterprises, including the following steps:

[0007] A multi-dimensional dataset is obtained, and a data fusion algorithm is used, combining principal component analysis and cluster analysis to reduce the difficulty of data integration. Key features are extracted from multi-source heterogeneous data to obtain a fused feature dataset.

[0008] The multi-dimensional dataset includes structured and unstructured data collected from coal type inventory, quality indicators, market demand, and historical sales records.

[0009] Based on the fused feature dataset, a long short-term memory neural network model is used, with input time series data and market demand data, to train a sales trend prediction model to obtain the predicted trends of coal sales volume and price. Then, a regression analysis model is used to analyze the price fluctuation pattern and obtain the predicted range of price fluctuation.

[0010] Based on sales trend forecasts and price fluctuation forecast ranges, a linear programming algorithm is used to input coal quality indicators and inventory data from a multi-dimensional dataset to obtain a decomposition plan for daily sales tasks. Then, real-time sales progress data and market feedback data are obtained, and a deviation dataset for sales execution is obtained through time series analysis.

[0011] When the sales execution deviation dataset shows that the deviation exceeds a preset threshold, a reinforcement learning algorithm is used. The deviation dataset and market feedback data are input, and the sales task allocation is dynamically adjusted to obtain an optimized sales task adjustment plan. The deviation dataset and market feedback data are real-time dynamic data derived from a multi-dimensional dataset.

[0012] Based on the optimized sales task adjustment plan, a real-time data stream processing framework is adopted to update the real-time monitoring mechanism and obtain dynamically updated sales execution status. Then, a decision tree model is used, combined with inventory data and market demand data, to obtain the final inventory optimization and adjustment plan.

[0013] Furthermore, the predicted trends in coal sales volume and price are obtained, specifically including:

[0014] By standardizing and fusing the feature dataset, a unified format input data for time series data and market demand data is obtained. Then, a long short-term memory neural network is used to train the sales forecasting model by inputting the time series data and market demand data, and the model parameters are obtained.

[0015] When the loss function value of the prediction model is low, the predicted results of coal sales volume and sales price are output. When the loss function value is high, the model hyperparameters are adjusted and the model is retrained to obtain optimized model parameters.

[0016] Based on the forecast results, time series analysis was used to generate trends in coal sales volume and sales price. By extracting key fluctuation points in sales volume and price, the dynamic characteristics of the supply and demand relationship in the coal market were obtained.

[0017] Using visualization technology, dynamic charts of sales volume and price trends are generated, providing an intuitive display of predicted trends.

[0018] Furthermore, the predicted range for price fluctuations is obtained, specifically including:

[0019] When sales volume fluctuations exceed a preset threshold, historical price data and market demand data are retrieved from the database to obtain an initial dataset. Then, the initial dataset is processed using a feature selection method to filter out features that are highly correlated with price fluctuations, resulting in an optimized feature set.

[0020] A linear regression model is used to train the optimized feature set, the relationship between price fluctuations and features is analyzed, the trained model parameters are obtained, and then combined with recent market demand data, the price fluctuation pattern is calculated to obtain the fluctuation trend prediction value.

[0021] When the predicted value of the fluctuation trend exceeds the normal range, the predicted value is evaluated by cross-validation to obtain the confidence interval of the prediction result.

[0022] Based on the confidence interval of the prediction results, the fluctuation range is refined using the quantile regression method to obtain the final price fluctuation prediction range. Then, the final price fluctuation prediction range is verified by backtesting with historical data to obtain the verified prediction range.

[0023] Furthermore, a breakdown plan for the daily sales targets is obtained, specifically including:

[0024] Based on sales trend forecasts and price fluctuation forecast ranges, determine whether the trend forecast results indicate an increase in sales volume. If an increase is indicated, prioritize the allocation of high-quality coal inventory; if the forecast results indicate a decrease in sales volume, prioritize the allocation of low-quality coal inventory to determine the initial allocation strategy.

[0025] By using a linear programming algorithm, inputting a structured dataset, trend prediction results, and a preliminary allocation strategy, the inventory allocation scheme is optimized to obtain a daily inventory allocation plan.

[0026] Based on the daily inventory allocation plan and the predicted sales trend, the daily sales tasks are broken down to obtain a task allocation table. Then, the actual sales data is obtained and compared with the task allocation table. When the deviation exceeds the preset threshold, the parameters of the linear programming algorithm are adjusted, the inventory allocation scheme is re-optimized, and the daily sales tasks are decomposed again to obtain the final sales task plan.

[0027] Furthermore, a deviation dataset of sales execution is obtained, specifically including:

[0028] Obtain the sales task breakdown plan, extract daily task allocation data from the preset database, generate a task plan dataset using structured query language, and then collect sales progress data and market feedback data in real time from the sales management system and market feedback platform.

[0029] A time series analysis model is used to process sales progress data and market feedback data, extract the changing trends of sales progress and market feedback, and generate a trend feature dataset. When the sales progress in the trend feature dataset deviates from the task plan dataset by more than a preset threshold, it is marked as a deviation record, and a preliminary deviation dataset is obtained.

[0030] The initial deviation dataset is grouped by cluster analysis to identify the patterns and correlation characteristics of deviation occurrences, generating a categorized deviation dataset. Then, a decision tree model is used to predict the probability of future sales execution deviations, generating a deviation prediction dataset.

[0031] Furthermore, the optimized sales target adjustment plan specifically includes:

[0032] When the deviation value of the sales execution deviation dataset exceeds the preset threshold, the distribution characteristics of the deviation value are obtained from the deviation dataset, the mean and variance of the deviation value are calculated using statistical analysis methods, and then consumer behavior characteristics and sales response characteristics are extracted from the market feedback data. Principal component analysis is used to reduce the dimensionality of the features to obtain a simplified market feedback feature set.

[0033] When the feature dimensions of the simplified market feedback feature set meet the preset conditions, the deviation value distribution model and the market feedback feature set are input into the reinforcement learning algorithm, and the sales task allocation strategy is iteratively updated using the Q-learning method to obtain a preliminary task allocation adjustment plan.

[0034] Based on the preliminary task allocation adjustment plan, the simulation method is used to evaluate the performance of the adjustment plan in the virtual market environment, calculate the task allocation efficiency index, and obtain the evaluated allocation efficiency value.

[0035] If the evaluated allocation efficiency value does not reach the preset efficiency threshold, then the low-efficiency task allocation points are extracted from the simulation results, the gradient descent method is used to optimize the task allocation parameters, the adjusted deviation value is obtained from the sales execution deviation dataset, and the statistical test method is used to determine whether the deviation value is lower than the preset threshold, so as to obtain the final task allocation scheme.

[0036] When the deviation of the final task allocation scheme is lower than the preset threshold, the optimized task allocation scheme is saved through the data storage system, and the scheme is distributed to the sales execution system using an automated scheduling tool to complete the task allocation adjustment.

[0037] Furthermore, the sales execution status is dynamically updated, specifically including:

[0038] Through a real-time data stream processing framework, sales progress and market demand changes are continuously collected to obtain raw data streams. When the integrity of the raw data streams meets a preset threshold, streaming processing technology is used to analyze the sales progress and market demand changes to obtain structured data.

[0039] Based on structured data, time series analysis algorithms are used to detect the changing trends of sales progress and market demand, determine the changing characteristics, and when the changing characteristics exceed the preset fluctuation range, the parameters of the real-time monitoring mechanism are adjusted according to preset rules to obtain an updated monitoring model.

[0040] The updated monitoring model is used to analyze the sales execution status in the structured data to obtain the dynamic execution status. Then, the decision tree algorithm is used to optimize the sales task allocation strategy and determine the adjusted task plan.

[0041] Based on the revised task plan, the sales execution process was updated to achieve an optimized sales execution status.

[0042] Furthermore, the final inventory optimization and adjustment plan is obtained, which specifically includes:

[0043] Based on the dynamically updated sales execution status, a decision tree model is used for training. The input features include sales execution status, inventory data, and market demand data. The output is the judgment result of whether to adjust the inventory allocation strategy, and the trained decision tree model is obtained.

[0044] By using a trained decision tree model, predictions are made on the comprehensive dataset acquired in real time. When the prediction results indicate an adjustment to the inventory allocation strategy, inventory adjustment rules are generated to obtain a preliminary adjustment plan. Then, combined with the predicted demand trend, a linear regression model is used to analyze the changing trend of market demand data to obtain the demand change trend.

[0045] By analyzing demand trends, the initial adjustment plan is adjusted. If the trend indicates an increase in demand, the inventory allocation ratio of related products is increased to obtain an optimized inventory allocation plan. Then, specific inventory adjustment instructions are generated and executed through the inventory management system to obtain the final inventory optimization result.

[0046] Furthermore, before obtaining the fused feature dataset, the process includes: acquiring multi-dimensional data, collecting structured and unstructured data from coal type inventory, quality indicators, market demand, and historical sales records, storing the data in a distributed database, and processing the data with different formats through data cleaning and standardization to obtain a multi-dimensional dataset in a unified format.

[0047] This invention provides a coal type sales planning and sales time collaborative management system for coal enterprises, used to implement a method for coal type sales planning and sales time collaborative management in coal enterprises, including:

[0048] The data acquisition and preprocessing module collects structured and unstructured data from multi-dimensional data sources, including coal type inventory, quality indicators, market demand, and historical sales records of coal enterprises. Through data cleaning and standardization, it unifies the data of different formats and stores it in a distributed database.

[0049] The data fusion and feature extraction module processes the collected multi-source heterogeneous data, extracts key features, and generates a fused feature dataset.

[0050] The sales trend and price fluctuation prediction module combines long short-term memory neural networks and regression analysis models. It takes time series data and market demand data as input, trains the sales trend prediction model, obtains the predicted trend of coal sales volume and price, and then analyzes the price fluctuation pattern through historical price data and market demand data to obtain the predicted range of price fluctuation.

[0051] The inventory allocation and sales task decomposition module uses a linear programming algorithm based on sales trend forecasts and price fluctuation forecast ranges. It takes coal quality indicators and inventory data from a multi-dimensional dataset as input, optimizes the inventory allocation scheme, generates a daily sales task decomposition plan, and dynamically adjusts the sales task allocation through real-time sales progress data and market feedback data.

[0052] The sales execution deviation analysis and adjustment module monitors the sales execution progress in real time, determines whether the sales execution deviation exceeds a preset threshold through time series analysis, generates a sales execution deviation dataset, and when the deviation exceeds the threshold, uses a reinforcement learning algorithm, inputs the deviation dataset and market feedback data, dynamically adjusts the sales task allocation, and optimizes the sales task adjustment plan.

[0053] The real-time monitoring and inventory optimization module adopts a real-time data stream processing framework to continuously collect sales progress and market demand change data, update the real-time monitoring mechanism, obtain dynamically updated sales execution status, and combine inventory data and market demand data to determine whether to adjust the inventory allocation strategy through a decision tree model, generating the final inventory optimization and adjustment plan.

[0054] The beneficial effects of this invention are as follows:

[0055] By employing data fusion algorithms, principal component analysis, and cluster analysis, key features are extracted from multi-source heterogeneous data and a fused feature dataset is generated. This effectively reduces the difficulty of data integration, improves data processing efficiency and quality, and solves the problems of scattered data sources, inconsistent formats, and high integration difficulty in traditional coal sales management. It provides high-quality, structured data support for subsequent sales trend prediction and inventory optimization, significantly improves data interpretability and analysis efficiency, and ensures the scientific and accurate nature of sales strategies.

[0056] By combining Long Short-Term Memory (LSTM) neural networks and regression analysis models, and inputting time series data and market demand data, a sales trend prediction model is trained to obtain predicted trends for coal sales volume and prices. At the same time, by analyzing price fluctuation patterns through historical price data and market demand data, a more accurate price fluctuation prediction range is obtained. This solves the shortcomings of traditional methods in terms of data processing capabilities and prediction accuracy, improves the accuracy and reliability of predictions, and helps enterprises better grasp market opportunities, optimize sales strategies, and reduce inventory backlog and missed market opportunities.

[0057] By employing linear programming and reinforcement learning algorithms, combined with real-time sales progress data and market feedback data, the system dynamically adjusts sales task allocation, generates daily sales task breakdown plans, and monitors sales execution status in real time. When sales execution deviations exceed a set reasonable range, the system automatically adjusts the sales task allocation strategy to ensure the smooth execution of sales tasks. This solves the problem of traditional sales management methods lacking a dynamic adjustment mechanism in the process of sales task breakdown and execution, making sales task breakdown and inventory management more flexible and efficient. It enables timely responses to market changes, improves the company's market responsiveness and sales efficiency, and ensures real-time optimization and execution of sales strategies. Attached Figure Description

[0058] To better understand and implement this application, the technical solution is described in detail below with reference to the accompanying drawings.

[0059] Figure 1 is a flowchart illustrating the method for coordinated management of coal type sales planning and sales time in coal enterprises provided in Embodiment 1 of this application;

[0060] Figure 2 is a flowchart illustrating the process of obtaining the decomposition plan of daily sales tasks based on the coal type sales planning and sales time collaborative management method of coal enterprises provided in Embodiment 1 of this application.

[0061] Figure 3 is a flowchart illustrating the optimized sales task adjustment scheme based on the coal type sales planning and sales time collaborative management method of coal enterprises provided in Embodiment 1 of this application;

[0062] Figure 4 is a schematic diagram of the structure of the coal type sales planning and sales time collaborative management system for coal enterprises provided in Embodiment 2 of this application. Detailed Implementation

[0063] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, exemplary embodiments will be described in detail below, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0064] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0065] The following detailed description of the specific implementation methods, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided in detail.

[0066] Example 1

[0067] Please refer to Figures 1-3. This embodiment provides a method for coordinated management of coal type sales planning and sales time for coal enterprises, including the following steps:

[0068] S1. Obtain a multi-dimensional dataset, use a data fusion algorithm, combine principal component analysis and cluster analysis to reduce the difficulty of data integration, extract key features from multi-source heterogeneous data, and obtain the fused feature dataset.

[0069] The multi-dimensional dataset includes structured and unstructured data collected from coal type inventory, quality indicators, market demand, and historical sales records. After data cleaning and standardization, it is formatted into a unified format.

[0070] Coal inventory data is extracted periodically via the API interface of the Enterprise Resource Planning (ERP) system; quality indicator data is collected in real time through a sensor network deployed on production lines and warehouses; market demand data is updated daily through the API interface of industry data platforms (such as coal trading centers and bulk commodity information service providers); historical sales record data is exported periodically from the company's internal sales management database; the collection frequency is set according to data characteristics and business needs, with inventory and quality data updated every minute, and market demand and price data updated daily.

[0071] Furthermore, the fused feature dataset is obtained, specifically including:

[0072] The original dataset is obtained from multi-source heterogeneous data. Standardization is applied to unify the data format and dimensions to obtain a standardized dataset. When the data dimension exceeds a preset threshold, principal component analysis is used to calculate eigenvalues ​​and eigenvectors, retain the principal components, and obtain a dimensionality-reduced dataset.

[0073] Based on the dimensionality reduction dataset, cluster analysis is used to calculate the distance between data points, generate cluster groups, obtain grouped datasets, extract key features from each group dataset, and use a weighted average method to fuse the features within the group to obtain a preliminary fused feature set;

[0074] Based on the initial fusion feature set, when the features are highly correlated, redundant features are removed and independent features are retained to obtain an optimized feature set. Then, a data fusion algorithm is used to calculate the weighted combination of features to generate the final fusion feature dataset.

[0075] Feature vectors are extracted from the final fused feature dataset and stored in a structured format to obtain a fused feature dataset that can be used for subsequent analysis.

[0076] The merged feature dataset is a structured two-dimensional data table. Each record corresponds to a single coal type at a specific point in time (e.g., daily). Key fields include: coal type code, date, calorific value (MJ / kg), sulfur content (%), inventory (tons), regional demand index, recent average selling price (yuan / ton), inventory turnover rate (times / month), and principal component scores generated after dimensionality reduction by principal component analysis. This dataset is stored in a distributed data warehouse in columnar storage format.

[0077] Specifically, by using data fusion algorithms, principal component analysis, and cluster analysis, key features are extracted from multi-source heterogeneous data and a fused feature dataset is generated. This process effectively reduces the difficulty of data integration, improves data processing efficiency and quality, and provides high-quality, structured data support for coal enterprises' coal type sales planning and sales time collaborative management. This helps enterprises accurately grasp market dynamics and formulate scientific and reasonable sales strategies.

[0078] S2. Based on the fused feature dataset, a long short-term memory neural network model is used. Time series data and market demand data are input to train a sales trend prediction model to obtain the predicted trends of coal sales volume and price. Then, a regression analysis model is used to analyze the price fluctuation pattern by combining historical price data and market demand data to obtain the predicted range of price fluctuation.

[0079] Furthermore, the predicted trends in coal sales volume and price are obtained, specifically including:

[0080] By standardizing and fusing the feature dataset, a unified format input data for time series data and market demand data is obtained. Then, a long short-term memory neural network is used to train the sales forecasting model by inputting the time series data and market demand data, and the model parameters are obtained.

[0081] When the loss function value of the prediction model is low, the predicted results of coal sales volume and sales price are output. When the loss function value is high, the model hyperparameters are adjusted and the model is retrained to obtain optimized model parameters.

[0082] Based on the forecast results, time series analysis was used to generate trends in coal sales volume and sales price. By extracting key fluctuation points in sales volume and price, the dynamic characteristics of the supply and demand relationship in the coal market were obtained.

[0083] Using visualization technology, dynamic charts of sales volume and price trends are generated, providing an intuitive display of predicted trends.

[0084] Specifically, a Long Short-Term Memory (LSTM) neural network model is used to predict sales trends, while a regression analysis model is combined to predict price fluctuations. LSTM can effectively handle long-term dependencies in time series data and accurately capture the changing trends of sales volume and prices; while regression analysis can further analyze the price fluctuation patterns to obtain a more accurate price fluctuation prediction range. This approach of combining deep learning with traditional statistical analysis methods can give full play to the advantages of both and improve the accuracy and reliability of predictions.

[0085] Furthermore, the predicted range for price fluctuations is obtained, specifically including:

[0086] When sales volume fluctuations exceed a preset threshold, historical price data and market demand data are retrieved from the database to obtain an initial dataset. Then, the initial dataset is processed using a feature selection method to filter out features that are highly correlated with price fluctuations, resulting in an optimized feature set.

[0087] A linear regression model is used to train the optimized feature set, the relationship between price fluctuations and features is analyzed, the trained model parameters are obtained, and then combined with recent market demand data, the price fluctuation pattern is calculated to obtain the fluctuation trend prediction value.

[0088] When the predicted value of the fluctuation trend exceeds the normal range, the predicted value is evaluated by cross-validation to obtain the confidence interval of the prediction result.

[0089] Based on the confidence interval of the prediction results, the fluctuation range is refined using the quantile regression method to obtain the final price fluctuation prediction range. Then, the final price fluctuation prediction range is verified by backtesting with historical data to obtain the verified prediction range.

[0090] Specifically, coal market prices fluctuate significantly, making accurate forecasting of sales volume and prices crucial for companies' sales planning and profit maximization. For instance, before the peak season in the coal market, the S2 step allows companies to predict the upward trend in sales volume and the range of price fluctuations in advance. This enables them to prepare production plans and inventory in advance, and rationally arrange sales strategies, such as adjusting sales prices and optimizing sales channels, to obtain higher profits. Simultaneously, during the off-season or when prices fluctuate greatly, companies can also adjust their sales strategies promptly based on the forecast results, reducing sales risks and ensuring stable operations. This accurate ability to predict sales trends and price fluctuations helps companies better seize market opportunities and improve their market competitiveness and economic efficiency.

[0091] S3. Based on the sales trend forecast and price fluctuation forecast range, a linear programming algorithm is used to input coal quality indicators and inventory data from the multi-dimensional dataset, optimize the inventory allocation scheme, obtain the decomposition plan of the daily sales task, and then obtain real-time sales progress data and market feedback data. Through time series analysis, it is determined whether the sales execution deviation exceeds the preset threshold, and the sales execution deviation dataset is obtained.

[0092] Furthermore, a breakdown plan for the daily sales targets is obtained, specifically including:

[0093] S31. Based on the sales trend forecast and price fluctuation forecast range, determine whether the trend forecast results show that the sales volume is increasing. If it is increasing, prioritize the allocation of high-quality coal inventory. If the forecast results show that the sales volume is decreasing, prioritize the allocation of low-quality coal inventory to determine the initial allocation strategy.

[0094] S32. Using a linear programming algorithm, input the structured dataset, trend prediction results, and preliminary allocation strategy to optimize the inventory allocation scheme and obtain the daily inventory allocation plan.

[0095] S33. Based on the daily inventory allocation plan and the predicted sales trend, decompose the daily sales tasks to obtain a task allocation table. Then, obtain the actual sales data and compare it with the task allocation table. When the deviation exceeds the preset threshold, adjust the linear programming algorithm parameters, re-optimize the inventory allocation scheme, and then re-decompose the daily sales tasks to obtain the final sales task plan.

[0096] Furthermore, a deviation dataset of sales execution is obtained, specifically including:

[0097] Obtain the sales task breakdown plan, extract daily task allocation data from the preset database, generate a task plan dataset using structured query language, and then collect sales progress data and market feedback data in real time from the sales management system and market feedback platform.

[0098] A time series analysis model is used to process sales progress data and market feedback data, extract the changing trends of sales progress and market feedback, and generate a trend feature dataset. When the sales progress in the trend feature dataset deviates from the task plan dataset by more than a preset threshold, it is marked as a deviation record, and a preliminary deviation dataset is obtained.

[0099] The initial deviation dataset is grouped by cluster analysis to identify the patterns and correlation characteristics of deviation occurrences, generating a categorized deviation dataset. Then, a decision tree model is used to predict the probability of future sales execution deviations, generating a deviation prediction dataset.

[0100] This step integrates multiple factors, including sales trend forecasting, price fluctuation prediction ranges, coal quality indicators, and inventory data, optimizing them using linear programming algorithms to ensure the scientific and rational nature of the inventory allocation plan. It also incorporates real-time sales progress data and market feedback data, dynamically adjusting sales task allocation through time series analysis. This combination of multi-dimensional data-driven and dynamic adjustment mechanisms makes sales task decomposition and inventory management more flexible and efficient, enabling timely responses to market changes and ensuring the smooth execution of sales tasks, demonstrating the flexibility and adaptability of sales time management.

[0101] Specifically, in daily sales management, this step can reasonably break down sales tasks to each day based on market demand and inventory status, ensuring that sales tasks can be completed on time and in the required quantity. In terms of inventory management, by optimizing inventory allocation plans, it ensures that inventory levels match market demand, reducing inventory costs. Through real-time monitoring and dynamic adjustment mechanisms, enterprises can quickly respond to market changes, adjust sales strategies in a timely manner, ensure that customer orders can be delivered on time, and improve customer satisfaction.

[0102] S4. When the sales execution deviation dataset shows that the deviation exceeds the preset threshold, a reinforcement learning algorithm is used. The deviation dataset and market feedback data are input to dynamically adjust the sales task allocation and obtain an optimized sales task adjustment plan.

[0103] The deviation dataset and market feedback data are real-time dynamic data derived from multi-dimensional datasets, used for subsequent sales task adjustments.

[0104] Furthermore, the optimized sales target adjustment plan specifically includes:

[0105] S41. When the deviation value of the sales execution deviation dataset exceeds the preset threshold, the deviation value distribution characteristics are obtained from the deviation dataset. The mean and variance of the deviation values ​​are calculated using statistical analysis methods to obtain the deviation value distribution model. Then, consumer behavior characteristics and sales response characteristics are extracted from the market feedback data. The principal component analysis method is used to reduce the dimensionality of the characteristics to obtain a simplified market feedback feature set.

[0106] S42. When the feature dimensions of the simplified market feedback feature set meet the preset conditions, the deviation value distribution model and the market feedback feature set are input into the reinforcement learning algorithm, and the sales task allocation strategy is iteratively updated using the Q-learning method to obtain a preliminary task allocation adjustment plan.

[0107] S43. Based on the preliminary task allocation adjustment plan, use simulation methods to evaluate the performance of the adjustment plan in the virtual market environment, calculate the task allocation efficiency index, and obtain the evaluated allocation efficiency value.

[0108] S44. When the evaluated allocation efficiency value does not reach the preset efficiency threshold, the low-efficiency task allocation points are extracted from the simulation results, the task allocation parameters are optimized using the gradient descent method, the optimized task allocation adjustment scheme is obtained, the adjusted deviation value is obtained from the sales execution deviation dataset, and the statistical test method is used to determine whether the deviation value is lower than the preset threshold, so as to obtain the final task allocation scheme.

[0109] S45. When the deviation of the final task allocation scheme is lower than the preset threshold, the optimized task allocation scheme is saved through the data storage system, and the scheme is distributed to the sales execution system using an automated scheduling tool to complete the task allocation adjustment.

[0110] Reinforcement learning, a machine learning method that dynamically adjusts strategies based on environmental feedback, is used in this step to automatically learn and update sales task allocation strategies based on sales execution deviations and market feedback, achieving better sales results. This dynamic adjustment mechanism based on reinforcement learning makes sales task allocation more flexible and intelligent, quickly adapting to market changes and various situations that arise during the sales process, thereby effectively reducing sales deviations and improving the completion rate of sales tasks and the company's market responsiveness. The coal market experiences frequent price fluctuations and demand changes, making deviations easy to occur during sales. For example, when market demand suddenly drops, the original sales task allocation may no longer be applicable, leading to problems such as inventory backlog or poor sales. Through step S4, the system can automatically adjust sales tasks using reinforcement learning algorithms based on real-time sales execution deviations and market feedback, such as adjusting the sales ratio of different coal types, changing sales regions or customer groups, thereby better responding to market changes, ensuring the smooth completion of sales tasks, and improving the company's sales efficiency and market competitiveness.

[0111] S5. Based on the optimized sales task adjustment plan, a real-time data stream processing framework is used to continuously collect sales progress and market demand change data, update the real-time monitoring mechanism, obtain dynamically updated sales execution status, and then use a decision tree model, combined with inventory data and market demand data, to determine whether to adjust the inventory allocation strategy and obtain the final inventory optimization adjustment plan.

[0112] Furthermore, the sales execution status is dynamically updated, specifically including:

[0113] Through a real-time data stream processing framework, sales progress and market demand changes are continuously collected to obtain raw data streams. When the integrity of the raw data streams meets a preset threshold, streaming processing technology is used to analyze the sales progress and market demand changes to obtain structured data.

[0114] Based on structured data, time series analysis algorithms are used to detect the changing trends of sales progress and market demand, determine the changing characteristics, and when the changing characteristics exceed the preset fluctuation range, the parameters of the real-time monitoring mechanism are adjusted according to preset rules to obtain an updated monitoring model.

[0115] The updated monitoring model is used to analyze the sales execution status in the structured data to obtain the dynamic execution status. Then, the decision tree algorithm is used to optimize the sales task allocation strategy and determine the adjusted task plan.

[0116] Based on the revised task plan, the sales execution process was updated to achieve an optimized sales execution status.

[0117] Furthermore, the final inventory optimization and adjustment plan is obtained, which specifically includes:

[0118] Based on the dynamically updated sales execution status, a decision tree model is used for training. The input features include sales execution status, inventory data, and market demand data. The output is the judgment result of whether to adjust the inventory allocation strategy, and the trained decision tree model is obtained.

[0119] By using a trained decision tree model, predictions are made on the comprehensive dataset acquired in real time. When the prediction results indicate an adjustment to the inventory allocation strategy, inventory adjustment rules are generated to obtain a preliminary adjustment plan. Then, combined with the predicted demand trend, a linear regression model is used to analyze the changing trend of market demand data to obtain the demand change trend.

[0120] By analyzing demand trends, the initial adjustment plan is adjusted. If the trend indicates an increase in demand, the inventory allocation ratio of related products is increased to obtain an optimized inventory allocation plan. Then, specific inventory adjustment instructions are generated and executed through the inventory management system to obtain the final inventory optimization result.

[0121] The real-time data stream processing framework continuously collects data on sales progress and changes in market demand, and updates sales execution status in real time. This real-time monitoring mechanism enables enterprises to keep abreast of market dynamics and sales execution, respond quickly to market changes, and avoid decision-making errors caused by information lag. Compared with traditional periodic data collection and analysis methods, real-time data stream processing can provide more timely and accurate information, providing strong support for dynamic decision-making by enterprises.

[0122] By combining inventory data and market demand data, a decision tree model is used to determine whether to adjust the inventory allocation strategy and generate a final inventory optimization adjustment plan. This approach, which combines real-time monitoring with inventory optimization, enables dynamic adjustment of inventory management, ensuring that inventory levels match market demand and reducing inventory backlogs or stockouts. Furthermore, the introduction of the decision tree model makes inventory adjustment decisions more scientific and rational, automatically formulating the optimal inventory allocation plan based on different market conditions and inventory statuses.

[0123] Specifically, in the daily operations of coal enterprises, changes in market demand and sales progress are commonplace. For example, when coal demand in a certain region suddenly increases, the enterprise needs to adjust its inventory allocation promptly, allocating more coal to that region to meet market demand. Through the S5 process, enterprises can quickly obtain this market change information using a real-time data stream processing framework, and analyze it in conjunction with inventory data using a decision tree model. This allows for timely adjustments to inventory allocation strategies, ensuring that coal is supplied to regions with high demand in a timely and accurate manner, improving customer satisfaction, and avoiding economic losses due to inventory backlog or shortages. Furthermore, this dynamic inventory optimization mechanism can help enterprises better cope with market fluctuations, reduce inventory costs, and improve operational efficiency and economic benefits.

[0124] Furthermore, before obtaining the fused feature dataset, the process includes: acquiring multi-dimensional data, collecting structured and unstructured data from coal type inventory, quality indicators, market demand, and historical sales records, storing the data in a distributed database, and, for cases with different data formats, obtaining a unified format multi-dimensional dataset through data cleaning and standardization.

[0125] Example 2

[0126] Please refer to Figure 4. This embodiment provides a coal type sales planning and sales time collaborative management system for coal enterprises, used to implement a method for coal type sales planning and sales time collaborative management for coal enterprises, including:

[0127] The data acquisition and preprocessing module collects structured and unstructured data from multi-dimensional data sources, including coal type inventory, quality indicators, market demand, and historical sales records of coal enterprises. Through data cleaning and standardization, it unifies the data of different formats and stores it in a distributed database, providing a high-quality data foundation for subsequent data fusion and analysis, and ensuring the accuracy and usability of the data.

[0128] The data fusion and feature extraction module processes the collected multi-source heterogeneous data, extracts key features, and generates a fused feature dataset. Through steps such as standardization, dimensionality reduction, clustering, and feature fusion, it reduces the difficulty of data integration, improves data processing efficiency and quality, provides structured data support for sales trend forecasting and inventory optimization, and enhances data interpretability and analysis efficiency.

[0129] The sales trend and price fluctuation prediction module combines a long short-term memory neural network (LSTM) and a regression analysis model. It takes time series data and market demand data as input, trains the sales trend prediction model, and obtains the predicted trends of coal sales volume and price. Then, by analyzing the price fluctuation patterns through historical price data and market demand data, it obtains the predicted range of price fluctuations, providing a scientific basis for the formulation of sales strategies, helping enterprises to make production plans and inventory preparations in advance, and improve market competitiveness.

[0130] The inventory allocation and sales task decomposition module, based on sales trend forecasts and price fluctuation forecast ranges, uses a linear programming algorithm, inputs coal quality indicators and inventory data from a multi-dimensional dataset, optimizes the inventory allocation scheme, generates a daily sales task decomposition plan, and dynamically adjusts the sales task allocation through real-time sales progress data and market feedback data to ensure the smooth execution of sales tasks, improve the flexibility and adaptability of sales time management, reduce inventory costs, and improve operational efficiency.

[0131] The sales execution deviation analysis and adjustment module monitors sales execution progress in real time, uses time series analysis to determine whether sales execution deviation exceeds a preset threshold, generates a sales execution deviation dataset, and when the deviation exceeds the threshold, uses a reinforcement learning algorithm, inputs the deviation dataset and market feedback data, dynamically adjusts the allocation of sales tasks, optimizes the sales task adjustment plan, ensures the efficient completion of sales tasks, reduces sales risks, and improves the company's market responsiveness.

[0132] The real-time monitoring and inventory optimization module adopts a real-time data stream processing framework to continuously collect sales progress and market demand change data, update the real-time monitoring mechanism, obtain dynamically updated sales execution status, and combine inventory data and market demand data to determine whether to adjust the inventory allocation strategy through a decision tree model. This generates the final inventory optimization and adjustment plan to ensure that inventory levels match market demand, reduce inventory costs, improve operational efficiency, and enhance the company's market adaptability.

[0133] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for coordinated management of coal type sales planning and sales time in coal enterprises, characterized by: The process includes the following steps: First, acquiring a multi-dimensional dataset. Second, employing a data fusion algorithm, combining principal component analysis and cluster analysis to reduce the difficulty of data integration. Third, extracting key features from multi-source heterogeneous data to obtain a fused feature dataset. The multi-dimensional dataset includes structured and unstructured data collected from coal type inventory, quality indicators, market demand, and historical sales records. Fourth, using the fused feature dataset, a long short-term memory neural network model is employed, inputting time series data and market demand data to train a sales trend prediction model, obtaining predicted trends for coal sales volume and price. Then, a regression analysis model is used to analyze price fluctuation patterns and obtain the predicted range for price fluctuations. Fifth, based on the sales trend prediction and the predicted range for price fluctuations, a linear programming algorithm is used, inputting coal type quality indicators and inventory data from the multi-dimensional dataset, to obtain a decomposition plan for daily sales tasks. Finally, real-time sales progress data and market feedback data are acquired, and through time series analysis, a deviation dataset for sales execution is obtained. The decomposition plan for daily sales tasks includes: determining whether the sales volume shows an increase based on the sales trend prediction and the predicted range for price fluctuations; if so, prioritizing the allocation of high-quality coal types. For inventory management, when the forecast indicates a decline in sales volume, priority is given to allocating low-quality coal inventory to determine the initial allocation strategy. A linear programming algorithm is then used, inputting a structured dataset, trend forecast results, and the initial allocation strategy, to optimize the inventory allocation scheme and obtain a daily inventory allocation plan. Based on the daily inventory allocation plan and the predicted sales trend, daily sales tasks are decomposed to obtain a task allocation table. Actual sales data is then compared with the task allocation table. If the deviation exceeds a preset threshold, the linear programming algorithm parameters are adjusted, the inventory allocation scheme is re-optimized, and the daily sales tasks are decomposed again to obtain the final sales task plan. When the sales execution deviation dataset shows a deviation exceeding a preset threshold, a reinforcement learning algorithm is used, inputting the deviation dataset and market feedback data, to dynamically adjust the sales task allocation and obtain an optimized sales task adjustment scheme. The deviation dataset and market feedback data are real-time dynamic data derived from a multi-dimensional dataset. Based on the optimized sales task adjustment scheme, a real-time data stream processing framework is used to update the real-time monitoring mechanism, obtaining dynamically updated sales execution status. Finally, a decision tree model is used, combined with inventory data and market demand data, to obtain the final optimized inventory adjustment scheme.

2. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: The process of obtaining predicted trends for coal sales volume and prices involves: standardizing and fusing feature datasets to obtain input data in a unified format for time-series data and market demand data; then training a sales forecasting model using a Long Short-Term Memory (LSTM) neural network with the input time-series data and market demand data to obtain model parameters; outputting predicted coal sales volume and prices when the loss function value of the forecasting model is low, and adjusting the model hyperparameters and retraining to obtain optimized model parameters when the loss function value is high; generating trends in coal sales volume and prices using time-series analysis methods based on the forecast results, extracting key fluctuation points in sales volume and prices to obtain dynamic characteristics of the coal market supply and demand relationship; and finally, generating dynamic charts of sales volume and price trends using visualization technology to provide an intuitive display of the predicted trends.

3. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: The process of obtaining the predicted price fluctuation range specifically includes: when sales volume fluctuations exceed a preset threshold, historical price data and market demand data are retrieved from the database to obtain an initial dataset. This initial dataset is then processed using feature selection methods to filter out features highly correlated with price fluctuations, resulting in an optimized feature set. A linear regression model is used to train the optimized feature set, analyzing the relationship between price fluctuations and the features to obtain the trained model parameters. These parameters are then combined with recent market demand data to calculate the price fluctuation pattern and obtain a predicted fluctuation trend value. When the predicted fluctuation trend value exceeds the normal range, cross-validation is used to evaluate the predicted value, obtaining a confidence interval for the prediction result. Based on the confidence interval of the prediction result, quantile regression is used to refine the fluctuation range, resulting in the final price fluctuation prediction range. Finally, backtesting with historical data is used to validate the final price fluctuation prediction range, yielding the validated prediction range.

4. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: The process of obtaining a sales execution deviation dataset specifically includes: acquiring a sales task breakdown plan; extracting daily task allocation data from a pre-defined database; generating a task plan dataset using structured query language; and collecting real-time sales progress data and market feedback data from the sales management system and market feedback platform. A time series analysis model is then used to process the sales progress and market feedback data, extracting trends in sales progress and market feedback to generate a trend feature dataset. When the sales progress in the trend feature dataset deviates from the task plan dataset by more than a pre-defined threshold, it is marked as a deviation record, resulting in a preliminary deviation dataset. Cluster analysis is then used to group the preliminary deviation dataset, identifying patterns and correlation characteristics of deviation occurrences to generate a categorized deviation dataset. Finally, a decision tree model is used to predict the likelihood of future sales execution deviations, generating a deviation prediction dataset.

5. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: The optimized sales task adjustment plan includes the following steps: When the deviation value of the sales execution deviation dataset exceeds a preset threshold, the deviation value distribution characteristics are obtained from the deviation dataset. Statistical analysis methods are used to calculate the mean and variance of the deviation values. Then, consumer behavior characteristics and sales response characteristics are extracted from market feedback data. Principal component analysis is used to reduce the dimensionality of the features, resulting in a simplified market feedback feature set. When the feature dimension of the simplified market feedback feature set meets preset conditions, the deviation value distribution model and the market feedback feature set are input into a reinforcement learning algorithm. The Q-learning method is used to iteratively update the sales task allocation strategy, resulting in a preliminary task allocation adjustment plan. Based on the preliminary task allocation adjustment plan... The proposed solution employs simulation to evaluate the performance of the adjusted plan in a virtual market environment, calculating the task allocation efficiency index to obtain the evaluated allocation efficiency value. If the evaluated allocation efficiency value does not reach the preset efficiency threshold, inefficient task allocation points are extracted from the simulation results, and the task allocation parameters are optimized using the gradient descent method. Then, the adjusted deviation value is obtained from the sales execution deviation dataset, and a statistical test method is used to determine whether the deviation value is lower than the preset threshold, resulting in the final task allocation plan. If the deviation value of the final task allocation plan is lower than the preset threshold, the optimized task allocation plan is saved through the data storage system, and the plan is distributed to the sales execution system using an automated scheduling tool to complete the task allocation adjustment.

6. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: The process of obtaining dynamically updated sales execution status includes: continuously collecting sales progress and market demand change data through a real-time data stream processing framework to obtain raw data streams; when the integrity of the raw data streams meets a preset threshold, using streaming processing technology to analyze sales progress and market demand changes to obtain structured data; based on the structured data, using time series analysis algorithms to detect the changing trends of sales progress and market demand, determining the changing characteristics; when the changing characteristics exceed a preset fluctuation range, adjusting the parameters of the real-time monitoring mechanism according to preset rules to obtain an updated monitoring model; using the updated monitoring model to analyze the sales execution status in the structured data to obtain the dynamic execution status; then using a decision tree algorithm to optimize the sales task allocation strategy and determine the adjusted task plan; and based on the adjusted task plan, updating the sales execution process to obtain the optimized sales execution status.

7. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: The final inventory optimization and adjustment plan is obtained through the following steps: First, a decision tree model is trained based on dynamically updated sales execution status. Input features include sales execution status, inventory data, and market demand data. The output is a judgment result indicating whether to adjust the inventory allocation strategy, resulting in a trained decision tree model. Second, the trained decision tree model is used to predict the real-time acquired comprehensive dataset. If the prediction indicates an adjustment to the inventory allocation strategy, inventory adjustment rules are generated, resulting in a preliminary adjustment plan. Third, combined with predicted demand trends, a linear regression model is used to analyze the changing trends of market demand data, obtaining the demand change trend. Fourth, the preliminary adjustment plan is adjusted based on the demand change trend. If the trend shows an increase in demand, the inventory allocation ratio of related products is increased, resulting in an optimized inventory allocation plan. Finally, specific inventory adjustment instructions are generated and executed through the inventory management system to obtain the final inventory optimization result.

8. The method for coordinated management of coal type sales planning and sales time in coal enterprises according to claim 1, characterized in that: Before obtaining the fused feature dataset, the process includes: acquiring multi-dimensional data, collecting structured and unstructured data from coal type inventory, quality indicators, market demand, and historical sales records, storing the data in a distributed database, and processing the data with different formats through data cleaning and standardization to obtain a multi-dimensional dataset in a unified format.

9. A coal type sales planning and sales time collaborative management system for coal enterprises, used to implement the coal type sales planning and sales time collaborative management method for coal enterprises as described in any one of claims 1-8, characterized in that: include: The data acquisition and preprocessing module collects structured and unstructured data from multi-dimensional data sources, including coal type inventory, quality indicators, market demand, and historical sales records of coal enterprises. Through data cleaning and standardization, it unifies the data of different formats and stores it in a distributed database. The data fusion and feature extraction module processes the collected multi-source heterogeneous data, extracts key features, and generates a fused feature dataset. The sales trend and price fluctuation prediction module combines a long short-term memory neural network and a regression analysis model. Inputting time series data and market demand data, it trains a sales trend prediction model to obtain predicted trends for coal sales volume and price. Then, using historical price data and market demand data, it analyzes price fluctuation patterns to obtain the predicted range for price fluctuations. The inventory allocation and sales task decomposition module, based on the sales trend prediction and price fluctuation prediction range, uses a linear programming algorithm. Inputting coal type quality indicators and inventory data from a multi-dimensional dataset, it optimizes the inventory allocation scheme, generates a daily sales task decomposition plan, and dynamically adjusts the sales task allocation based on real-time sales progress data and market feedback data. The sales execution deviation analysis and adjustment module monitors sales execution progress in real time. Through time series analysis, it determines whether sales execution deviations exceed preset thresholds, generating a sales execution deviation dataset. When deviations exceed the threshold, a reinforcement learning algorithm is used, inputting the deviation dataset and market feedback data, to dynamically adjust sales task allocation and optimize the sales task adjustment plan. The real-time monitoring and inventory optimization module employs a real-time data stream processing framework to continuously collect sales progress and market demand change data, updating the real-time monitoring mechanism to obtain dynamically updated sales execution status. Combined with inventory data and market demand data, a decision tree model is used to determine whether to adjust the inventory allocation strategy, generating the final inventory optimization adjustment plan.

Images