Coal mining machine adaptive energy-saving optimization system based on reinforcement learning

Abstract

The invention discloses a coal mining machine adaptive energy-saving optimization system based on reinforcement learning, and relates to the field of energy supply optimization, and the system comprises a central control unit which is used for carrying out the butt joint with a program engineering end of a coal mining machine to be optimized, and obtaining the submission and control authority of a data instruction; the production identification module is used for reading a production yield target and a production speed index of the current production cycle, and converting various indexes into production tasks of a plurality of time periods; by collecting energy consumption data, hardware states, production tasks and environmental conditions of all the operation parts in real time and utilizing a deep learning algorithm to construct an optimization prediction model, dynamic changes of the production environment are continuously learned and adapted, the optimal energy consumption level of all the operation parts in different production stages is accurately recognized, and the production efficiency is improved. And dynamically generating standard energy supply templates of each running part in different time periods, and timely finding and tracking the part causing the abnormality by setting the energy supply fluctuation tolerance and identifying the potential abnormal energy consumption trend.

Description

Technical Field

[0001] The present invention relates to the technical field of energy supply optimization, and in particular to a coal mining machine adaptive energy-saving optimization system based on reinforcement learning. Background Art

[0002] As an important equipment for coal mining, coal mining machines are mainly used in underground mining. Modern coal mines usually adopt continuous coal mining technology. Coal mining machines can continuously feed on the working face, improve coal mining efficiency and reduce labor costs. The use of advanced electronic control technology can achieve precise control of equipment and improve the stability and safety of operations. With the development of remote control technology, coal mining machines can realize automated operation to a certain extent, reduce dependence on manpower, and improve safety. With the promotion of energy conservation and emission reduction policies, the design and application of coal mining machines need to emphasize energy efficiency management and use advanced drive and control technologies to reduce energy consumption;

[0003] During the operation of coal mining machines, production demand, hardware status and environmental conditions often change dynamically, but traditional optimization methods often fail to adjust energy-saving strategies in a timely manner. Energy-saving strategies are often rigid and lack the ability to automatically adjust according to real-time and historical data. Many existing energy consumption optimization models use static methods and fail to respond to changes in production environment and equipment status in a timely and effective manner. They are not conducive to tracking the real-time energy consumption status and production load of each operating component, and it is difficult to detect and respond to energy consumption anomalies in a timely manner.

[0004] There is a lack of reasonable scheduling of energy supply strategies for different operating components. The response mechanism for abnormal fluctuations in energy consumption is usually slow, making it difficult to quickly identify and locate problematic components. The existing system is usually unable to adjust the energy supply in a timely manner according to the actual needs and status of each component, which leads to excessive energy consumption in some components and insufficient energy consumption in other components. It is difficult to effectively link production progress feedback and energy efficiency analysis, and it is impossible to update the optimization strategy in time according to the actual production situation. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the above-mentioned shortcomings of the prior art, the present invention provides a coal mining machine adaptive energy-saving optimization system based on reinforcement learning, which can effectively solve the problems of the prior art.

[0007] (II) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] The invention discloses a coal mining machine adaptive energy-saving optimization system based on reinforcement learning, comprising:

[0010] The central control unit is used to obtain the submission and control authority of data instructions from the program engineering end of the reception optimization coal mining machine;

[0011] The production identification module is used to read the production output target and production speed index of the current production cycle and convert various indicators into production tasks for several time periods;

[0012] The attribute collection module is used to obtain the data reading authority of several operating components related to energy consumption, read the setting parameters and energy consumption characteristic parameters of each operating component in the current production cycle, integrate them into hardware data, and analyze the standard energy consumption data set to store in the database;

[0013] The energy consumption analysis module is used to collect historical data on the energy consumption status of each operating component under different production tasks, hardware data, environmental data and energy supply capacity, and output historical energy consumption data sets;

[0014] A model building unit is used to build an optimization prediction model based on the historical energy consumption data set, analyze the potential abnormal energy consumption trend of each operating component in the current production cycle, define the tolerance of energy supply fluctuations of each operating component, and output the standard energy supply template of each operating component in different time periods of the current cycle;

[0015] The detection and analysis module is used to monitor the energy consumption of each component in real time during the current production cycle, compare it with the standard energy supply template, calculate the fluctuation coefficient, and mark it as an abnormal event if the fluctuation exceeds the set range;

[0016] The abnormality tracing module is used to receive the abnormal events marked by the abnormality tracing module, perform data backtracking of abnormal fluctuations, and identify several operating components associated with abnormal fluctuations based on potential abnormal energy consumption trends;

[0017] The priority identification module is used to analyze the correlation coefficients of the operating components involved in each production task in the current production cycle, define the components with priority energy supply according to the correlation coefficients and energy consumption performance of each component, and formulate an energy supply priority list;

[0018] The feedback adjustment module is used to collect and analyze the actual execution status of the production schedule, compare it with the prediction results of the model building unit, and generate an adjustment plan for the current optimized prediction model based on the feedback information output by the abnormality tracing module and the priority identification module, and submit it to the central control unit for execution.

[0019] Furthermore, the model building unit includes submodules: a training module, a trend output module, a fluctuation formulation module and a template generation module, and the training module, the trend output module, the fluctuation formulation module and the template generation module are interactively connected via a wireless network;

[0020] The training module is used to build an optimized prediction model framework through a deep learning algorithm, obtain historical energy consumption data sets for preprocessing, use the preprocessed energy consumption data sets as training sets for model training, and identify the energy consumption law parameters of each operating component under different production tasks, hardware data, environmental data, and energy supply capabilities;

[0021] The trend output module is used to receive the current production tasks, hardware data, environmental data and energy supply capacity through the trained model, and predict and output the potential abnormal energy consumption trend of each component;

[0022] The fluctuation formulation module sets the energy consumption tolerance range of each operating component in the current cycle based on the output of potential abnormal energy consumption trends;

[0023] The template generation module is used to obtain the standard energy consumption data set of each operating component, analyze the energy consumption tolerance range of each operating component in the current cycle, and generate a standard energy supply template for each operating component in different time periods of the current cycle.

[0024] Furthermore, the training module constructs an optimized prediction model framework through a deep learning algorithm, including:

[0025] Input layer: historical energy consumption data set; production task information: including production output, production speed and task type; hardware data: equipment working control status, operating time and wear degree; environmental data: climate conditions and ore layer properties; energy supply capacity indicators: maximum load of the power system and historical energy supply fluctuations;

[0026] Hidden layer: A multi-layer fully connected neural network is used. Each hidden layer uses a certain number of neurons. The activation function is ReLU. The LSTM structure is used to capture the energy consumption change trend in the time series.

[0027] Output layer: Output the predicted energy consumption of each running component in a specific time period in the future, build several output nodes, and predict the energy consumption of different components respectively.

[0028] Furthermore, the training module obtains the historical energy consumption data set for preprocessing by: cleaning the original data, selecting features related to energy consumption in the current production cycle, performing unique hot encoding on the classification features, converting them into numerical data, normalizing the numerical features, and creating time series features through a sliding window.

[0029] Furthermore, the calculation formula of the energy consumption law parameter of each operating component in the training module is:

[0030]

[0031] In the formula, E represents the total energy consumption of a certain operating component, wo represents the intercept of the model, X i represents the production task characteristics, Y j Represents hardware data, Z k Represents environmental data, H l represents the energy supply capability indicator, n, m, p and q represent the quantity of corresponding production tasks, hardware data, environmental data and energy supply capability characteristics respectively, and w i 、w j 、w k and w l They represent the weight coefficients associated with production tasks, hardware data, environmental data and energy supply capacity respectively, and ∈ represents the error term.

[0032] Furthermore, the calculation logic of the calculation formula of the energy consumption law parameters is: by determining the model parameters and features, initializing the weights, calculating the linear combination of each feature, forming the contribution value of each feature to the total energy consumption, and integrating the random fluctuation error to generate the total energy consumption of a certain operating component.

[0033] Furthermore, the template generation module is interactively connected to an evaluation module via a wireless network, and the evaluation module is used to cross-validate the standard energy supply template output by the template generation module, and synchronously receive adjustment plan feedback from the feedback adjustment module, readjust the template generation module, and verify the completion of the adjustment plan in the next cycle, and add the adjustment plan data with a completion degree higher than a preset threshold as a training sample to the training module to participate in the retraining of the optimization prediction model.

[0034] Furthermore, the definition indicators of the priority energy supply component list of the priority identification module include: average energy consumption, energy consumption standard deviation, correlation coefficient score and comprehensive priority score. The specific definition process is: collect energy consumption data of each component, calculate the average energy consumption, energy consumption standard deviation and correlation coefficient of all components, filter operating components according to the average energy consumption, energy consumption standard deviation and correlation coefficient performance, calculate the comprehensive priority score, sort the scores of each operating component, and obtain a priority energy supply component list.

[0035] Furthermore, after being triggered, the priority identification module calls up the associated operating components of the current production task and a preset number of future production tasks, and simultaneously obtains the hardware data of each associated operating component and the real-time energy consumption performance data collected by the detection and analysis module through the attribute collection module, and integrates and analyzes the output of the operating components with priority power supply, and supports manual and automatic adjustment of the operating components with priority functions.

[0036] Furthermore, the central control unit is interactively connected to the production identification module, the attribute collection module and the feedback adjustment module through an electrical medium, the energy consumption analysis module is interactively connected to the production identification module, the attribute collection module and the model building unit through a wireless network, the detection and analysis module is interactively connected to the attribute collection module, the model building unit, the abnormality tracing module and the priority identification module through a wireless network, and the feedback adjustment module is interactively connected to the abnormality tracing module and the priority identification module through a wireless network.

[0037] (III) Beneficial effects

[0038] Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects:

[0039] 1. By collecting energy consumption data, hardware status, production tasks and environmental conditions of each operating component in real time, using deep learning algorithms to build an optimized prediction model, continuously learning and adapting to the dynamic changes of the production environment, continuously optimizing feature selection and model parameters during the deep learning process, analyzing historical and real-time data, and accurately identifying the optimal energy consumption level of each operating component in different production stages, accurate energy consumption prediction can be achieved.

[0040] 2. By continuously monitoring the energy consumption status of running components and comparing them with standard energy supply templates, an alarm will be immediately issued and traceability analysis will be performed when an abnormality occurs. The real-time detection and feedback mechanism allows the system to quickly adapt to changes in the production environment and ensure dynamic management of energy consumption. By analyzing the relevance of production tasks, environmental factors and the status of running components, and combining historical data, standard energy supply templates for each running component in different time periods are dynamically generated to optimize energy consumption configuration. According to changes in the real-time production environment and demand, the energy supply strategy is automatically adjusted to improve production efficiency.

[0041] 3. By setting the tolerance for energy supply fluctuations and identifying potential abnormal energy consumption trends, the system can promptly detect and track components that cause abnormalities, implement targeted energy efficiency management for each component, accurately detect abnormal energy consumption events, and avoid unnecessary energy waste. By optimizing the energy consumption of operating components, it significantly reduces energy costs in the coal mining process, identifies components that require maintenance or replacement in advance, reduces downtime and improves overall equipment utilization, and when troubleshooting cannot be performed in a timely manner, the fault factor can be used as a reference for energy consumption analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0043] Figure 1 It is a schematic diagram of the architecture of the present invention;

[0044] Figure 2 It is a schematic diagram of the framework of the optimization prediction model in the present invention.

[0045] The numbers in the figure represent: 1. Central control unit; 2. Production identification module; 3. Attribute collection module; 4. Energy consumption analysis module;

[0046] 5. Model building unit; submodules include: 51. training module; 52. trend output module; 53. fluctuation formulation module; 54. template generation module;

[0047] 6. Detection and analysis module; 7. Abnormal tracing module; 8. Priority identification module; 9. Feedback adjustment module; 10. Evaluation module. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0049] The present invention will be further described below in conjunction with the embodiments.

[0050] Example 1

[0051] The coal mining machine adaptive energy-saving optimization system based on reinforcement learning in this embodiment is as follows: Figure 1 As shown, including:

[0052] The central control unit 1 is used to obtain the submission and control authority of data instructions from the program engineering end of the reception optimization coal mining machine;

[0053] Production identification module 2 is used to read the production output target and production speed index of the current production cycle, and convert various indicators into production tasks for several time periods; by converting production targets into actionable production tasks, it ensures the clarity and executability of the production plan and provides basic data for subsequent optimization;

[0054] The attribute collection module 3 is used to obtain the data reading authority of several operating components related to energy consumption, read the setting parameters and energy consumption characteristic parameters of each operating component in the current production cycle, integrate them into hardware data, and analyze the standard energy consumption data set to store in the database; comprehensively understand the energy consumption characteristics of each operating component, provide an accurate data basis for subsequent energy consumption analysis, and set reasonable standards for standard energy consumption;

[0055] Energy consumption analysis module 4 is used to collect historical data of energy consumption status of each operating component under different production tasks, hardware data, environmental data and energy supply capacity, and output historical energy consumption data set; through the accumulation of historical data, it can realize trend analysis of energy consumption, provide historical support for the establishment of optimization model, and make the prediction more accurate and reliable;

[0056] The model building unit 5 is used to build an optimization prediction model based on the historical energy consumption data set, analyze the potential abnormal energy consumption trend of each operating component in the current production cycle, define the tolerance of energy supply fluctuation of each operating component, and output the standard energy supply template of each operating component in different periods of the current cycle; through the analysis of energy consumption data, the potential abnormal energy consumption trend can be identified in advance, and equipment failure and energy waste can be effectively prevented to achieve energy saving;

[0057] The detection and analysis module 6 is used to monitor the energy consumption status of each component in real time during the current production cycle, compare it with the standard energy supply template, calculate the fluctuation coefficient, and mark it as an abnormal event if the fluctuation exceeds the set range; the real-time monitoring of the energy consumption status enables the abnormal energy consumption to be quickly identified and processed, thereby achieving dynamic regulation of production efficiency;

[0058] The abnormality tracing module 7 is used to receive the abnormal events marked by the abnormality tracing module 7, perform data backtracking of abnormal fluctuations, and identify several operating components associated with abnormal fluctuations based on potential abnormal energy consumption trends; through the backtracking analysis of abnormal fluctuations, key components that affect energy consumption can be identified, thereby improving the pertinence of problem solving;

[0059] The priority identification module 8 is used to analyze the correlation coefficients of the operating components involved in each production task in the current production cycle, define the components with priority energy supply according to the correlation coefficients and energy consumption performance of each component, and formulate an energy supply priority list; the definition indicators of the priority energy supply component list of the priority identification module 8 include: average energy consumption, energy consumption standard deviation, correlation coefficient score and comprehensive priority score, and the specific definition process is: collect energy consumption data of each component, calculate the average energy consumption, energy consumption standard deviation and correlation coefficient of all components, screen the operating components according to the average energy consumption, energy consumption standard deviation and correlation coefficient performance, calculate the comprehensive priority score, sort the scores of each operating component, and obtain a priority energy supply component list; after being triggered, the priority identification module 8 retrieves the associated operating components of the current production task and the future preset number of production tasks, and simultaneously obtains the hardware data of the associated operating components and the real-time energy consumption performance data collected by the detection and analysis module 6 through the attribute collection module 3, integrates and analyzes the output of the operating components with priority energy supply, and supports manual automatic adjustment of the operating components with priority functions; optimizes the energy supply strategy, prioritizes the energy consumption requirements of important components, improves the overall energy efficiency performance of the system, and helps to achieve reasonable allocation of resources;

[0060] The feedback adjustment module 9 is used to collect and analyze the actual execution status of the production schedule, compare it with the prediction results of the model building unit 5, and generate an adjustment plan for the current optimization prediction model based on the feedback information output by the abnormality tracing module 7 and the priority identification module 8, and submit it to the central control unit 1 for execution. By adjusting the optimization model through real-time feedback, a closed-loop optimization system is formed, which can better match the prediction with the actual execution and further improve production efficiency.

[0061] As an implementation method in this embodiment, Figure 1 As shown, the central control unit 1 is interactively connected with the production identification module 2, the attribute collection module 3 and the feedback adjustment module 9 through an electrical medium, the energy consumption analysis module 4 is interactively connected with the production identification module 2, the attribute collection module 3 and the model building unit 5 through a wireless network, the detection and analysis module 6 is interactively connected with the attribute collection module 3, the model building unit 5, the abnormality tracing module 7 and the priority identification module 8 through a wireless network, and the feedback adjustment module 9 is interactively connected with the abnormality tracing module 7 and the priority identification module 8 through a wireless network.

[0062] Compared with the existing technology, this embodiment is based on real-time monitoring and historical data analysis, can quickly adapt to changes in the production process, provide accurate energy consumption management, and introduces a reinforcement learning algorithm, so that the optimization model can continuously self-learn and improve to cope with complex production environments. Through real-time monitoring and adjustment strategies, it significantly reduces energy waste and improves the operating efficiency of equipment, achieves higher energy-saving effects, realizes efficient circulation and use of data, forms a complete feedback control system, and improves the operability of the entire system.

[0063] Example 2

[0064] At other levels, this embodiment also provides a construction and operation process of an optimization prediction model, wherein the model construction unit 5 includes submodules: a training module 51, a trend output module 52, a fluctuation formulation module 53, and a template generation module 54, wherein the training module 51, the trend output module 52, the fluctuation formulation module 53, and the template generation module 54 are interactively connected via a wireless network;

[0065] The training module 51 is used to construct an optimized prediction model framework through a deep learning algorithm, obtain a historical energy consumption data set for preprocessing, use the preprocessed energy consumption data set as a training set for model training, and identify the energy consumption law parameters of each operating component under different production tasks, hardware data, environmental data and energy supply capacity; the process of obtaining the historical energy consumption data set for preprocessing is: data cleaning of the original data, selecting features related to energy consumption in the current production cycle, performing unique hot encoding on the classification features, converting them into numerical data, normalizing the numerical features, and creating time series features through a sliding window; the model framework constructed by the deep learning algorithm can effectively capture the complex patterns in the historical energy consumption data, and the preprocessing process ensures the data quality, providing a high-quality training set for subsequent model training, thereby improving the accuracy and reliability of the prediction;

[0066] The trend output module 52 is used to receive the current production tasks, hardware data, environmental data and energy supply capacity through the trained model, and predict and output the potential abnormal energy consumption trend of each component;

[0067] The fluctuation formulation module 53 sets the energy consumption tolerance range of each operating component in the current cycle based on the output of the potential abnormal energy consumption trend;

[0068] The template generation module 54 is used to obtain a standard energy consumption data set for each operating component, analyze the energy consumption tolerance range of each operating component in the current cycle, and generate a standard energy supply template for each operating component in different time periods of the current cycle. The template generation module 54 is interactively connected to the evaluation module 10 via a wireless network. The evaluation module 10 is used to cross-validate the standard energy supply template output by the template generation module 54, and synchronously receive the adjustment plan feedback from the feedback adjustment module 9, readjust the template generation module 54, and verify the completion of the adjustment plan in the next cycle. The adjustment plan data with a completion degree higher than the preset threshold is added as a training sample to the training module 51 to participate in the retraining of the optimization prediction model, and the standard energy supply template generated by the comprehensive analysis of the energy consumption tolerance provides a reference energy supply standard for each operating component, ensuring that the energy consumption effect of each component is good in different production periods, thereby optimizing the energy supply strategy of the entire system.

[0069] As a preferred implementation in this embodiment, Figure 2 As shown, the training module 51 constructs an optimized prediction model framework through a deep learning algorithm, including:

[0070] Input layer: historical energy consumption data set; production task information: including production output, production speed and task type; hardware data: equipment working control status, operating time and wear degree; environmental data: climate conditions and ore layer properties; energy supply capacity indicators: maximum load of the power system and historical energy supply fluctuations;

[0071] Hidden layer: A multi-layer fully connected neural network is used. Each hidden layer uses a certain number of neurons. The activation function is ReLU. The LSTM structure is used to capture the energy consumption change trend in the time series.

[0072] Output layer: Output the predicted energy consumption of each running component in a specific time period in the future, build several output nodes, and predict the energy consumption of different components respectively.

[0073] In the specific implementation of this embodiment, the historical energy consumption data is deeply mined through the deep learning algorithm. Compared with the traditional method, it can identify more complex and subtle energy consumption patterns, improve the data processing and model prediction capabilities, receive the latest production and environmental data in real time, and self-adjust the prediction model and energy supply strategy, making energy consumption management more flexible and dynamic.

[0074] Through the combination of multi-layer neural networks and time series analysis, the processing capabilities of time-sensitive data have been effectively improved, avoiding decision-making errors caused by delays or static analysis. Continuous learning and optimization based on the feedback mechanism ensure that the model always has a high accuracy rate and can quickly respond to changes in the production environment to achieve true adaptive optimization.

[0075] Example 3

[0076] In this embodiment, a calculation process of the energy consumption law parameters of each operating component is provided, and the specific calculation formula is:

[0077]

[0078] In the formula, E represents the total energy consumption of a certain operating component, which means the energy consumption of the operating component under specific conditions, and w o represents the intercept of the model, indicating the baseline energy consumption when all eigenvalues ​​are zero, X i represents the production task characteristics, including information about the production task, production output, production speed and task type, where i ranges from 1 to n, representing different production task characteristics, Y j Represents hardware data, involving the operating status, control parameters and wear degree of the equipment. j ranges from 1 to m, indicating different hardware characteristics. Zk Represents environmental data, such as mineral layer characteristics hardness and humidity and climate conditions temperature and pressure. k ranges from 1 to p, indicating different environmental characteristics. H l represents the energy supply capacity index, reflecting the power supply capacity and fluctuation current and voltage changes of the power system. l ranges from 1 to q, indicating different energy supply capacity characteristics. n, m, p and q represent the number of corresponding production tasks, hardware data, environmental data and energy supply capacity characteristics, respectively. w i 、w j 、w k and w l Respectively represent the weight coefficients associated with production tasks, hardware data, environmental data, and energy supply capacity, indicating the degree of influence of each factor on the total energy consumption. These weights are obtained through model training. ∈ represents the error term, which represents random noise and usually obeys a normal distribution, indicating unpredictable energy consumption fluctuations;

[0079] The calculation logic of the calculation formula for the energy consumption law parameters is as follows: by determining the model parameters and features, initializing the weights, calculating the linear combination of each feature, forming the contribution value of each feature to the total energy consumption, and integrating the random fluctuation error to generate the total energy consumption of a certain operating component.

[0080] Compared with the existing technology, the calculation process of energy consumption law parameters provides strong support for energy consumption monitoring and management through systematic analysis, data-driven optimization, reasonable and explainable model construction, and appropriate handling of random fluctuations, making this method have broad application potential and value in practical applications such as energy management, production optimization and fault warning.

[0081] Working principle: The present invention connects the central control system of the coal mining machine through the central control unit 1, edits and sends the operating indicators of each functional unit and functional module, obtains the production demand of the coal mining machine in the current cycle through the production identification module 2, obtains the attributes of the energy-consuming running components involved in production through the attribute collection module 3, and analyzes the energy consumption status of each energy-consuming running component under different production states and environmental conditions through the energy consumption analysis module 4;

[0082] The training module 51 constructs an optimization prediction model through a deep neural network learning algorithm, and uses historical data as sample training. The trend output module 52 outputs the potential abnormal energy consumption trend of the operating components based on the hardware status, production demand, environmental conditions and energy supply capacity. The fluctuation formulation module 53 formulates the energy supply fluctuation tolerance of each operating component based on the trend. The template generation module 54 generates and formulates the standard energy supply template of each operating component in different time periods of the current cycle. The evaluation module 10 evaluates the availability of the model.

[0083] The energy consumption status of each operating component in the production process is detected in real time through the feedback adjustment module 9, and the fluctuation coefficient is analyzed through the standard energy supply template. When abnormal fluctuations occur, the abnormal tracing module 7 is used to trace the source and obtain the abnormal components associated with the abnormal fluctuations. Based on the potential abnormal energy consumption trend of the operating components output by the current cycle model, the specific abnormal components that cause the abnormal fluctuations are judged. The priority identification module 8 analyzes the correlation of various production tasks of the current production demand and outputs the priority energy supply components. The feedback adjustment module 9 generates an adjustment plan with the energy supply priority components as the priority energy supply target, obtains production progress feedback, generates an adjustment plan and feeds it back to the training module 51 for training. The template generation module 54 outputs the standard energy supply template for the next time period of the current cycle.

[0084] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. The adaptive energy-saving optimization system for coal mining machines based on reinforcement learning is characterized by: include: The central control unit (1) is used to obtain the submission and control authority of data instructions from the program engineering end of the reception optimization coal mining machine; The production identification module (2) is used to read the production output target and production speed index of the current production cycle and convert various indicators into production tasks for a certain period of time; The attribute collection module (3) is used to obtain data reading permissions for several operating components related to energy consumption, read the setting parameters and energy consumption characteristic parameters of each operating component in the current production cycle, integrate them into hardware data, and analyze the standard energy consumption data set and store it in the database; An energy consumption analysis module (4) is used to collect historical data of energy consumption status of each operating component under different production tasks, hardware data, environmental data and energy supply capacity, and output a historical energy consumption data set; A model building unit (5) is used to build an optimization prediction model based on the historical energy consumption data set, analyze the potential abnormal energy consumption trend of each operating component in the current production cycle, define the energy supply fluctuation tolerance of each operating component, and output a standard energy supply template for each operating component in different time periods of the current cycle; The detection and analysis module (6) is used to monitor the energy consumption of each component in real time during the current production cycle, compare it with the standard energy supply template, calculate the fluctuation coefficient, and mark it as an abnormal event if the fluctuation exceeds the set range; An abnormality tracing module (7) is used to receive abnormal events marked by the abnormality tracing module (7), perform data backtracking of abnormal fluctuations, and identify several operating components associated with abnormal fluctuations based on potential abnormal energy consumption trends; A priority identification module (8) is used to analyze the correlation coefficients of the operating components involved in each production task in the current production cycle, define the components with priority energy supply according to the correlation coefficients and energy consumption performance of each component, and formulate an energy supply priority list; The feedback adjustment module (9) is used to collect and analyze the actual execution status of the production schedule, compare it with the prediction results of the model building unit (5), and generate an adjustment plan for the current optimization prediction model based on the feedback information output by the abnormality tracing module (7) and the priority identification module (8) and submit it to the central control unit (1) for execution.

2. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 1 is characterized in that: The model building unit (5) comprises submodules: a training module (51), a trend output module (52), a fluctuation formulation module (53) and a template generation module (54), wherein the training module (51), the trend output module (52), the fluctuation formulation module (53) and the template generation module (54) are interactively connected via a wireless network; A training module (51) is used to construct an optimization prediction model framework through a deep learning algorithm, obtain a historical energy consumption data set for preprocessing, use the preprocessed energy consumption data set as a training set for model training, and identify energy consumption law parameters of each operating component under different production tasks, hardware data, environmental data and energy supply capacity; A trend output module (52) is used to receive current production tasks, hardware data, environmental data and energy supply capacity through a trained model, and predict and output potential abnormal energy consumption trends of each component; A fluctuation setting module (53) sets the energy consumption tolerance range of each operating component in the current cycle based on the output of the potential abnormal energy consumption trend; The template generation module (54) is used to obtain a standard energy consumption data set of each operating component, analyze the energy consumption tolerance range of each operating component in the current cycle, and generate a standard energy supply template for each operating component in different time periods of the current cycle.

3. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 2 is characterized in that: The training module (51) constructs an optimized prediction model framework through a deep learning algorithm, including: Input layer: historical energy consumption data set; production task information: including production output, production speed and task type; hardware data: equipment working control status, operating time and wear degree; environmental data: climate conditions and ore layer properties; energy supply capacity indicators: maximum load of the power system and historical energy supply fluctuations; Hidden layer: A multi-layer fully connected neural network is used. Each hidden layer uses a certain number of neurons. The activation function is ReLU. The LSTM structure is used to capture the energy consumption change trend in the time series. Output layer: Output the predicted energy consumption of each running component in a specific time period in the future, build several output nodes, and predict the energy consumption of different components respectively.

4. The self-adaptive energy-saving optimization system for coal mining machines based on reinforcement learning according to claim 2 is characterized in that: The training module (51) obtains the historical energy consumption data set for preprocessing by: cleaning the original data, selecting features related to energy consumption in the current production cycle, performing unique-hot encoding on the classification features, converting them into numerical data, normalizing the numerical features, and creating time series features by means of a sliding window.

5. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 2 is characterized in that: The calculation formula for the energy consumption law parameter of each operating component in the training module (51) is: Where E represents the total energy consumption of a certain operating component, w o represents the intercept of the model, X i represents the production task characteristics, Y j Represents hardware data, Z k Represents environmental data, H l represents the energy supply capability indicator, n, m, p and q represent the quantity of corresponding production tasks, hardware data, environmental data and energy supply capability characteristics respectively, and w i 、w j 、w k and w l They represent the weight coefficients associated with production tasks, hardware data, environmental data and energy supply capacity respectively, and ∈ represents the error term.

6. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 5 is characterized in that: The calculation logic of the calculation formula of the energy consumption law parameter is: by determining the model parameters and features, initializing the weights, calculating the linear combination of each feature, forming the contribution value of each feature to the total energy consumption, and integrating the random fluctuation error to generate the total energy consumption of a certain operating component.

7. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 2 is characterized in that: The template generation module (54) is interactively connected to the evaluation module (10) via a wireless network. The evaluation module (10) is used to cross-validate the standard energy supply template output by the template generation module (54), and synchronously receive the adjustment plan feedback from the feedback adjustment module (9), readjust the template generation module (54), and verify the completion of the adjustment plan in the next cycle, and add the adjustment plan data with a completion degree higher than a preset threshold as a training sample to the training module (51) to participate in the retraining of the optimization prediction model.

8. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 1 is characterized in that: The definition indicators of the priority energy supply component list of the priority identification module (8) include: average energy consumption, energy consumption standard deviation, correlation coefficient score and comprehensive priority score. The specific definition process is: collecting energy consumption data of each component, calculating the average energy consumption, energy consumption standard deviation and correlation coefficient of all components, screening operating components according to the average energy consumption, energy consumption standard deviation and correlation coefficient performance, calculating the comprehensive priority score, sorting the scores of each operating component, and obtaining a priority energy supply component list.

9. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 1 is characterized in that: After being triggered, the priority identification module (8) retrieves the associated operating components of the current production task and the future preset number of production tasks, and simultaneously obtains the hardware data of the associated operating components and the real-time energy consumption performance data collected by the detection and analysis module (6) through the attribute collection module (3), performs integrated analysis and outputs the operating components with priority energy supply, and supports manual and automatic adjustment of the operating components with priority functions.

10. The coal mining machine adaptive energy-saving optimization system based on reinforcement learning according to claim 1, characterized in that: The central control unit (1) is interactively connected with the production identification module (2), the attribute collection module (3) and the feedback adjustment module (9) via an electrical medium; the energy consumption analysis module (4) is interactively connected with the production identification module (2), the attribute collection module (3) and the model construction unit (5) via a wireless network; the detection analysis module (6) is interactively connected with the attribute collection module (3), the model construction unit (5), the abnormality tracing module (7) and the priority identification module (8) via a wireless network; and the feedback adjustment module (9) is interactively connected with the abnormality tracing module (7) and the priority identification module (8) via a wireless network.

Images