Coal mill hydraulic system control method based on machine learning

Abstract

The invention provides a coal mill hydraulic system control method based on machine learning, and belongs to the technical field of monitoring control, and the method comprises the steps: obtaining the historical operation data of a coal mill hydraulic system, and collecting the real-time operation data of the coal mill hydraulic system in real time; determining a starting condition matrix, an optimal shutdown matrix and a real-time operation matrix of the coal mill hydraulic system; the working state of the coal mill hydraulic system is determined, and start-stop control over the coal mill hydraulic system is achieved; constructing a control model based on the historical operation data, and generating a real-time control signal; and based on the real-time fault type and the real-time control signal of the coal mill hydraulic system, fault control over the coal mill hydraulic system is achieved. The self-adaption of various working conditions is realized, the fault detection accuracy is improved, the fault rate is reduced, the intelligent start-stop and fault control of the system are realized, the real-time performance and the accuracy of control are improved, and the stability, the intelligent level and the operation efficiency of the hydraulic system of the coal mill are improved.

Description

Technical Field

[0001] The present invention relates to the field of monitoring and control technology, and in particular to a coal mill hydraulic system control method based on machine learning. Background Art

[0002] In traditional coal mill hydraulic systems, control usually relies on fixed logic rules. However, this method has limitations when dealing with complex working conditions and nonlinear faults, and it is difficult to accurately cope with the rapid changes in system status and the complexity of different fault types. Especially when the system is in dynamic operation, the control method based on fixed rules is difficult to optimize the startup and shutdown process in real time, and it is also difficult to accurately identify and promptly handle different types of faults. In addition, the traditional method is not intelligent enough.

[0003] Therefore, the present invention provides a coal mill hydraulic system control method based on machine learning. Summary of the invention

[0004] The present invention provides a coal mill hydraulic system control method based on machine learning. The method determines the coal mill hydraulic system startup condition matrix and the optimal shutdown matrix by analyzing the historical operation data of the coal mill hydraulic system, determines the coal mill hydraulic system real-time operation matrix by analyzing the real-time operation data, and determines the working state, real-time fault type and control signal of the coal mill hydraulic system. The coal mill hydraulic system realizes adaptation to various working conditions, improves the fault detection accuracy and reduces the fault rate, realizes the intelligent start and stop and fault control of the system, improves the real-time and accuracy of the control, and improves the stability, intelligence level and operation efficiency of the coal mill hydraulic system.

[0005] The present invention provides a coal mill hydraulic system control method based on machine learning, comprising:

[0006] 101: Acquire historical operation data of the hydraulic system of the coal mill, and collect real-time operation data of the hydraulic system of the coal mill in real time;

[0007] 102: Analyze historical operation data to determine a start condition matrix and an optimal shutdown matrix of a coal mill hydraulic system, and analyze real-time operation data to determine a real-time operation matrix of the coal mill hydraulic system;

[0008] 103: Determine the working state of the coal mill hydraulic system based on the real-time operation data, the start condition matrix and the optimal stop matrix, and realize the start and stop control of the coal mill hydraulic system;

[0009] 104: construct a control model based on historical operation data, determine the real-time fault type of the coal mill hydraulic system based on the control model and the real-time operation matrix, and generate a real-time control signal;

[0010] 105: Based on the real-time fault type and real-time control signal of the coal mill hydraulic system, the fault control of the coal mill hydraulic system is realized.

[0011] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, historical operation data is analyzed to determine a start-up condition matrix and an optimal shutdown matrix of the coal mill hydraulic system, including:

[0012] Extracting data of the coal mill hydraulic system in the first specified time period before each start time from the historical operation data, and determining a plurality of historical start data after preprocessing the extracted data, wherein each historical start data includes sub-start data of all equipment in the coal mill hydraulic system before the corresponding start time;

[0013] Feature extraction is performed on the sub-startup data of each device in the coal mill hydraulic system before each start-up time, and the start-up vector of each device before each start-up time is determined;

[0014] Determine a fitting start vector for each device based on all start vectors of each device, and determine a start condition matrix for a coal mill hydraulic system based on the fitting start vectors of all devices;

[0015] Extracting data of the coal mill hydraulic system in the second specified time period before the stop time from the historical operation data to determine the historical stop data, wherein the historical stop data includes sub-stop data of all equipment in the coal mill hydraulic system;

[0016] Feature extraction is performed on the sub-stop data of each device in the coal mill hydraulic system before each stop time, and the stop vector of each device before each stop time is determined;

[0017] A fitting stop vector of each device is determined based on all the stop vectors of each device, and an optimal stop matrix of the coal mill hydraulic system is determined based on the fitting stop vectors of all the devices.

[0018] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, real-time operation data is analyzed to determine a real-time operation matrix of the coal mill hydraulic system, including:

[0019] Based on the first sensor group, real-time sub-operation data of each device in the coal mill hydraulic system is collected in real time, and real-time operation data is determined based on the real-time sub-operation data of all devices in the coal mill hydraulic system;

[0020] Extract features from the pre-processed real-time sub-operation data of each device in the coal mill hydraulic system to determine the real-time operation vector of each device;

[0021] The real-time operation matrix of the coal mill hydraulic system is determined based on the real-time operation vectors of all equipment in the coal mill hydraulic system.

[0022] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, the working state of the coal mill hydraulic system is determined based on real-time operation data, a start condition matrix and an optimal stop matrix, and the start and stop control of the coal mill hydraulic system is realized, including:

[0023] Determine the similarity between the real-time operation matrix and the start-up condition matrix, and at the same time, determine the similarity between the real-time operation matrix and the optimal shutdown matrix;

[0024] The working state includes: waiting to start state, waiting to stop state and running state;

[0025] If the similarity between the real-time operation matrix and the start-up condition matrix is ​​greater than or equal to the start-up threshold, the coal mill hydraulic system is determined to be in a waiting-to-start state, a start-up signal is determined based on the real-time operation matrix, and start-up control of the coal mill hydraulic system is implemented based on the start-up signal;

[0026] If the similarity between the real-time operation matrix and the optimal stop matrix is ​​greater than or equal to the stop threshold, the coal mill hydraulic system is determined to be in a waiting-to-stop state, a stop signal is determined based on the real-time operation matrix, and stop control of the coal mill hydraulic system is implemented based on the stop signal;

[0027] Otherwise, determine that the coal mill hydraulic system is in working condition.

[0028] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, a control model is constructed based on historical operation data, including:

[0029] Extract sub-fault data and fault labels of each fault of the coal mill hydraulic system in the historical operation data;

[0030] Determine the fault label data of each fault label based on the pre-processed sub-fault data of all faults with the same fault label;

[0031] Determine a label target value and a label threshold for each fault label based on the fault label data of each fault label;

[0032] The fault label data of each fault label are divided into a training set and a test set respectively, a control model is constructed based on the label target values ​​and label thresholds of all fault labels, and the control model is trained based on the training set in the fault label data of all fault labels;

[0033] Input the test set of the fault label data of all fault labels into the control model, and determine the predicted label of each sub-fault data based on the output result of the control model;

[0034] The control model is evaluated based on the prediction labels of all sub-fault data and the fault labels of all sub-fault data, and the control model is optimized based on the evaluation results.

[0035] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, based on a control model and a real-time operation matrix, a real-time fault type of the coal mill hydraulic system is determined and a real-time control signal is generated, including:

[0036] Inputting the real-time operation matrix of the coal mill hydraulic system into the control model, if it is determined based on the output result of the control model that the coal mill hydraulic system has a fault, then determining the real-time fault type of the coal mill hydraulic system;

[0037] Generate real-time control signals based on the real-time operation matrix, real-time fault types, all tag target values, and all tag thresholds.

[0038] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, a real-time control signal is generated based on a real-time operation matrix, a real-time fault type, all label target values ​​and all label thresholds, including:

[0039]

[0040] Where CS represents the real-time control signal, PL represents pressure fault, PT represents flow fault, PF represents temperature fault, PV represents vibration fault, Rt represents the real-time fault type, 1≤j≤N2, N2 represents the number of eigenvalues ​​in the real-time operation vector, k1 represents the first gain parameter, LT p represents the target value of the label based on pressure, X ij represents the jth eigenvalue of the real-time operation vector of the i-th device, Y p represents the characteristic value based on pressure, TN represents the nominal temperature of the hydraulic system of the coal mill, and Y t represents the characteristic value based on temperature, β represents the first adjustment coefficient, k2 represents the second gain parameter, LT q represents the target value of the label based on the flow, Y q represents the characteristic value based on flow, k3 represents the third gain parameter, TM represents the label threshold based on temperature, k4 represents the fourth gain parameter, τ and λ represent the second adjustment coefficient and the third adjustment coefficient respectively, α represents the fourth adjustment coefficient, γ represents the exponential increase parameter, k6 represents the fifth gain parameter, k6 represents the sixth gain parameter, k7 represents the seventh gain parameter, VM represents the label threshold based on vibration, Y v represents the vibration-based characteristic value, θ represents the fifth adjustment coefficient, LT o represents the target value of the label based on label o, TV orepresents the label threshold based on label o, N3 represents the number of fault labels other than pressure, temperature, flow, and vibration, and Y o represents the feature value based on label o, W o represents the weight of the control signal based on label o, A1 ij 、A2 ij 、A3 ij 、A4 ij 、A5 ij The first label value, the second label value, the third label value, the fourth label value, and the fifth label value respectively represent the j-th characteristic value of the real-time operation vector of the i-th device.

[0041] According to a coal mill hydraulic system control method based on machine learning provided by the present invention, based on the real-time fault type and real-time control signal of the coal mill hydraulic system, the fault control of the coal mill hydraulic system is realized, including:

[0042] The coal mill hydraulic system is fault controlled based on the real-time control signal, and a control report is generated based on the real-time fault type, the real-time control signal and the control result.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] By analyzing the historical operation data of the coal mill hydraulic system, the starting condition matrix and the optimal shutdown matrix of the coal mill hydraulic system are determined. By analyzing the real-time operation data, the real-time operation matrix of the coal mill hydraulic system is determined, and the working status, real-time fault type and control signal of the coal mill hydraulic system are determined. The coal mill hydraulic system realizes adaptation to various working conditions, improves the accuracy of fault detection and reduces the failure rate, realizes the intelligent start and stop and fault control of the system, improves the real-time and accuracy of the control, and improves the stability, intelligence level and operation efficiency of the coal mill hydraulic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 It is a flow chart of a coal mill hydraulic system control method based on machine learning provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings 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 creative work are within the scope of protection of the present invention.

[0048] Embodiment 1:

[0049] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, such as Figure 1 As shown, including:

[0050] 101: Acquire historical operation data of the hydraulic system of the coal mill, and collect real-time operation data of the hydraulic system of the coal mill in real time;

[0051] 102: Analyze historical operation data to determine a start condition matrix and an optimal shutdown matrix of a coal mill hydraulic system, and analyze real-time operation data to determine a real-time operation matrix of the coal mill hydraulic system;

[0052] 103: Determine the working state of the coal mill hydraulic system based on the real-time operation data, the start condition matrix and the optimal stop matrix, and realize the start and stop control of the coal mill hydraulic system;

[0053] 104: construct a control model based on historical operation data, determine the real-time fault type of the coal mill hydraulic system based on the control model and the real-time operation matrix, and generate a real-time control signal;

[0054] 105: Based on the real-time fault type and real-time control signal of the coal mill hydraulic system, the fault control of the coal mill hydraulic system is realized.

[0055] In this embodiment, historical operating data of the coal mill hydraulic system are collected. These data include the past status and parameter changes of the system, providing a data basis for subsequent analysis. At the same time, the system collects the current operating data of the coal mill in real time to grasp the current actual status of the system.

[0056] In this embodiment, after the historical operation data are collected, the data are analyzed to obtain the system startup condition matrix and the optimal shutdown matrix. The startup condition matrix contains the conditions required for the stable startup of the system, and the optimal shutdown matrix contains the optimal state conditions when the system is shut down. At the same time, by analyzing the operation data collected in real time, the current real-time operation matrix can be generated to understand the real-time state of the system.

[0057] In this embodiment, by combining real-time operating data, the start condition matrix and the optimal shutdown matrix, the system can determine whether the current state is suitable for starting or stopping. Based on the judgment result, the system automatically controls the start or shutdown of the coal mill hydraulic system to realize intelligent start and stop control.

[0058] In this embodiment, based on the identified real-time fault type and the generated real-time control signal, the system will immediately execute corresponding control measures to control the development of the fault and minimize the impact of the fault on the system. Through real-time feedback adjustment, the system can continuously monitor the fault status and dynamically adjust the control measures, thereby realizing a closed loop of fault control.

[0059] The beneficial effects of the above technical scheme are as follows: by analyzing the historical operation data of the coal mill hydraulic system, the starting condition matrix and the optimal shutdown matrix of the coal mill hydraulic system are determined, and the real-time operation data are analyzed to determine the real-time operation matrix of the coal mill hydraulic system, and the working state, real-time fault type and control signal of the coal mill hydraulic system are determined. The coal mill hydraulic system realizes adaptation to various working conditions, improves the accuracy of fault detection and reduces the fault rate, realizes the intelligent start and stop and fault control of the system, improves the real-time and accuracy of the control, and improves the stability, intelligence level and operation efficiency of the coal mill hydraulic system.

[0060] Embodiment 2:

[0061] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which analyzes historical operation data to determine a start-up condition matrix and an optimal shutdown matrix of the coal mill hydraulic system, including:

[0062] Extracting data of the coal mill hydraulic system in the first specified time period before each start time from the historical operation data, and determining a plurality of historical start data after preprocessing the extracted data, wherein each historical start data includes sub-start data of all equipment in the coal mill hydraulic system before the corresponding start time;

[0063] Feature extraction is performed on the sub-startup data of each device in the coal mill hydraulic system before each start-up time, and the start-up vector of each device before each start-up time is determined;

[0064] Determine a fitting start vector for each device based on all start vectors of each device, and determine a start condition matrix for a coal mill hydraulic system based on the fitting start vectors of all devices;

[0065] Extracting data of the coal mill hydraulic system in the second specified time period before the stop time from the historical operation data to determine the historical stop data, wherein the historical stop data includes sub-stop data of all equipment in the coal mill hydraulic system;

[0066] Feature extraction is performed on the sub-stop data of each device in the coal mill hydraulic system before each stop time, and the stop vector of each device before each stop time is determined;

[0067] A fitting stop vector of each device is determined based on all the stop vectors of each device, and an optimal stop matrix of the coal mill hydraulic system is determined based on the fitting stop vectors of all the devices.

[0068] In this embodiment, the historical startup data represents data within the first specified time period before each startup of the coal mill hydraulic system, including the operating status of the equipment before startup.

[0069] In this embodiment, the sub-startup data includes the operation data of all devices before a certain startup.

[0070] In this embodiment, the startup vector refers to a feature vector extracted based on the sub-startup data, reflecting feature information of a device before startup.

[0071] In this embodiment, the fitted startup vector refers to a representative vector obtained by fitting startup vectors of each device in multiple startups.

[0072] In this embodiment, the start-up condition matrix is ​​composed of the fitted start-up vectors of all devices and is used to determine the start-up conditions of the system.

[0073] In this embodiment, the historical stop data represents the operation data within the second specified time period before each stop of the coal mill hydraulic system.

[0074] In this embodiment, the sub-stop data represents the operation data of all devices before a certain stop.

[0075] In this embodiment, the stop vector represents a feature vector extracted based on the sub-stop data, reflecting the state of the device before stopping.

[0076] In this embodiment, the fitted stop vector is a representative vector obtained by fitting the stop vectors of each device in multiple stops.

[0077] In this embodiment, the optimal stop matrix is ​​composed of the fitted stop vectors of all devices and is used to determine the optimal stop condition of the system.

[0078] The beneficial effects of the above technical solution are as follows: by analyzing historical operating data to determine the start-up condition matrix and the optimal shutdown matrix of the coal mill hydraulic system, the start-up and shutdown conditions of the system can be accurately determined, thus improving the data basis for realizing high-precision automatic control.

[0079] Embodiment 3:

[0080] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which analyzes real-time operation data to determine a real-time operation matrix of the coal mill hydraulic system, including:

[0081] Based on the first sensor group, real-time sub-operation data of each device in the coal mill hydraulic system is collected in real time, and real-time operation data is determined based on the real-time sub-operation data of all devices in the coal mill hydraulic system;

[0082] Extract features from the pre-processed real-time sub-operation data of each device in the coal mill hydraulic system to determine the real-time operation vector of each device;

[0083] The real-time operation matrix of the coal mill hydraulic system is determined based on the real-time operation vectors of all equipment in the coal mill hydraulic system.

[0084] In this embodiment, the real-time sub-operation data is collected in real time by the sensor group, and reflects the operation data of the current state of each device.

[0085] In this embodiment, the real-time operation data is a summary of the real-time sub-operation data of all devices, indicating the current overall operation status of the system.

[0086] In this embodiment, the real-time operation vector is a vector extracted based on the real-time sub-operation data of each device, which simplifies the state expression of the device.

[0087] In this embodiment, the real-time operation matrix is ​​composed of the real-time operation vectors of each device, representing the overall real-time status of the system.

[0088] The beneficial effects of the above technical solution are: analyzing real-time operating data to determine the real-time operating matrix of the coal mill hydraulic system, improving the real-time and accuracy of status monitoring and the response speed of intelligent decision-making, and facilitating system optimization and fault prevention.

[0089] Embodiment 4:

[0090] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which determines the working state of the coal mill hydraulic system based on real-time operation data, a start condition matrix and an optimal stop matrix, and realizes the start and stop control of the coal mill hydraulic system, including:

[0091] Determine the similarity between the real-time operation matrix and the start-up condition matrix, and at the same time, determine the similarity between the real-time operation matrix and the optimal shutdown matrix;

[0092] The working state includes: waiting to start state, waiting to stop state and running state;

[0093] If the similarity between the real-time operation matrix and the start-up condition matrix is ​​greater than or equal to the start-up threshold, the coal mill hydraulic system is determined to be in a waiting-to-start state, a start-up signal is determined based on the real-time operation matrix, and start-up control of the coal mill hydraulic system is implemented based on the start-up signal;

[0094] If the similarity between the real-time operation matrix and the optimal stop matrix is ​​greater than or equal to the stop threshold, the coal mill hydraulic system is determined to be in a waiting-to-stop state, a stop signal is determined based on the real-time operation matrix, and stop control of the coal mill hydraulic system is implemented based on the stop signal;

[0095] Otherwise, determine that the coal mill hydraulic system is in working condition.

[0096] In this embodiment, the similarity measures the matching degree between the two matrices, indicating the closeness of the real-time state to the expected state (startup or shutdown condition).

[0097] In this embodiment, the startup threshold represents a similarity threshold of the startup state. When this value is reached or exceeded, the system enters the waiting-to-start state.

[0098] In this embodiment, the stop threshold represents a similarity threshold of the stop state, and when this value is reached or exceeded, the system enters the waiting-to-stop state.

[0099] In this embodiment, the start signal refers to a signal generated after the system meets the start condition, which is used to trigger the system to start.

[0100] In this embodiment, the stop signal refers to a signal generated after the system meets the shutdown condition, which is used to trigger the system to stop.

[0101] In this embodiment, the start signal includes real-time adjustment of the loading oil pressure to the level required for starting, rapid response of the loading oil pressure, and ensuring that a normal working state is achieved in the shortest time.

[0102] In this embodiment, the stop signal includes gradually reducing the load and reducing the loading oil pressure at the same time, so that the real-time adjustment of the loading oil pressure is no longer performed, so that the system can reduce the load smoothly.

[0103] The beneficial effects of the above technical solution are as follows: based on real-time operating data, the starting condition matrix and the optimal shutdown matrix, the working state of the coal mill hydraulic system is determined, and the start and stop control of the coal mill hydraulic system is realized. The current state of the coal mill hydraulic system can be intelligently judged, the system automation and response capabilities are improved, human intervention is reduced, and safety and energy efficiency are improved.

[0104] Embodiment 5:

[0105] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which constructs a control model based on historical operation data, including:

[0106] Extract sub-fault data and fault labels of each fault of the coal mill hydraulic system in the historical operation data;

[0107] Determine the fault label data of each fault label based on the pre-processed sub-fault data of all faults with the same fault label;

[0108] Determine a label target value and a label threshold for each fault label based on the fault label data of each fault label;

[0109] The fault label data of each fault label are divided into a training set and a test set respectively, a control model is constructed based on the label target values ​​and label thresholds of all fault labels, and the control model is trained based on the training set in the fault label data of all fault labels;

[0110] Input the test set of the fault label data of all fault labels into the control model, and determine the predicted label of each sub-fault data based on the output result of the control model;

[0111] The control model is evaluated based on the prediction labels of all sub-fault data and the fault labels of all sub-fault data, and the control model is optimized based on the evaluation results.

[0112] In this embodiment, the sub-fault data represents the operating data of all sub-devices in the system when a certain fault occurs, such as abnormal data such as temperature, pressure, flow, etc.

[0113] In this embodiment, the fault label is a label for identifying each fault type and is used to distinguish different fault types.

[0114] In this embodiment, the fault label data represents a collection of all sub-fault data with the same fault label.

[0115] In this embodiment, the label target value represents a value representing a typical feature of a certain type of fault label, and is used to determine the occurrence of a fault.

[0116] In this embodiment, the tag threshold is a threshold standard used to determine whether a fault corresponding to the tag occurs.

[0117] In this embodiment, the control model is a machine learning model built based on target values ​​and thresholds of various fault labels, which can identify and predict different fault types.

[0118] In this embodiment, the predicted label represents the fault type label predicted by the model based on the input data, which is used to determine whether the system is currently in a certain fault state.

[0119] The beneficial effects of the above technical solution are as follows: by building a control model based on historical operating data, various types of faults in the hydraulic system of the coal mill can be intelligently identified and predicted, efficient self-diagnosis can be achieved, fault detection accuracy can be improved, and the failure rate can be reduced, thus helping to improve the stability and safety of system operation.

[0120] Embodiment 6:

[0121] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which determines the real-time fault type of the coal mill hydraulic system based on a control model and a real-time operation matrix and generates a real-time control signal, including:

[0122] Inputting the real-time operation matrix of the coal mill hydraulic system into the control model, if it is determined based on the output result of the control model that the coal mill hydraulic system has a fault, then determining the real-time fault type of the coal mill hydraulic system;

[0123] Generate real-time control signals based on the real-time operation matrix, real-time fault types, all tag target values, and all tag thresholds.

[0124] In this embodiment, the control model analyzes the real-time operation matrix. If the model output indicates that the system has a fault, the model will further give a specific fault type. This fault type is based on different fault labels learned in previous training, thereby determining the actual fault type of the system.

[0125] In this embodiment, after determining the fault type, a "real-time control signal" is generated by combining the data of the real-time operation matrix, the real-time fault type, and the target values ​​and thresholds of all labels. This signal can automatically adjust the system operation status to respond to the currently detected fault type. The control signal is adjusted according to the severity of the fault and the label threshold to ensure timely and accurate response to the fault and prevent further damage to the system.

[0126] The beneficial effects of the above technical solution are: based on the control model and the real-time operation matrix, the real-time fault type of the coal mill hydraulic system is determined and a real-time control signal is generated, which effectively improves the system's adaptability and automation level, reduces the need for manual intervention, and enhances the safety and response efficiency of the system operation.

[0127] Embodiment 7:

[0128] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which generates a real-time control signal based on a real-time operation matrix, a real-time fault type, all label target values, and all label thresholds, including:

[0129]

[0130] Where CS represents the real-time control signal, PL represents pressure fault, PT represents flow fault, PF represents temperature fault, PV represents vibration fault, Rt represents the real-time fault type, 1≤j≤N2, N2 represents the number of eigenvalues ​​in the real-time operation vector, k1 represents the first gain parameter, LT p represents the target value of the label based on pressure, X ij represents the jth eigenvalue of the real-time operation vector of the i-th device, Y p represents the characteristic value based on pressure, TN represents the nominal temperature of the hydraulic system of the coal mill, and Y t represents the characteristic value based on temperature, β represents the first adjustment coefficient, k2 represents the second gain parameter, LT q represents the target value of the label based on the flow, Y q represents the characteristic value based on flow, k3 represents the third gain parameter, TM represents the label threshold based on temperature, k4 represents the fourth gain parameter, τ and λ represent the second adjustment coefficient and the third adjustment coefficient respectively, α represents the fourth adjustment coefficient, γ represents the exponential increase parameter, k6 represents the fifth gain parameter, k6 represents the sixth gain parameter, k7 represents the seventh gain parameter, VM represents the label threshold based on vibration, Y v represents the vibration-based characteristic value, θ represents the fifth adjustment coefficient, LT o represents the target value of the label based on label o, TV o represents the label threshold based on label o, N3 represents the number of fault labels other than pressure, temperature, flow, and vibration, and Y o represents the feature value based on label o, W o represents the weight of the control signal based on label o, A1 ij 、A2 ij 、A3 ij 、A4 ij 、A5 ij The first label value, the second label value, the third label value, the fourth label value, and the fifth label value respectively represent the j-th characteristic value of the real-time operation vector of the i-th device.

[0131] In this embodiment, Indicates a pressure-based control signal.

[0132] In this embodiment, Represents a temperature-based control signal.

[0133] In this embodiment, Indicates a flow-based control signal.

[0134] In this embodiment, Represents a vibration-based control signal.

[0135] In this embodiment, the first label value A1 ij Indicates whether the jth eigenvalue of the real-time operation vector of the i-th device is based on the eigenvalue of label o. If so, A1 ij The value is 1. If not, A1 ij The value is 0.

[0136] In this embodiment, the second label value A2 ij Indicates whether the jth eigenvalue of the real-time operation vector of the i-th device is a pressure-based eigenvalue. If so, A2 ij The value is 1. If not, A2 ij The value is 0.

[0137] In this embodiment, the third tag value A3 ij Indicates whether the jth eigenvalue of the real-time operation vector of the i-th device is a eigenvalue based on the flow rate. If so, A3 ij The value is 1. If not, A3 ij The value is 0.

[0138] In this embodiment, the fourth tag value A4 ij Indicates whether the jth eigenvalue of the real-time operation vector of the i-th device is a temperature-based eigenvalue. If so, A4 ij The value is 1, if not, A4 ij The value is 0.

[0139] In this embodiment, the fifth tag value A5 ij Indicates whether the jth eigenvalue of the real-time operation vector of the i-th device is based on the vibration eigenvalue. If so, A5 ij The value is 1. If not, A5 ij The value is 0.

[0140] The beneficial effects of the above technical solution are: based on the real-time operation matrix, the real-time fault type, all label target values ​​and all label thresholds, a real-time control signal is generated, which can automatically generate a control signal for a rapid diagnosis result of a coal mill hydraulic system fault.

[0141] Embodiment 8:

[0142] The embodiment of the present invention provides a coal mill hydraulic system control method based on machine learning, which implements fault control of the coal mill hydraulic system based on the real-time fault type and real-time control signal of the coal mill hydraulic system, including:

[0143] The coal mill hydraulic system is fault controlled based on the real-time control signal, and a control report is generated based on the real-time fault type, the real-time control signal and the control result.

[0144] In this embodiment, the real-time control signal represents an automated control instruction generated according to the fault type detected in real time, and is used to adjust the operating state of the system to cope with the fault.

[0145] In this embodiment, fault control is a process in which the system receives and executes real-time control signals to make adjustments, aiming to control or mitigate the impact of the fault and prevent further damage to the system.

[0146] In this embodiment, the control result indicates the operating effect or state of the system after the control signal is executed, reflecting whether the fault is alleviated or eliminated.

[0147] In this embodiment, the control report is a file that records the fault type, control measures and control effects in detail, and is used to analyze, evaluate and optimize the fault handling capability of the system.

[0148] The beneficial effects of the above technical solution are as follows: the historical operating data of the coal mill hydraulic system is analyzed to determine the starting condition matrix and the optimal shutdown matrix of the coal mill hydraulic system, the real-time operating data is analyzed to determine the real-time operating matrix of the coal mill hydraulic system, and the working status, real-time fault type and control signal of the coal mill hydraulic system are determined. The coal mill hydraulic system realizes adaptation to various working conditions, improves the accuracy of fault detection and reduces the fault rate, realizes the intelligent start and stop and fault control of the system, improves the real-time and accuracy of the control, and improves the stability, intelligence level and operation efficiency of the coal mill hydraulic system.

[0149] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0150] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0151] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A coal mill hydraulic system control method based on machine learning, characterized in that: include: 101: Acquire historical operation data of the hydraulic system of the coal mill, and collect real-time operation data of the hydraulic system of the coal mill in real time; 102: Analyze historical operation data to determine a start condition matrix and an optimal shutdown matrix of a coal mill hydraulic system, and analyze real-time operation data to determine a real-time operation matrix of the coal mill hydraulic system; 103: Determine the working state of the coal mill hydraulic system based on the real-time operation data, the start condition matrix and the optimal stop matrix, and realize the start and stop control of the coal mill hydraulic system; 104: construct a control model based on historical operation data, determine the real-time fault type of the coal mill hydraulic system based on the control model and the real-time operation matrix, and generate a real-time control signal; 105: Based on the real-time fault type and real-time control signal of the coal mill hydraulic system, the fault control of the coal mill hydraulic system is realized.

2. The method for controlling a coal mill hydraulic system based on machine learning according to claim 1, characterized in that: Analyze historical operation data to determine the startup condition matrix and optimal shutdown matrix of the coal mill hydraulic system, including: Extracting data of the coal mill hydraulic system in the first specified time period before each start time from the historical operation data, and determining a plurality of historical start data after preprocessing the extracted data, wherein each historical start data includes sub-start data of all equipment in the coal mill hydraulic system before the corresponding start time; Feature extraction is performed on the sub-startup data of each device in the coal mill hydraulic system before each start-up time, and the start-up vector of each device before each start-up time is determined; Determine a fitting start vector for each device based on all start vectors of each device, and determine a start condition matrix for a coal mill hydraulic system based on the fitting start vectors of all devices; Extracting data of the coal mill hydraulic system in the second specified time period before the stop time from the historical operation data to determine the historical stop data, wherein the historical stop data includes sub-stop data of all equipment in the coal mill hydraulic system; Feature extraction is performed on the sub-stop data of each device in the coal mill hydraulic system before each stop time, and the stop vector of each device before each stop time is determined; A fitting stop vector of each device is determined based on all the stop vectors of each device, and an optimal stop matrix of the coal mill hydraulic system is determined based on the fitting stop vectors of all the devices.

3. The method for controlling a coal mill hydraulic system based on machine learning according to claim 1, characterized in that: Analyze the real-time operation data to determine the real-time operation matrix of the coal mill hydraulic system, including: Based on the first sensor group, real-time sub-operation data of each device in the coal mill hydraulic system is collected in real time, and real-time operation data is determined based on the real-time sub-operation data of all devices in the coal mill hydraulic system; Extract features from the pre-processed real-time sub-operation data of each device in the coal mill hydraulic system to determine the real-time operation vector of each device; The real-time operation matrix of the coal mill hydraulic system is determined based on the real-time operation vectors of all equipment in the coal mill hydraulic system.

4. The method for controlling a coal mill hydraulic system based on machine learning according to claim 1, characterized in that: Based on real-time operation data, the start condition matrix and the optimal shutdown matrix, the working state of the coal mill hydraulic system is determined, and the start and stop control of the coal mill hydraulic system is realized, including: Determine the similarity between the real-time operation matrix and the start-up condition matrix, and at the same time, determine the similarity between the real-time operation matrix and the optimal shutdown matrix; The working state includes: waiting to start state, waiting to stop state and running state; If the similarity between the real-time operation matrix and the start-up condition matrix is ​​greater than or equal to the start-up threshold, the coal mill hydraulic system is determined to be in a waiting-to-start state, a start-up signal is determined based on the real-time operation matrix, and start-up control of the coal mill hydraulic system is implemented based on the start-up signal; If the similarity between the real-time operation matrix and the optimal stop matrix is ​​greater than or equal to the stop threshold, the coal mill hydraulic system is determined to be in a waiting-to-stop state, a stop signal is determined based on the real-time operation matrix, and stop control of the coal mill hydraulic system is implemented based on the stop signal; Otherwise, determine that the coal mill hydraulic system is in working condition.

5. The method for controlling a coal mill hydraulic system based on machine learning according to claim 1, characterized in that: Build control models based on historical operating data, including: Extract sub-fault data and fault labels of each fault of the coal mill hydraulic system in the historical operation data; Determine the fault label data of each fault label based on the pre-processed sub-fault data of all faults with the same fault label; Determine a label target value and a label threshold for each fault label based on the fault label data of each fault label; The fault label data of each fault label are divided into a training set and a test set respectively, a control model is constructed based on the label target values ​​and label thresholds of all fault labels, and the control model is trained based on the training set in the fault label data of all fault labels; Input the test set of the fault label data of all fault labels into the control model, and determine the predicted label of each sub-fault data based on the output result of the control model; The control model is evaluated based on the prediction labels of all sub-fault data and the fault labels of all sub-fault data, and the control model is optimized based on the evaluation results.

6. The method for controlling a coal mill hydraulic system based on machine learning according to claim 5, characterized in that: Based on the control model and the real-time operation matrix, the real-time fault type of the coal mill hydraulic system is determined and a real-time control signal is generated, including: Inputting the real-time operation matrix of the coal mill hydraulic system into the control model, if it is determined based on the output result of the control model that the coal mill hydraulic system has a fault, then determining the real-time fault type of the coal mill hydraulic system; Generate real-time control signals based on the real-time operation matrix, real-time fault types, all tag target values, and all tag thresholds.

7. The method for controlling a coal mill hydraulic system based on machine learning according to claim 6, characterized in that: Generate real-time control signals based on the real-time operation matrix, real-time fault types, all tag target values, and all tag thresholds, including: Where CS represents the real-time control signal, PL represents pressure fault, PT represents flow fault, PF represents temperature fault, PV represents vibration fault, Rt represents the real-time fault type, 1≤j≤N2, N2 represents the number of eigenvalues ​​in the real-time operation vector, k1 represents the first gain parameter, LT p represents the target value of the label based on pressure, X ij represents the jth eigenvalue of the real-time operation vector of the i-th device, Y p represents the characteristic value based on pressure, TN represents the nominal temperature of the hydraulic system of the coal mill, and Y t represents the characteristic value based on temperature, β represents the first adjustment coefficient, k2 represents the second gain parameter, LT q represents the target value of the label based on the flow, Y q represents the characteristic value based on flow, k3 represents the third gain parameter, TM represents the label threshold based on temperature, k4 represents the fourth gain parameter, τ and λ represent the second adjustment coefficient and the third adjustment coefficient respectively, α represents the fourth adjustment coefficient, γ represents the exponential increase parameter, k6 represents the fifth gain parameter, k6 represents the sixth gain parameter, k7 represents the seventh gain parameter, VM represents the label threshold based on vibration, Y v represents the vibration-based characteristic value, θ represents the fifth adjustment coefficient, LT o represents the target value of the label based on label o, TV o represents the label threshold based on label o, N3 represents the number of fault labels other than pressure, temperature, flow, and vibration, and Y o represents the feature value based on label o, W o represents the weight of the control signal based on label o, A1 ij 、A2 ij 、A3 ij 、A4 ij 、A5 ij The first label value, the second label value, the third label value, the fourth label value, and the fifth label value respectively represent the j-th characteristic value of the real-time operation vector of the i-th device.

8. The method for controlling a coal mill hydraulic system based on machine learning according to claim 1, characterized in that: Based on the real-time fault type and real-time control signal of the coal mill hydraulic system, the fault control of the coal mill hydraulic system is realized, including: The coal mill hydraulic system is fault controlled based on the real-time control signal, and a control report is generated based on the real-time fault type, the real-time control signal and the control result.

Images