Management method and system for coal dressing quality intelligent control

Abstract

The invention provides a management method and system for coal preparation quality intelligent control, and the method comprises the steps: firstly obtaining raw material quality, equipment operation and environment state index sets of a target production line, then carrying out the dynamic association processing of the index sets, and generating a technological process feature set; then calling a preset optimization decision model to carry out parameter prediction on the technological process feature set to obtain a process regulation parameter set, generating a quality optimization instruction set based on the process regulation parameter set, transmitting the quality optimization instruction set to a production control unit to execute dynamic regulation operation, and finally, according to a quality feedback index set after regulation execution, carrying out dynamic regulation operation. And the parameter weight distribution of the optimization decision model is updated, intelligent control and continuous optimization of the coal dressing quality are realized, and the stability of coal dressing production and the product quality are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular, to a management method and system for intelligent control of coal preparation quality. Background Art

[0002] In the field of coal washing and processing, traditional coal preparation quality control methods mainly rely on manual experience judgment and fixed parameter regulation modes. In the prior art, raw material quality monitoring usually adopts an offline sampling and analysis method, and there is a significant lag in data acquisition, making it difficult to reflect the dynamic fluctuation characteristics of the incoming coal quality in real time. The monitoring of equipment operating status is mostly limited to single-parameter threshold alarms, and the collaborative relationship between various equipment is not effectively utilized, resulting in a lack of systematic support for fault warning and performance optimization. Although some environmental state parameters (such as temperature, humidity, vibration, etc.) are collected, they are only used as independent monitoring items and are not organically associated with the process, and their potential value has not been fully explored.

[0003] The current process parameter regulation is mainly based on a preset static rule library, and parameter adjustment is performed through a look-up table method or simple logical judgment. This rigid control strategy cannot adapt to the complex changes in raw material properties, equipment status, and environmental conditions. More critically, the mapping relationship between regulation parameters and process effects in the prior art lacks a dynamic learning mechanism. When production conditions deviate, it is neither possible to autonomously correct the control strategy nor to achieve iterative upgrading of the optimization model through historical data accumulation. This open-loop control mode leads to large fluctuations in coal preparation quality, low resource utilization rate, and problems such as regulation failure or over-intervention when facing complex working conditions. Summary of the Invention

[0004] In view of the problems mentioned above, in combination with the first aspect of the present invention, embodiments of the present invention provide a management method for intelligent control of coal preparation quality, and the method includes: Obtain a set of raw material quality indicators, a set of equipment operation indicators, and a set of environmental state indicators for a target production line; Perform dynamic association processing on the set of raw material quality indicators, the set of equipment operation indicators, and the set of environmental state indicators to generate a set of process characteristic sets; Call a preset optimization decision model to perform parameter prediction on the set of process characteristic sets to generate a set of process regulation parameters; Generate a set of quality optimization instructions based on the set of process regulation parameters, and transmit the set of quality optimization instructions to the production control unit of the target production line to perform dynamic regulation operations; Update the parameter weight distribution of the optimization decision model according to the set of quality feedback indicators after regulation execution.

[0005] On the other hand, an embodiment of the present invention also provides a management system for intelligent control of coal preparation quality, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0006] Based on the above aspects, the embodiment of the present invention forms a representation space that can fully reflect the characteristics of the process by integrating the dynamic association processing mechanism of multi-source heterogeneous data. On this basis, the preset optimization decision model deeply mines the process feature set through machine learning algorithms, and realizes the dynamic prediction of key control parameters. Its output parameter set not only has real-time response capabilities, but also can form a two-way interaction with the production control unit through a closed-loop feedback mechanism, so that the generation and execution of control instructions have adaptive optimization characteristics, thereby transforming the coal preparation quality optimization process from experience-dependent to model-predictive, significantly improving the accuracy and timeliness of process parameter adjustment. More importantly, through the dynamic update mechanism of the model parameter weights by the quality feedback index, an intelligent control system with continuous evolution capabilities is constructed, so that the optimization model can automatically correct the decision logic as the production conditions change, thereby maintaining long-term stable optimization performance in a complex and changeable industrial environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a schematic diagram of the execution flow of the management method for intelligent control of coal preparation quality provided by an embodiment of the present invention.

[0008] Figure 2 It is a schematic diagram of exemplary hardware and software components of a management system for intelligent control of coal preparation quality provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0009] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 It is a flow chart of a management method for intelligent control of coal preparation quality provided by an embodiment of the present invention. The management method for intelligent control of coal preparation quality is introduced in detail below.

[0010] Step S110: Obtain a set of raw material quality indicators, a set of equipment operation indicators, and a set of environmental status indicators of a target production line.

[0011] For example, in a coal preparation plant, a coal preparation production line that processes a specific type of coal is selected as the target production line. While the production process is ongoing, various sets of indicators are systematically obtained.

[0012] Specifically, for the set of raw material quality indicators, relevant data are obtained with the help of professional testing equipment. Among them, the raw material ash fluctuation data are recorded every 15 minutes by a high-precision ash detector, such as A1, A2, A3... An, which reflect the fluctuation of the raw material ash content at different time points. The sulfur content distribution data are obtained through multi-point sampling and detection. For example, sampling points are selected at set intervals on the raw material conveyor belt, and after detecting each sampling point, B1, B2, B3... Bm are obtained, reflecting the sulfur content distribution at different positions of the raw material. The particle size change data are measured at different times using a particle size detection instrument, denoted as C1, C2, C3... Ck, showing the dynamic change of the raw material particle size over time.

[0013] The set of equipment operation indicators covers various aspects of data. The set value of the sorting medium density of the equipment is adjusted in real time according to the raw material characteristics and production process requirements. For example, it is set as D1 at the initial stage of production and then adjusted to D2, D3, etc. according to the actual situation. The equipment vibration frequency parameters are set as needed in different production stages, namely E1, E2, E3... The sorting time control parameters are set respectively for different batches of raw materials as F1, F2, F3... The actual sorting efficiency of the equipment is obtained by analyzing and counting products such as clean coal and gangue produced. In different production periods, such as the first period is G1, the second period is G2, and the third period is G3... The energy consumption rate is monitored and recorded in real time by an energy monitoring device, and each monitoring value is recorded as H1, H2, H3... The equipment vibration amplitude data are collected in real time by a vibration sensor, and each collection value is I1, I2, I3...

[0014] The set of environmental status indicators is obtained through corresponding monitoring equipment. The environmental temperature change data are measured every 30 minutes in a specific monitoring area of the coal preparation plant. The first measurement value is recorded as J1, the second is J2, and the xth is Jx. The humidity fluctuation data are measured at set intervals, and each measurement value is successively K1, K2, K3... Ky. The dust concentration data are obtained through dust concentration detection equipment at different detection points and detection times. For example, the first detection value at the first detection point is L1, the first detection value at the second detection point is L2, and the first detection value at the zth detection point is Lz.

[0015] Step S120: Perform dynamic association processing on the set of raw material quality indicators, the set of equipment operation indicators, and the set of environmental status indicators to generate a set of process characteristics.

[0016] After completely obtaining the above various types of indicator sets, the dynamic association processing work is immediately carried out to generate a set of process characteristics.

[0017] Step S121: Perform time-series fluctuation analysis on the raw material characteristic change data in the set of raw material quality indicators to generate raw material dynamic change characteristics.

[0018] Step S1211: Perform standardized transformation and time series alignment processing on the raw material ash content fluctuation data, sulfur content distribution data, and particle size change data in the raw material quality index set to generate a raw material multi-dimensional time series dataset.

[0019] In this step, for the raw material ash content fluctuation data A1, A2, ……, An, a standardization method is adopted. This standardization method may be based on statistical characteristics such as the mean and standard deviation of the data, and convert these raw material ash content fluctuation data into A1', A2', ……, An', so that the ash content data has a unified dimension and comparable scale. Similarly, perform standardization operations on the sulfur content distribution data B1, B2, ……, Bm to obtain B1', B2', ……, Bm'. Also perform corresponding standardization on the particle size change data C1, C2, ……, Ck to obtain C1', C2', ……, Ck'. After completing the standardization, arrange the standardized ash content, sulfur content, and particle size data in the order of detection time, ensuring that there are corresponding three types of data at each time point, thereby generating a raw material multi-dimensional time series dataset. For example, at the first time point, there are corresponding standardized ash content data A1', sulfur content data B1', and particle size data C1'; at the second time point, there are A2', B2', C2', and so on.

[0020] Step S1212: Calculate the moving window mean of the raw material multi-dimensional time series dataset to generate a smooth feature of raw material characteristics.

[0021] Assume that the moving window size is set to 4. For the ash content data part in the raw material multi-dimensional time series dataset, starting from A1', calculate (A1'+A2'+A3'+A4')÷4 to get M1, (A2'+A3'+A4'+A5')÷4 to get M2, and so on, to obtain a series of smooth feature values M1, M2, M3... for the ash content. Similarly, for the sulfur content data part, starting from B1', calculate (B1'+B2'+B3'+B4')÷4 to get N1, (B2'+B3'+B4'+B5')÷4 to get N2, to obtain smooth feature values N1, N2, N3... for the sulfur content. For the particle size data part, starting from C1', calculate (C1'+C2'+C3'+C4')÷4 to get O1, (C2'+C3'+C4'+C5')÷4 to get O2, to obtain smooth feature values O1, O2, O3... for the particle size. These smooth feature values for ash content, sulfur content, and particle size respectively together constitute the smooth feature of raw material characteristics.

[0022] Step S1213: Perform fluctuation amplitude detection based on the smooth feature of raw material characteristics to generate a raw material abnormal fluctuation mark.

[0023] Taking the smoothed eigenvalue of ash as an example, the fluctuation amplitude is determined by calculating the difference between adjacent smoothed eigenvalues. Calculate |M2 - M1| to obtain the fluctuation amplitude value P1, |M3 - M2| to obtain P2, and so on, to obtain a series of fluctuation amplitude values P1, P2, P3... for ash. The same method is used for sulfur content and smoothed eigenvalues. Calculate |N2 - N1| to obtain Q1, |N3 - N2| to obtain Q2, to obtain the fluctuation amplitude values Q1, Q2, Q3... for sulfur content; for the particle size smoothed eigenvalue, calculate |O2 - O1| to obtain R1, |O3 - O2| to obtain R2, to obtain the fluctuation amplitude values R1, R2, R3... for particle size. Compare these fluctuation amplitude values with their respective preset smaller thresholds (such as the threshold T1 for ash, the threshold T2 for sulfur content, and the threshold T3 for particle size). If the fluctuation amplitude value is greater than the corresponding threshold, it is marked as abnormal, and a raw material abnormal fluctuation mark is generated. For example, if P1 is greater than T1, the ash fluctuation at this time point is marked as abnormal, denoted as the abnormal mark S1; if Q1 is greater than T2, the sulfur content fluctuation at this time point is marked as abnormal, denoted as the abnormal mark T4; if R1 is greater than T3, the particle size fluctuation at this time point is marked as abnormal, denoted as the abnormal mark U1.

[0024] Step S1214: Compare the raw material abnormal fluctuation mark with a preset process tolerance threshold to generate a raw material quality stability evaluation parameter.

[0025] In this embodiment, the preset process tolerance thresholds for ash, sulfur content, and particle size are set as T1', T2', and T3' respectively. For the ash abnormal fluctuation mark, such as S1, if S1 is greater than T1', it indicates that the fluctuation of the raw material ash at this time point exceeds the process tolerance range, and an evaluation parameter V1 representing poor quality stability of the raw material ash is generated; if S1 is less than or equal to T1', an evaluation parameter V2 representing relatively stable quality of the raw material ash is generated. Similarly, for the sulfur content abnormal fluctuation mark T4, compared with T2', if T4 is greater than T2', an evaluation parameter W1 representing poor quality stability of the raw material sulfur content is generated; if T4 is less than or equal to T2', an evaluation parameter W2 representing relatively stable quality of the raw material sulfur content is generated. For the particle size abnormal fluctuation mark U1, compared with T3', if U1 is greater than T3', an evaluation parameter X1 representing poor quality stability of the raw material particle size is generated; if U1 is less than or equal to T3', an evaluation parameter X2 representing relatively stable quality of the raw material particle size is generated. These evaluation parameters comprehensively reflect the quality stability of the raw material.

[0026] Step S1215: Perform feature weighting processing on the raw material multi-dimensional time series dataset according to the raw material quality stability evaluation parameter to generate a raw material dynamic change feature including a fluctuation period feature, an abnormal offset feature, and a trend prediction feature.

[0027] Taking the ash data as an example, if the evaluation parameter V1 is obtained, that is, the mass stability of the raw material ash is poor, higher weights are assigned to the features related to the abnormal fluctuations of the ash in the multi-dimensional time series dataset of the raw material, and lower weights are assigned to the features of normal fluctuations. For example, for the standardized ash data A1', A2',..., An', the data near the time points marked with abnormal fluctuations have their weights increased in subsequent calculations. Through this weighted processing method, combined with the analysis of the data, the fluctuation period features are extracted, such as determining the fluctuation period by observing the interval time between the occurrences of consecutive abnormal fluctuation marks; the abnormal deviation features, that is, the degree to which the data deviates from the normal range during abnormal fluctuations; and the trend prediction features, by analyzing the weighted data to predict the future change trend of the raw material ash. The same method is used for the sulfur content and particle size data, and finally the dynamic change features of the raw material including the fluctuation period features, abnormal deviation features, and trend prediction features for ash, sulfur content, and particle size are generated. For example, the fluctuation period feature data generated for ash are Y1, Y2, Y3,..., the abnormal deviation feature data are Z1, Z2, Z3,..., and the trend prediction feature data are AA1, AA2, AA3,...; the corresponding feature data generated for sulfur content are BB1, BB2, BB3,..., CC1, CC2, CC3,..., DD1, DD2, DD3,...; the corresponding feature data generated for particle size are EE1, EE2, EE3,..., FF1, FF2, FF3,..., GG1, GG2, GG3,..., and these feature data together constitute the dynamic change features of the raw material.

[0028] Step S122: Perform a coupling and matching analysis on the equipment control parameters and operation efficiency data in the equipment operation index set to generate the dynamic features of the equipment efficiency.

[0029] Step S1221: Extract the setpoint of the equipment sorting medium density, the equipment vibration frequency parameter, and the sorting time control parameter in the equipment operation index set as the equipment control parameter set.

[0030] From the equipment operation index set, the setpoint of the equipment sorting medium density, such as D1, D2,..., Dn, the equipment vibration frequency parameter E1, E2,..., Em, and the sorting time control parameter F1, F2,..., Fk are extracted to form the equipment control parameter set. This equipment control parameter set contains the key adjustable parameters during the operation of the equipment, which have a direct impact on the operation efficiency of the equipment.

[0031] Step S1222: Collect the actual sorting efficiency, energy consumption rate, and equipment vibration amplitude data of the equipment in the equipment operation index set as the operation efficiency dataset.

[0032] Meanwhile, collect the actual sorting efficiency data of the acquisition device, such as G1, G2, ……, Gn, the energy consumption rate data H1, H2, ……, Hm, and the device vibration amplitude data I1, I2, ……, Ik, to form an operation efficiency dataset. These data reflect the working effect, energy consumption, vibration and other states of the device during actual operation, and are important bases for evaluating the operation efficiency of the device.

[0033] Step S1223: Align the timestamps of the device control parameter set and the operation efficiency dataset to generate a time series record of the device operation state.

[0034] Associate each parameter value in the device control parameter set with the corresponding efficiency data at the same time point in the operation efficiency dataset. For example, when the sorting medium density of the device is set to D1, find the actual sorting efficiency G1, energy consumption rate H1, and device vibration amplitude I1 of the device at the same moment, and so on. Arrange all the data in chronological order to generate a time series record of the device operation state. In this way, the corresponding relationship between the device control parameters and the operation efficiency can be clearly seen at each time point.

[0035] Step S1224: Perform standardized transformation and control parameter-efficiency response correlation analysis on the time series record of the device operation state in sequence to generate device response lag characteristics and parameter sensitivity characteristics.

[0036] For example, step S12241: Perform z-score standardized transformation on the device control parameter set and the operation efficiency dataset in the time series record of the device operation state respectively to generate a dimensionless control parameter sequence and a dimensionless efficiency index sequence.

[0037] For the sorting medium density setting values D1, D2, ……, Dn, device vibration frequency parameters E1, E2, ……, Em, and sorting time control parameters F1, F2, ……, Fk in the device control parameter set, use the z-score standardization method. This method is based on the mean and standard deviation of the data, and converts each parameter value into a dimensionless value to obtain a dimensionless control parameter sequence D1', D2', ……, Dn', E1', E2', ……, Em', F1', F2', ……, Fk'. Similarly, perform z-score standardized transformation on the actual sorting efficiency G1, G2, ……, Gn, energy consumption rate H1, H2, ……, Hm, and device vibration amplitude data I1, I2, ……, Ik in the operation efficiency dataset to generate a dimensionless efficiency index sequence G1', G2', ……, Gn', H1', H2', ……, Hm', I1', I2', ……, Ik'. Through standardized transformation, different types of data have a unified dimension and comparable scale, which is convenient for subsequent analysis.

[0038] Step S12242: Perform time-sliding window matching on the dimensionless control parameter sequence and the dimensionless performance index sequence, and determine the response delay time window between each control parameter and the corresponding performance index through cross-correlation calculation.

[0039] Set a suitable time-sliding window size, for example, the window size is 5 time units. Take the sequence of set values of the equipment sorting medium density D1', D2', ……, Dn' in the dimensionless control parameter sequence as an example, and match it with the sequence of the actual sorting efficiency G1', G2', ……, Gn' of the equipment in the dimensionless performance index sequence. Within the sliding window, through the cross-correlation calculation method, analyze the correlation between the change in the set value of the equipment sorting medium density and the change in the actual sorting efficiency of the equipment. Through this calculation, determine the response delay time window for the change in the set value of the equipment sorting medium density to affect the actual sorting efficiency of the equipment. For example, after calculation, it is found that when the set value of the equipment sorting medium density changes, the actual sorting efficiency of the equipment shows an obvious response after 3 time units, then these 3 time units are the response delay time window between the set value of the equipment sorting medium density and the actual sorting efficiency of the equipment. Using the same method, determine the response delay time windows between the equipment vibration frequency parameter and the energy consumption rate, the equipment vibration amplitude, and between the sorting time control parameter and the corresponding performance index.

[0040] Step S12243: Perform time shift compensation processing on the dimensionless control parameter sequence according to the response delay time window to generate a time-shifted control parameter sequence that is time-aligned with the dimensionless performance index sequence.

[0041] According to the response delay time window determined above, adjust the dimensionless control parameter sequence. Take the sequence of set values of the equipment sorting medium density D1', D2', ……, Dn' as an example. If the response delay time window between it and the sequence of the actual sorting efficiency G1', G2', ……, Gn' of the equipment is 3 time units, shift the sequence D1', D2', ……, Dn' as a whole backward by 3 time units to obtain the time-shifted sequence D1'', D2'', ……, Dn'', so that this time-shifted sequence is time-aligned with the sequence of the actual sorting efficiency G1', G2', ……, Gn' of the equipment. Similarly, perform corresponding time shift compensation processing on the equipment vibration frequency parameter sequence and the sorting time control parameter sequence to generate a time-shifted control parameter sequence that is time-aligned with the corresponding dimensionless performance index sequence.

[0042] Step S12244: Perform multiple linear regression analysis on the time-shifted control parameter sequence and the dimensionless performance index sequence, and extract the regression coefficients of each control parameter as parameter sensitivity characteristics.

[0043] Perform multiple linear regression analysis on the time-shift control parameter sequence and the dimensionless efficiency index sequence. Taking the time-shifted setpoint sequences D1'', D2'', …, Dn'' of the equipment separation medium density and the actual separation efficiency sequences G1', G2', …, Gn' of the equipment as examples, a linear relationship model between the two is established through the multiple linear regression analysis method. In this linear relationship model, the coefficient corresponding to each control parameter is the influence degree of this control parameter on the efficiency index, that is, the parameter sensitivity. For example, after regression analysis, the regression coefficient of the setpoint of the equipment separation medium density on the actual separation efficiency of the equipment is a1. This a1 represents the change degree of the actual separation efficiency of the equipment when the setpoint of the equipment separation medium density changes by one unit, that is, the parameter sensitivity of the setpoint of the equipment separation medium density. Using the same method, the regression coefficients of the equipment vibration frequency parameter and the separation time control parameter on their respective corresponding efficiency indexes are obtained as their parameter sensitivity characteristics.

[0044] Step S12245: Perform standard deviation statistics and sliding window mean filtering on the delay time data within the response delay time window to generate the equipment response lag feature.

[0045] Analyze the delay time data within the response delay time window between each control parameter and the corresponding efficiency index. Taking the response delay time window between the setpoint of the equipment separation medium density and the actual separation efficiency of the equipment as an example, assuming that the delay time data within the response delay time window is t1, t2, t3 (corresponding to the previously determined delay times), first calculate the standard deviation of these data to measure the dispersion degree of the delay time. Then, adopt the sliding window mean filtering method, set a suitable sliding window size, for example, the window size is 2, and process the delay time data. First calculate (t1 + t2) ÷ 2 to get the first filtered value, then calculate (t2 + t3) ÷ 2 to get the second filtered value, and so on, to obtain a series of processed values. These values together form the equipment response lag feature, reflecting the delay characteristic between the change of the equipment control parameter and the response of the efficiency index.

[0046] Step S1225: Construct an equipment dynamic response model based on the equipment response lag feature to generate an equipment control compensation coefficient.

[0047] Utilize the device response lag feature to construct a device dynamic response model. This device dynamic response model can be established based on general mathematical principles or empirical formulas in the prior art, which describes the relationship between changes in device control parameters and device responses. Taking the set value of the sorting medium density of the device as an example, according to its response lag feature and other relevant factors, a device control compensation coefficient is calculated and generated through the device dynamic response model. For example, the device dynamic response model can take into account factors such as response delay time, stability of the delay time (reflected by the standard deviation), etc., and through calculation, obtain the device control compensation coefficient b1 for the set value of the sorting medium density of the device. Similarly, for the device vibration frequency parameter and the sorting time control parameter, according to their respective response lag features, the corresponding device control compensation coefficients b2 and b3 are calculated and generated respectively through the device dynamic response model. These device control compensation coefficients reflect the degree of adjustment required for each control parameter to more effectively control the device operation and compensate for the impact brought by the response lag.

[0048] Step S1226: Perform feature fusion processing on the parameter sensitivity feature and the device control compensation coefficient to generate a device effectiveness dynamic feature including control delay compensation feature, effectiveness fluctuation suppression feature, and parameter optimization margin feature.

[0049] First, comprehensively consider each regression coefficient in the parameter sensitivity feature, such as the regression coefficient a1 corresponding to the set value of the sorting medium density of the device, the regression coefficient a2 corresponding to the device vibration frequency parameter, and the regression coefficient a3 corresponding to the sorting time control parameter, together with the corresponding device control compensation coefficients b1, b2, and b3.

[0050] For the control delay compensation feature, combine the parameter sensitivity and the device control compensation coefficient to analyze how to adjust the control parameter to compensate for the response delay. For example, since the parameter sensitivity of the set value of the sorting medium density of the device is a1 and the control compensation coefficient is b1, then according to these two values and the device response lag feature, determine how to adjust the set value when controlling the sorting medium density of the device to compensate for the response delay. This adjustment method and degree constitute a part of the control delay compensation feature. Similar analyses are also performed for the device vibration frequency parameter and the sorting time control parameter, and comprehensively form control delay compensation feature data c1, c2, c3..., which reflect the compensation situation for the control delay from different control parameter perspectives.

[0051] In terms of the performance fluctuation suppression feature, based on the parameter sensitivity and the device control compensation coefficient, consider how to suppress the fluctuation of the device operation performance. For example, the actual sorting efficiency of the device will fluctuate due to various factors, and the parameter sensitivity a1 and the control compensation coefficient b1 of the set value of the sorting medium density of the device indicate the influence degree and adjustment direction of this control parameter on the sorting efficiency. By analyzing these relationships, determine what measures to take to suppress the fluctuation of the actual sorting efficiency of the device. Similarly, conduct similar analyses on performance indicators such as the energy consumption rate and the vibration amplitude of the device to form the performance fluctuation suppression feature data d1, d2, d3... These data reflect the relevant features of suppressing performance fluctuations.

[0052] The parameter optimization margin feature is based on the parameter sensitivity and the device control compensation coefficient, and evaluates how much optimization space each control parameter has on the premise of ensuring the normal operation of the device and achieving the expected performance. For example, under the current parameter sensitivity a1 and control compensation coefficient b1 of the set value of the sorting medium density of the device, analyze how much it can be adjusted within a certain range to further optimize the device performance without affecting the stable operation of the device and the sorting efficiency. And so on, analyze the device vibration frequency parameter and the sorting time control parameter to generate the parameter optimization margin feature data e1, e2, e3...

[0053] Finally, splice the control delay compensation feature data c1, c2, c3..., the performance fluctuation suppression feature data d1, d2, d3... and the parameter optimization margin feature data e1, e2, e3... to form the device performance dynamic feature containing multi-dimensional data, comprehensively reflecting various dynamic characteristics related to the control parameters during the operation of the device.

[0054] Step S123: Perform an association mapping process on the environmental interference data in the environmental state index set and the process stability index to generate an environmental disturbance influence factor.

[0055] Step S1231: Collect the environmental temperature change data, humidity fluctuation data and dust concentration data as the environmental interference data set.

[0056] Within the production area of the coal preparation plant, data is collected by arranging multiple environmental monitoring points and using high-precision temperature sensors, humidity sensors, and dust concentration detectors. For example, the environmental temperature change data is recorded every 20 minutes, denoted as T1, T2, T3... Tx, which reflects the fluctuation of the environmental temperature over time. The humidity fluctuation data is also recorded at set time intervals, denoted as H1, H2, H3... Hy, reflecting the change in environmental humidity. The dust concentration data is obtained at different monitoring points and different times. For example, the dust concentration data obtained from the first detection at the first monitoring point is denoted as D1, and that obtained from the first detection at the second monitoring point is denoted as D2, and so on, obtaining D1, D2, D3... Dz. These data together constitute the environmental interference dataset, comprehensively reflecting the changes in interference factors in the production environment of the coal preparation plant.

[0057] Step S1232: Obtain the standard deviation of the separation efficiency, the volatility of the product quality, and the equipment failure frequency in the historical process stability indicators.

[0058] Extract relevant process stability indicators from the historical production data records of the coal preparation plant. The standard deviation of the separation efficiency is obtained by statistically calculating the actual separation efficiency data of the equipment over a past period of time, which reflects the fluctuation degree of the equipment separation efficiency. For example, calculate the standard deviation of the actual separation efficiency of the equipment every day in the past week, denoted as S1. The volatility of the product quality is calculated based on the changes in product quality indicators (such as ash content, sulfur content, etc. of clean coal), reflecting the stability of the product quality. Suppose the volatility of the product quality indicators in the past month is calculated, denoted as V1. The equipment failure frequency is the number of times the equipment fails within a set time period. For example, the equipment failure frequency in the past six months is denoted as F1. These historical process stability indicators provide a reference basis for analyzing the impact of environmental interference on process stability.

[0059] Step S1233: Perform a normalization operation on the environmental interference dataset that adapts to multiple metric dimensions to generate an environmental interference composite index.

[0060] Since the ambient temperature change data, humidity fluctuation data and dust concentration data have different dimensions, normalization operations are required to comprehensively analyze their impact on process stability. For the ambient temperature change data T1, T2, T3...Tx, a normalization method based on the temperature data range and statistical characteristics is used to convert them into T1', T2', T3'...Tx', so that they are in the numerical range of 0 to 1 and have a unified dimension. Similarly, the humidity fluctuation data H1, H2, H3...Hy are normalized to obtain H1', ​​H2', H3'...Hy', and the dust concentration data D1, D2, D3...Dz are normalized to obtain D1', D2', D3'...Dz'. Then, through a comprehensive calculation method, such as the weighted summation method (the weights are determined based on actual experience or data analysis, assuming that the temperature weight is w1, the humidity weight is w2, and the dust concentration weight is w3), the comprehensive environmental interference index I=w1×T1'+w2×H1'+w3×D1' is calculated (here the data at the first time point is taken as an example, and the subsequent time points are analogous). The comprehensive environmental interference index comprehensively reflects the overall degree of environmental interference.

[0061] Step S1234: performing regression analysis on the environmental interference comprehensive index and the standard deviation of the sorting efficiency to generate a temperature-efficiency correlation coefficient and a humidity-quality correlation parameter.

[0062] Using the regression analysis method, the relationship between environmental interference and the stability of sorting efficiency is analyzed based on the comprehensive index of environmental interference I and the standard deviation of sorting efficiency S1. In the process of regression analysis, the influence of ambient temperature changes (reflected by normalized temperature data) on sorting efficiency is determined, and the temperature-efficiency correlation coefficient r1 is generated. For example, through analysis, it is found that when the ambient temperature rises to a certain extent, the sorting efficiency will change according to the proportion of r1. Similarly, the influence of ambient humidity changes (reflected by normalized humidity data) on product quality fluctuation rate (here the product quality fluctuation rate is used as an indicator related to quality) is analyzed to generate the humidity-quality correlation parameter r2. This humidity-quality correlation parameter reflects the quantitative relationship between humidity changes and product quality fluctuations.

[0063] Step S1235: Generate environmental dust impact weight based on the correlation analysis between the equipment failure frequency and the dust concentration data.

[0064] Perform a correlation analysis on the equipment failure frequency F1 and the dust concentration data D1, D2, D3... Dz. Through statistical analysis methods, determine the degree of association between the change in dust concentration and the equipment failure frequency. For example, calculate the correlation coefficient between the dust concentration and the equipment failure frequency, and determine the influence degree of environmental dust on equipment failure according to the magnitude of the correlation coefficient, and then generate the environmental dust influence weight w4. If the correlation coefficient is high, it indicates that the dust concentration has a greater impact on the equipment failure frequency, and the value of w4 is relatively large; conversely, the value of w4 is small.

[0065] Step S1236: Perform standardization transformation on the temperature-efficiency correlation coefficient, humidity-quality correlation parameter, and environmental dust influence weight, and then perform dynamic weighted summation to generate an environmental disturbance influence factor.

[0066] Perform standardization transformation on the temperature-efficiency correlation coefficient r1, humidity-quality correlation parameter r2, and environmental dust influence weight w4 to make them have a unified dimension and comparable scale. For example, adopt a standardization method based on the data range and distribution characteristics to convert r1, r2, and w4 into r1', r2', and w4' respectively. Then, according to the real-time environmental state and process requirements, determine the dynamic weights. Assume that in the current production situation, the weight of the temperature-efficiency correlation coefficient is v1, the weight of the humidity-quality correlation parameter is v2, and the weight of the environmental dust influence weight is v3. Calculate the environmental disturbance influence factor E = v1×r1' + v2×r2' + v3×w4' by means of dynamic weighted summation. This environmental disturbance influence factor comprehensively reflects the influence degree of environmental factors on process stability and provides an important reference for subsequent process adjustment.

[0067] Step S124: Based on the dynamic change characteristics of the raw materials and the dynamic characteristics of the equipment efficiency, perform feature space fusion processing to generate process fitness characteristics.

[0068] Step S1241: After unifying the time units of the fluctuation period feature in the dynamic change characteristics of the raw materials and the control delay compensation feature in the dynamic characteristics of the equipment efficiency, perform time domain matching analysis to generate process time sequence synchronization parameters.

[0069] In the dynamic change characteristics of raw materials, the fluctuation cycle characteristics include a series of data reflecting the fluctuation cycle of raw material characteristics. For example, the fluctuation cycle data of raw material ash is denoted as P1, P2, P3... In the dynamic characteristics of equipment efficiency, the control delay compensation characteristics also have corresponding data, such as the control delay compensation data for the set value of the equipment separation medium density, denoted as Q1, Q2, Q3... First, check whether the time units of the two are consistent. If not, perform a unified conversion of the time units. Assume that the time unit of the fluctuation cycle characteristic data is hours, while the time unit of the control delay compensation characteristic data is minutes. Convert the time unit of the control delay compensation characteristic data to hours. Then, perform a time-domain matching analysis to compare the changes of the fluctuation cycle characteristic data and the control delay compensation characteristic data on the time axis. For example, observe the time correspondence between the change of the raw material ash fluctuation cycle and the adjustment of the control delay compensation of the equipment separation medium density set value. By analyzing the synchronization or delay of the two changes, generate process timing synchronization parameters R1, R2, R3... These parameters reflect the matching degree of the raw material characteristic fluctuation and the equipment control response in time.

[0070] Step S1242: Perform a spatial superposition process on the abnormal offset characteristic and the equipment efficiency fluctuation suppression characteristic to generate a process abnormal suppression factor.

[0071] The abnormal offset characteristic in the dynamic change characteristics of raw materials, such as the abnormal offset data S1, S2, S3... for raw material sulfur content, reflects the degree to which the raw material characteristics deviate from the normal range. The efficiency fluctuation suppression characteristic in the dynamic characteristics of equipment efficiency, such as the data T1, T2, T3... for suppressing the fluctuation of the actual separation efficiency of the equipment, reflects the suppression of the equipment operation efficiency fluctuation. Perform a spatial superposition process on these two sets of characteristics, and conduct a combined analysis of the raw material abnormal offset characteristic data and the equipment efficiency fluctuation suppression characteristic data in multiple dimensions. For example, for each time point, combine the degree of abnormal offset of the raw material sulfur content with the suppression of the fluctuation of the actual separation efficiency of the equipment, and consider whether the measures for suppressing the fluctuation of the actual separation efficiency of the equipment are effective when the raw material sulfur content shows a large abnormal offset. Through this analysis, generate process abnormal suppression factors U1, U2, U3... These factors reflect the ability of the equipment and the process to suppress abnormalities in the face of raw material abnormalities.

[0072] Step S1243: Perform a probability distribution fitting on the trend prediction characteristic and the parameter optimization margin characteristic to generate a process optimization potential evaluation value.

[0073] The trend prediction features in the dynamic change characteristics of raw materials, such as the trend prediction data V1, V2, V3... for the change in the particle size of raw materials, predict the future change trends of raw material characteristics. The parameter optimization margin feature in the dynamic characteristics of equipment efficiency, such as the data W1, W2, W3... for the parameter optimization margin of equipment vibration frequency, evaluates the optimization space of equipment control parameters. Probability distribution fitting is performed on these two sets of features to analyze the relationship between the change trend of raw material characteristics and the optimization margin of equipment control parameters. For example, assuming that the particle size of raw materials has an increasing trend, analyze how the optimization margin of the equipment vibration frequency parameter changes under this trend. Through the analysis of a large amount of data, the probability distribution relationship between the two is fitted to generate process optimization potential evaluation values X1, X2, X3... These values reflect the potential ability of the process in terms of equipment parameter optimization based on the change trend of raw material characteristics.

[0074] Step S1244: Based on the process timing synchronization parameters, construct a process dynamic response model and generate a process stability prediction index.

[0075] Construct a process dynamic response model using the process timing synchronization parameters R1, R2, R3... This process dynamic response model can describe the dynamic relationship between the fluctuation of raw material characteristics and the equipment control response based on common mathematical principles or empirical formulas in the prior art. For example, the process dynamic response model may consider the time matching degree reflected by the process timing synchronization parameters, as well as other relevant factors, such as the change range of raw material characteristics and the adjustment range of equipment control parameters. Through this process dynamic response model, predict the process stability and generate process stability prediction indicators Y1, Y2, Y3... These process stability prediction indicators can help determine the stability degree of the process in the future for a period of time under the current raw material and equipment states.

[0076] Step S1245: Perform multi-dimensional space mapping processing on the process anomaly suppression factor and the process optimization potential evaluation value to generate a process fitness feature including real-time adjustment ability features, anti-interference ability features, and optimization space features.

[0077] Perform multi-dimensional space mapping processing on process anomaly suppression factors U1, U2, U3... and process optimization potential evaluation values X1, X2, X3... In the multi-dimensional space, analyze the mutual relationship between the process anomaly suppression factors and the process optimization potential evaluation values in different dimensions. For example, from the perspective of real-time adjustment ability, consider that when a process anomaly occurs, based on the process anomaly suppression factors and the process optimization potential evaluation values, analyze whether the equipment control parameters can be quickly and effectively adjusted to adapt to raw material changes, and generate real-time adjustment ability characteristic data Z1, Z2, Z3... From the perspective of anti-interference ability, combine the process anomaly suppression factors and the process optimization potential evaluation values to evaluate the resistance ability of the process in the face of environmental interference and raw material anomalies, and generate anti-interference ability characteristic data A1, A2, A3... From the perspective of optimization space, based on the process optimization potential evaluation values and the process anomaly suppression factors, determine how much optimization space the process has on the premise of ensuring stable operation, and generate optimization space characteristic data B1, B2, B3... Finally, splice the real-time adjustment ability characteristic data Z1, Z2, Z3..., the anti-interference ability characteristic data A1, A2, A3... and the optimization space characteristic data B1, B2, B3... to form a process fitness characteristic containing multi-dimensional data, comprehensively reflecting the adaptability of the process to changes in raw materials, equipment and environment.

[0078] Step S125: Perform dynamic weight superposition processing on the process fitness characteristic and the environmental disturbance influence factor to generate a process process characteristic set including process deviation compensation characteristics, equipment load balancing characteristics and environmental anti-interference characteristics.

[0079] For example, step S1251: Calculate the dynamic weight distribution coefficient according to the equipment load rate and the environmental disturbance comprehensive index in the real-time process state data.

[0080] In the real-time process state data, the equipment load rate is obtained through monitoring equipment, assumed to be recorded as L1, L2, L3..., which reflects the working load degree of the equipment at different time points. The environmental disturbance comprehensive index I is calculated previously and reflects the overall degree of environmental disturbance. A calculation method based on the equipment load rate and the environmental disturbance comprehensive index is used to generate the dynamic weight distribution coefficient. For example, a functional relationship may be adopted, such as the dynamic weight distribution coefficient K = f(L1, I), where the function f is determined according to the actual production situation and data analysis, and comprehensively considers the influence degrees of equipment load and environmental disturbance on the process, so as to determine the proportion of weight distribution.

[0081] Step S1252: Decompose the dynamic weight distribution coefficient into a process fitness weight and an environmental disturbance suppression weight.

[0082] Decompose the dynamic weight distribution coefficient K to obtain the process fitness weight k1 and the environmental disturbance suppression weight k2. This decomposition process can be determined according to process requirements and experience. For example, based on past production data and analysis of process stability, determine the values of the process fitness weight k1 and the environmental disturbance suppression weight k2 when the equipment load rate is high and the comprehensive environmental interference index is large. Assume that in the current situation, the proportional relationship between k1 and k2 is determined through analysis, such that k1 + k2 = K, and the values of k1 and k2 can reasonably reflect the importance of the process fitness characteristics and the environmental disturbance impact factors in subsequent processing.

[0083] Step S1253: After performing feature normalization based on the process threshold on the process fitness characteristics and the environmental disturbance impact factors, perform weighted amplification processing on the normalized process fitness characteristics based on the process fitness weight to generate enhanced process characteristics, and perform attenuation processing on the normalized environmental disturbance impact factors based on the environmental disturbance suppression weight to generate suppressed environmental characteristics.

[0084] Perform feature normalization based on the process threshold on the process fitness characteristics, such as including real-time adjustment ability characteristic data Z1, Z2, Z3..., anti-interference ability characteristic data A1, A2, A3..., and optimization space characteristic data B1, B2, B3..., and the environmental disturbance impact factor E. The process threshold is determined according to process requirements and historical data. For example, for the real-time adjustment ability characteristic data, normalize it to the range of 0 to 1. Assume that through a threshold-based normalization method, Z1, Z2, Z3... are converted to Z1', Z2', Z3'..., and similarly, other characteristic data and the environmental disturbance impact factor are normalized to obtain A1', A2', A3'..., B1', B2', B3... and E'. Then, perform weighted amplification processing on the normalized process fitness characteristics based on the process fitness weight k1. For example, for the real-time adjustment ability characteristic data Z1', Z2', Z3..., calculate k1×Z1', k1×Z2', k1×Z3... to obtain the enhanced real-time adjustment ability characteristic data, and similarly perform weighted amplification on other process fitness characteristics to generate enhanced process characteristics. At the same time, perform attenuation processing on the normalized environmental disturbance impact factor E' based on the environmental disturbance suppression weight k2, calculate k2×E' to obtain the suppressed environmental characteristics.

[0085] Step S1254: Perform feature channel splicing on the enhanced process characteristics and the suppressed environmental characteristics to generate a process process feature set including process-dominant characteristics and environmental auxiliary characteristics, where the process-dominant characteristics include the process deviation compensation characteristics and the equipment load balancing characteristics, and the environmental auxiliary characteristics include the environmental anti-interference characteristics.

[0086] Splice the dimensional data in the enhanced process features, such as the real-time adjustment ability feature data after enhancement, the anti-interference ability feature data after enhancement, the optimization space feature data after enhancement, etc., with the suppressed environmental features. For example, arrange the real-time adjustment ability feature data after enhancement in order, and then successively splice the anti-interference ability feature data after enhancement, the optimization space feature data after enhancement, and the suppressed environmental feature data to form a multi-dimensional feature set.

[0087] In this feature set, the process-dominant features play a major role, which include process deviation compensation features and equipment load balancing features. The process deviation compensation features can be reflected in aspects such as the real-time adjustment ability feature and the optimization space feature in the enhanced process features. For example, the real-time adjustment ability feature data after enhancement reflects the ability of the process to quickly adjust to compensate for process deviations in the face of dynamic changes in raw materials and changes in equipment operating conditions. Integrate these relevant data to form a process deviation compensation feature data group P1, P2, P3... The equipment load balancing features can be obtained from the optimization space feature in the enhanced process features and the analysis related to the equipment load rate. For example, by analyzing the optimization space feature data after enhancement and the equipment load rate data, determine how the equipment adjusts control parameters under different operating conditions to achieve load balancing, thereby generating an equipment load balancing feature data group Q1, Q2, Q3...

[0088] The environmental auxiliary features include environmental anti-interference features, which mainly come from the suppressed environmental features and the anti-interference ability features in the process fitness features. The suppressed environmental features reflect the weakening degree of the influence of environmental factors on the process after processing the environmental interference impact factors, while the anti-interference ability features in the process fitness features reflect the resistance ability of the process itself to environmental interference. Integrate these two parts of relevant data to generate an environmental anti-interference feature data group R1, R2, R3...

[0089] Finally, splice the process deviation compensation feature data group P1, P2, P3..., the equipment load balancing feature data group Q1, Q2, Q3... and the environmental anti-interference feature data group R1, R2, R3... together to form a complete process feature set. This process feature set comprehensively reflects the comprehensive features of the process under the influence of multiple factors such as raw materials, equipment, and environment.

[0090] Step S130: Invoke a preset optimization decision model to perform parameter prediction on the process feature set, and generate a process control parameter set.

[0091] After obtaining the process feature set, the pre-set optimization decision model is then called. This model is constructed based on the data accumulated by the coal preparation plant over a long period and process knowledge, aiming to predict appropriate process control parameters according to the current process features.

[0092] Step S131: Input the process deviation compensation feature into the first analysis module of the optimization decision model to generate an initial adjustment value for the raw material separation parameters.

[0093] Extract the process deviation compensation feature data groups P1, P2, P3... from the process feature set and input them into the first analysis module of the optimization decision model. The first analysis module is the part specifically for analyzing the process deviation compensation feature. It uses the algorithms and data structures inside the model (these algorithms and data structures are trained based on a large amount of historical data and process experience) to process the process deviation compensation feature data. For example, this first analysis module may analyze the deviation situations of the process in different aspects included in the process deviation compensation feature data, such as the deviation caused by the change of raw material characteristics and the deviation caused by the equipment operation status. Through the analysis of this deviation information, the first analysis module generates an initial adjustment value for the raw material separation parameters. Suppose the generated initial adjustment values for the raw material separation parameters are A1, A2, A3..., and these values represent the initial adjustment amplitude or direction of each relevant parameter in the raw material separation process to compensate for the process deviation.

[0094] Step S132: Input the equipment load balance feature into the second analysis module of the optimization decision model to generate a dynamic compensation value for the equipment operation parameters.

[0095] Input the equipment load balance feature data groups Q1, Q2, Q3... in the process feature set into the second analysis module of the optimization decision model. The second analysis module focuses on analyzing the equipment load balance related information. It analyzes the equipment load situation at different working stages reflected by the equipment load balance feature data and the parameter adjustment direction required to achieve load balance based on the internal mechanism of the model (also a parameter mechanism trained based on historical data and process knowledge). Through this analysis, the second analysis module generates a dynamic compensation value for the equipment operation parameters. Suppose the generated dynamic compensation values for the equipment operation parameters are B1, B2, B3..., and these dynamic compensation values represent the amount that the equipment operation parameters need to be dynamically compensated to make the equipment reach the load balance state.

[0096] Step S133: Input the environment anti-interference feature into the third analysis module of the optimization decision model to generate a set of process correction coefficients.

[0097] Input the environmental anti-interference feature data groups R1, R2, R3... in the process feature set into the third analysis module of the optimization decision model. The third analysis module is mainly responsible for processing information related to environmental anti-interference. Using the internal training processing method of the model (a processing method obtained by training data on the impact of historical environmental factors on the process), it analyzes the environmental anti-interference feature data. For example, the third analysis module will analyze the degree of influence of the environmental interference reflected by the environmental anti-interference feature data on the process stability, and the resistance ability of the process to these interferences. Through such analysis, the third analysis module generates a set of process correction coefficients. Suppose the generated set of process correction coefficients is {C11, C12, C13...; C21, C22, C23...; C31, C32, C33...}, where the coefficients in different groups may correspond to different process links or parameters, and are used to correct the process deviation that may be caused by environmental interference.

[0098] Step S134: Jointly optimize the initial adjustment value, dynamic compensation value, and the set of process correction coefficients based on the set of prior optimal control parameters associated with the current working condition label of the target production line, and generate a set of process control parameters that meet the multi-objective constraint conditions; the joint optimization process includes a hierarchical calibration operation based on process priority and a parameter boundary constraint operation based on the safe operating range of the equipment.

[0099] First, obtain the set of prior optimal control parameters associated with the current working condition label of the target production line. This set of prior optimal control parameters is obtained based on the production experience and data analysis of the coal preparation plant under the same or similar working conditions in the past, and it contains the control parameters that have been proven to be optimal under specific working conditions. Suppose the current working condition label is "specific coal type, medium output demand, moderate environmental interference", and the associated set of prior optimal control parameters is {D11, D12, D13...; D21, D22, D23...; D31, D32, D33...}.

[0100] Then, a hierarchical calibration operation based on process priorities is carried out. Process priorities are determined according to the importance of the process and the degree of influence on product quality. For example, the raw material sorting process may be set as a high priority, the equipment operation stability as a secondary priority, and the environmental adaptability as a lower priority. For the initial adjustment values A1, A2, A3... of the raw material sorting parameters, calibration is first carried out according to the process priorities. If the raw material sorting parameters have a greater impact on product quality (high priority), then based on the corresponding part {D11, D12, D13...} of the prior optimal control parameter set for the raw material sorting parameters, A1, A2, A3... are adjusted. Assuming the adjustment method is a weighted calculation according to the priority weights, and the high-priority weight is w1, then the new adjusted value of the raw material sorting parameter A1' = w1×A1 + (1 - w1)×D11 (this is only an example calculation method), and so on for calibrating other initial adjustment values.

[0101] Next, similar calibration is carried out for the dynamic compensation values B1, B2, B3... of the equipment operation parameters according to the secondary priority. Assuming the secondary priority weight is w2, referring to the corresponding part {D21, D22, D23...} of the prior optimal control parameter set for the equipment operation parameters, calculate the new dynamic compensation value of the equipment operation parameter B1' = w2×B1 + (1 - w2)×D21.

[0102] For the process correction coefficient set {C11, C12, C13...; C21, C22, C23...; C31, C32, C33...}, calibration is carried out according to the lower priority. Assuming the lower priority weight is w3, referring to the corresponding part {D31, D32, D33...} of the prior optimal control parameter set for the process correction, perform the corresponding calculation to obtain the calibrated process correction coefficient set.

[0103] After that, a parameter boundary constraint operation based on the safe operation range of the equipment is carried out. Each equipment operation parameter has its safe operation range. For example, the safe range of the equipment vibration frequency parameter is [Emin, Emax], and the safe range of the set value of the equipment sorting medium density is [Fmin, Fmax], etc. For the dynamic compensation values of the equipment operation parameters and the adjusted values of the raw material sorting parameters after hierarchical calibration, it is necessary to ensure that they are within the safe operation range of the equipment. If a calibrated dynamic compensation value B1' of an equipment operation parameter exceeds its corresponding safe range, such as B1' > Emax, then it is adjusted to Emax; if B1' < Emin, then it is adjusted to Emin. Similarly, similar boundary constraint operations are carried out for other parameters to ensure the safe operation of the equipment.

[0104] After this series of combined optimization processes, a set of process control parameters that meet multi-objective constraint conditions is generated. This set of process control parameters comprehensively considers process priorities, the safe operating range of equipment, and the characteristics of the current process, ensuring that the control parameters can both optimize the process and guarantee the safe and stable operation of the equipment. Suppose the generated set of process control parameters is {E11, E12, E13...; E21, E22, E23...; E31, E32, E33...}, where the parameters in different groups respectively correspond to different process links or equipment operating parameters.

[0105] Step S135: Correlate and store the set of process control parameters with real-time process state data to form a process control knowledge base.

[0106] Correlate the generated set of process control parameters {E11, E12, E13...; E21, E22, E23...; E31, E32, E33...} with real-time process state data. The real-time process state data includes the current set of raw material quality indicators, equipment operation indicators, environmental state indicators, and other real-time information related to the process. For example, the ash content data A1, sulfur content data B1, and particle size data C1 of the current raw material, the set value D1 of the sorting medium density of the equipment, the vibration frequency parameter E1, the environmental temperature data T1, the humidity data H1, etc.

[0107] Correspondingly store each parameter in the set of process control parameters with the corresponding real-time process state data. For example, store the part E11, E12, E13... of the process control parameters for raw material sorting parameters in association with the quality indicator data such as the ash content, sulfur content, and particle size of the current raw material; store the part E21, E22, E23... of the equipment operation parameters in association with the operation indicator data such as the set value of the sorting medium density of the equipment and the vibration frequency parameter; store the part E31, E32, E33... of the process correction parameters related to the environment in association with the state indicator data such as environmental temperature and humidity.

[0108] Through this method of associated storage, a process control knowledge base is formed. This process control knowledge base records the optimized process control parameters corresponding to different real-time process states, provides a reference basis for subsequent process control, facilitates quickly obtaining the corresponding control parameters when similar process states occur, and improves the efficiency and accuracy of process control.

[0109] Step S140: Generate a set of quality optimization instructions based on the set of process control parameters, and transmit the set of quality optimization instructions to the production control unit of the target production line to perform dynamic control operations.

[0110] After obtaining the set of process control parameters, a set of quality optimization instructions is generated based on this, and then this set of instructions is sent to the production control unit of the target production line to perform dynamic regulation operations to optimize the coal preparation quality.

[0111] Step S141: Conduct trend prediction analysis on the initial adjustment values of the raw material separation parameters to generate a separation density gradient adjustment instruction.

[0112] Extract the initial adjustment values of the raw material separation parameters from the set of process control parameters, such as A1, A2, A3,.... Conduct trend prediction analysis on these initial adjustment values. By analyzing the changes of these values in the time series and the comparison with previous similar data, predict the future change trend of the raw material separation parameters. For example, observe whether the initial adjustment values A1, A2, A3,... show an upward, downward, or stable trend.

[0113] Based on this trend prediction, generate a separation density gradient adjustment instruction. If it is predicted that the raw material separation parameters have an upward trend, and this trend indicates that it is necessary to increase the separation density gradient to ensure the coal preparation quality, then generate a separation density gradient adjustment instruction to instruct the production control unit to gradually increase the separation density gradient. Suppose the instruction content is "In the next n production cycles, increase the separation density gradient by x units in each cycle", where the values of n and x are determined according to the results of the trend prediction and the process requirements.

[0114] Step S142: Conduct load fluctuation analysis on the dynamic compensation values of the equipment operation parameters to generate an equipment vibration frequency compensation instruction.

[0115] Take out the dynamic compensation values of the equipment operation parameters from the set of process control parameters, such as B1, B2, B3,.... Conduct load fluctuation analysis on these dynamic compensation values to analyze the parameter adjustments required by the equipment due to load changes at different operation stages. For example, by analyzing the dynamic compensation values B1, B2, B3,... and the real-time data of the equipment load rate (assuming the real-time equipment load rate data is L1, L2, L3,...), determine the equipment load fluctuation situation.

[0116] According to the results of the load fluctuation analysis, generate an equipment vibration frequency compensation instruction. If it is found that the equipment load fluctuates greatly, and in order to ensure the stable operation of the equipment and the coal preparation quality, it is necessary to adjust the equipment vibration frequency, then generate the corresponding instruction. For example, the instruction may be "When the equipment load rate exceeds a certain threshold (assumed to be Lth), increase the equipment vibration frequency by y units", where the threshold Lth and the adjustment amount y are determined according to the results of the load fluctuation analysis and the equipment operation characteristics.

[0117] Step S143: Generate a separation cycle optimization instruction according to the matching degree between the set of process correction coefficients and the real-time process indicators.

[0118] Perform a matching degree analysis on the process correction coefficient set {C11, C12, C13...; C21, C22, C23...; C31, C32, C33...} and the real-time process indicators. The real-time process indicators include raw material quality indicators, equipment operation indicators, environmental status indicators, etc. For example, compare and analyze the part related to raw materials in the process correction coefficient set, C11, C12, C13..., with the quality indicators such as ash content, sulfur content, and particle size of the current raw materials; perform a matching analysis on the part related to equipment, C21, C22, C23..., with the operating parameters of the equipment (such as the set value of the sorting medium density, vibration frequency parameters, etc.); perform a comparison analysis on the part related to the environment, C31, C32, C33..., with the status indicators such as environmental temperature, humidity, and dust concentration.

[0119] Generate a sorting cycle optimization instruction according to the results of the matching degree analysis. If it is found that the matching degree between the process correction coefficient set and the real-time process indicators indicates that the current sorting cycle needs to be adjusted to adapt to the changes in raw materials, equipment, and environment, then generate the corresponding instruction. For example, when it is found that the changes in raw material characteristics and equipment operating status make the current sorting cycle too long or too short, the instruction may be "adjust the sorting cycle from the current t1 to t2", and the values of t1 and t2 are determined according to the matching degree analysis and process requirements.

[0120] Step S144: Perform strategy coding processing on the sorting density gradient adjustment instruction, equipment vibration frequency compensation instruction, and sorting cycle optimization instruction to generate a quality optimization instruction set including the execution priority order.

[0121] Collect the sorting density gradient adjustment instruction, equipment vibration frequency compensation instruction, and sorting cycle optimization instruction. Perform strategy coding processing on these instructions, and determine the execution priority order of each instruction according to various factors such as the urgency of the process, the impact on product quality, and the safety of equipment operation.

[0122] For example, assume that the raw material sorting has a greater impact on product quality, then the sorting density gradient adjustment instruction may be given a higher priority; if the instability of the equipment vibration frequency may lead to equipment failure and affect production safety, the equipment vibration frequency compensation instruction will also have a higher priority; while the sorting cycle optimization instruction may have a relatively lower priority.

[0123] Through a coding method, these instructions and their execution priority order are encoded to generate a set of quality optimization instructions. Assume that the coding method is to combine the instruction content and priority information in a set format, such as "Instruction 1 (Priority 1); Instruction 2 (Priority 2); Instruction 3 (Priority 3)". Here, Instruction 1 represents the instruction for adjusting the sorting density gradient, Instruction 2 represents the instruction for compensating the vibration frequency of the device, and Instruction 3 represents the instruction for optimizing the sorting cycle. The numbers in the parentheses indicate the execution priority order. The set of quality optimization instructions formed in this way clarifies the execution order of each instruction to ensure the orderly progress of process control.

[0124] Step S145: During the policy coding process, monitor the response status of the device execution unit in real time and dynamically adjust the execution timing parameters in the set of quality optimization instructions.

[0125] While performing policy coding on the instruction for adjusting the sorting density gradient, the instruction for compensating the vibration frequency of the device, and the instruction for optimizing the sorting cycle, monitor the response status of the device execution unit in real time. After receiving the instruction, the device execution unit will have corresponding responses, such as whether the device successfully executes the instruction, the time delay in executing the instruction, and whether any abnormalities occur during the execution process.

[0126] According to the response status of the device execution unit, dynamically adjust the execution timing parameters in the set of quality optimization instructions. For example, if the device execution unit experiences a delay or abnormality when executing the instruction for adjusting the sorting density gradient, it indicates that the execution of this instruction may be affected by other factors. At this time, it is necessary to re-evaluate the execution priority and execution time of the instruction. It may advance the execution time of the instruction for compensating the vibration frequency of the device or the instruction for optimizing the sorting cycle, or adjust the timing parameters such as the number of executions and the execution interval of the instruction for adjusting the sorting density gradient.

[0127] Through this real-time monitoring and dynamic adjustment, ensure that the set of quality optimization instructions can be optimized according to the actual execution situation of the device, and ensure that the process control operation can be carried out smoothly and efficiently to achieve the purpose of optimizing the coal preparation quality.

[0128] Step S146: Perform format adaptation processing on the set of quality optimization instructions and the device control protocol to generate an executable control signal stream.

[0129] After the set of quality optimization instructions is generated, since different devices may have different control protocols, it is necessary to perform format adaptation processing on the set of quality optimization instructions and the device control protocol. The device control protocol stipulates the instruction formats that the device can recognize and execute.

[0130] For example, for a specific model of coal preparation equipment, its control protocol requires that instructions be sent in a set binary coding format, and each instruction field has a set length and meaning. The instructions in the quality optimization instruction set are format-converted according to the requirements of the equipment control protocol. For the sorting density gradient adjustment instruction, its content is encoded according to the format specified by the equipment control protocol to determine the start bit, data bit, check bit, etc. of the instruction; similarly, format conversions are performed on the equipment vibration frequency compensation instruction and the sorting cycle optimization instruction.

[0131] After format adaptation processing, an executable control signal stream is generated. This executable control signal stream can be accurately received and executed by the execution unit of the target device. According to the provisions of the equipment control protocol, the executable control signal stream arranges and encodes each instruction in the quality optimization instruction set in a suitable format to form a continuous signal sequence.

[0132] For example, each instruction segment in the signal stream contains content such as the instruction type identifier, parameter information, and check code. Taking the sorting density gradient adjustment instruction as an example, in the signal stream, first there is a set identifier bit to indicate that this is the sorting density gradient adjustment instruction, followed by the adjustment parameter information encoded according to the equipment control protocol, such as the adjustment amplitude and the start time of the adjustment, and finally the check code for verifying the correctness of the instruction transmission. Similarly, the equipment vibration frequency compensation instruction and the sorting cycle optimization instruction are also encoded and arranged in the signal stream in a similar manner.

[0133] When the executable control signal stream is transmitted to the production control unit of the target production line, the production control unit parses the signal stream according to the equipment control protocol. It first identifies the type identifier of each instruction segment to determine whether it is a sorting density gradient adjustment instruction, an equipment vibration frequency compensation instruction, or a sorting cycle optimization instruction. Then, according to the instruction type, the corresponding parameter information is extracted from the signal stream. For example, for the sorting density gradient adjustment instruction, parameters such as the adjustment amplitude and the start time are extracted. Finally, the extracted instructions and parameters are verified using the check code to ensure that no errors occur during the instruction transmission.

[0134] If the verification passes, the production control unit will execute each quality optimization instruction in sequence according to the execution priority order specified in the quality optimization instruction set. For the sorting density gradient adjustment instruction with a higher priority, the production control unit will first control the relevant equipment to adjust the sorting density gradient according to the extracted parameter information. For example, by adjusting the working parameters of some components of the sorting equipment, the change of the sorting density gradient is realized. Then, the equipment vibration frequency compensation instruction is executed in sequence, and the vibration frequency of the equipment is compensated and adjusted according to the parameters in the instruction to ensure the stable operation of the equipment under different load conditions. Finally, the sorting cycle optimization instruction is executed to adjust the sorting cycle of the equipment to match the current raw material characteristics, equipment status and environmental conditions.

[0135] During the execution process, the production control unit will monitor the operation status of the equipment in real time to ensure that each instruction is correctly executed. If an abnormal situation occurs during the execution of a certain instruction, such as the equipment being unable to adjust the sorting density gradient as required by the instruction, the production control unit will record the relevant error information and take corresponding measures according to the preset exception handling mechanism. This may include resending the instruction, adjusting the instruction parameters or pausing the entire regulation operation, and issuing an alarm to the operator at the same time for timely problem handling.

[0136] In this way, the executable control signal flow accurately transmits the quality optimization instruction set to the production control unit and guides it to perform dynamic regulation operations, realizing precise control of the coal washing process, thereby improving the coal washing quality.

[0137] Step S150: Update the parameter weight distribution of the optimization decision model according to the quality feedback index set after the regulation execution.

[0138] After the production control unit executes the dynamic regulation operation, it is necessary to collect the relevant quality feedback index set to update the parameter weight distribution of the optimization decision model, so that the model can better adapt to the actual production situation and continuously improve the control effect of the coal washing quality.

[0139] Step S151: Collect the product quality index, equipment energy consumption index and process stability index after the dynamic regulation operation is executed.

[0140] After the dynamic regulation operation is executed for a period of time, special detection equipment and methods are used to collect various quality feedback indicators. For the product quality index, it is obtained by comprehensively detecting the produced clean coal and other products. For example, detecting the ash content, sulfur content, particle size and other indicators of the clean coal. Suppose the ash content index values of the clean coal are M1, M2, M3... respectively, the sulfur content index values are N1, N2, N3... respectively, and the particle size index values are O1, O2, O3... respectively. These values reflect the situation of the product quality after the regulation operation.

[0141] The equipment energy consumption indicators are collected by energy monitoring equipment. The monitoring equipment records the energy consumption of the equipment after the regulation operation is executed, such as the electrical energy consumption of the equipment per unit time. Suppose the collected equipment energy consumption indicator values are P1, P2, P3... These values reflect the energy consumption level of the equipment under the new regulation parameters.

[0142] The process stability indicators are obtained by analyzing various parameters in the process. For example, analyzing the operation stability of the equipment, the fluctuation of product quality, etc. The process stability can be measured by calculating the standard deviation of the equipment operation parameters, the coefficient of variation of the product quality indicators, etc. Suppose the calculated standard deviation of the equipment operation parameters is Q1, Q2, Q3... and the coefficient of variation of the product quality indicators is R1, R2, R3... These are used as the process stability indicators to reflect the stability degree of the process after the regulation operation.

[0143] Step S152: Calculate the deviation degree of the product quality indicator from the preset quality standard to generate a quality error compensation factor.

[0144] Compare the collected product quality indicators, such as the clean coal ash content indicator values M1, M2, M3..., the sulfur content indicator values N1, N2, N3..., and the particle size indicator values O1, O2, O3... with the preset quality standard respectively. The preset quality standard is determined according to the product quality requirements of the coal preparation plant and industry standards. For example, the preset standard value of clean coal ash content is M0, the preset standard value of sulfur content is N0, and the preset standard value of particle size is O0.

[0145] For the ash content indicator, the deviation degree can be measured by the ratio of the absolute value of the difference between the two to the preset standard value, that is, calculate |M1 - M0| / M0 to get the ash content deviation degree value S1, and calculate |M2 - M0| / M0 to get S2, and so on. Similar methods are used for the sulfur content and particle size indicators. Calculate |N1 - N0| / N0 to get the sulfur content deviation degree value T1, |N2 - N0| / N0 to get T2, and so on; calculate |O1 - O0| / O0 to get the particle size deviation degree value U1, |O2 - O0| / O0 to get U2, and so on.

[0146] Then, generate a quality error compensation factor through a comprehensive calculation method. For example, different weights can be assigned according to the influence degree of indicators such as ash content, sulfur content, and particle size on product quality. Suppose the ash content weight is w1, the sulfur content weight is w2, and the particle size weight is w3. Calculate the quality error compensation factor V = w1×S1 + w2×T1 + w3×U1 (taking the data of the first time point as an example, and so on for subsequent time points). This quality error compensation factor comprehensively reflects the deviation degree of the product quality from the preset standard and is used for subsequent adjustment of the optimization decision-making model.

[0147] Step S153: Compare and analyze the equipment energy consumption index with the historical optimal energy consumption data to generate an energy consumption optimization correction amount.

[0148] Obtain the historical optimal energy consumption data, which are the lowest energy consumption levels achieved by the coal preparation plant under similar production conditions during previous production processes. Assume that the energy consumption values of the historical optimal energy consumption data within the same time period are P01, P02, P03, ….

[0149] Compare and analyze the collected equipment energy consumption index values P1, P2, P3, … with the historical optimal energy consumption data. Calculate the difference between the equipment energy consumption index and the historical optimal energy consumption data at each time point. For example, P1 - P01 gives the energy consumption difference W1, P2 - P02 gives W2, ….

[0150] Then, generate an energy consumption optimization correction amount by analyzing and processing these energy consumption differences. For example, the average value or weighted average value of these energy consumption differences can be calculated (the weights can be determined according to factors such as the similarity of production conditions). Assume that the calculated energy consumption optimization correction amount is X, which reflects the gap between the current equipment energy consumption and the historical optimal level and provides an adjustment basis for the energy consumption aspect of the optimization decision model.

[0151] Step S154: Perform a dynamic trend prediction on the process stability index to generate a process fluctuation warning parameter.

[0152] Perform a dynamic trend prediction on the process stability index, such as the standard deviation of equipment operation parameters Q1, Q2, Q3, …, and the coefficient of variation of product quality indicators R1, R2, R3, …. Utilize the historical process stability data accumulated by the coal preparation plant and relevant prediction models to analyze the changing trend of the current process stability index.

[0153] For example, observe whether the standard deviation of equipment operation parameters Q1, Q2, Q3, … shows a gradually increasing or decreasing trend, and the changes in the coefficient of variation of product quality indicators R1, R2, R3, …. Through the trend analysis of these data, predict the changes in process stability in the future period.

[0154] Generate a process fluctuation warning parameter according to the prediction results. If it is predicted that the process stability has a deteriorating trend, for example, the standard deviation of equipment operation parameters gradually increases and the coefficient of variation of product quality indicators also rises, indicating that the process may have large fluctuations, generate the corresponding warning parameter. Assume that the generated process fluctuation warning parameter is Y, which can be a value comprehensively considering the changing trends of equipment operation parameters and product quality indicators and is used to remind the optimization decision model to pay more attention to process stability and make corresponding adjustments.

[0155] Step S155: Input the quality error compensation factor, energy consumption optimization correction amount, and process fluctuation warning parameter into the weight update module of the optimization decision model to generate a parameter weight adjustment gradient.

[0156] Input the quality error compensation factor V, the energy consumption optimization correction amount X, and the process fluctuation warning parameter Y into the weight update module of the optimization decision model. The weight update module is a part of the optimization decision model that is specifically used to adjust the parameter weights of the model according to the feedback information. It processes the input information based on its internal algorithms and mechanisms.

[0157] The weight update module analyzes the product quality deviation reflected by the quality error compensation factor V, the energy consumption gap reflected by the energy consumption optimization correction amount X, and the process stability trend indicated by the process fluctuation warning parameter Y, and comprehensively considers the influence of these factors on the model parameter weights. For example, according to the quality error compensation factor V, if the product quality deviation is large, the weight update module may adjust the parameter weights related to product quality, so that the model pays more attention to the optimization of product quality in subsequent predictions; according to the energy consumption optimization correction amount X, if the energy consumption is high, it will adjust the weights related to the equipment operation parameters to seek the control parameters for reducing energy consumption; according to the process fluctuation warning parameter Y, if the process stability shows a deteriorating trend, it will adjust the parameter weights related to process stability.

[0158] Through this analysis and adjustment, the weight update module generates a parameter weight adjustment gradient. Suppose the generated parameter weight adjustment gradient is a set of values {Z1, Z2, Z3...}, and these values respectively correspond to the weight adjustment directions and amplitudes of different parameters in the optimization decision model.

[0159] Step S156: Based on the parameter weight adjustment gradient, perform iterative optimization processing on the decision layer parameters of the optimization decision model to complete the dynamic update of the model parameters, where the decision layer parameters include the parameters of the first parsing module, the second parsing module, and the third parsing module.

[0160] Use the generated parameter weight adjustment gradient {Z1, Z2, Z3...} to perform iterative optimization processing on the decision layer parameters of the optimization decision model. The decision layer parameters include the parameters of the first parsing module, the second parsing module, and the third parsing module.

[0161] For the first parsing module, assuming its parameters are {A11, A12, A13...}, adjust the corresponding part Z1 in the gradient according to the parameter weights to adjust these parameters. For example, an update formula may be used, such as A11' = A11 + Z1 × ΔA11 (where ΔA11 is an adjustment step determined according to the model structure and algorithm), and so on to update other parameters of the first parsing module, obtaining the updated parameters {A11', A12', A13...}.

[0162] Similarly, for the parameters {B11, B12, B13...} of the second parsing module, adjust the corresponding part Z2 in the gradient according to the parameter weights and perform adjustments according to a similar update formula, such as B11' = B11 + Z2 × ΔB11, obtaining the updated parameters {B11', B12', B13...}.

[0163] For the parameters {C11, C12, C13...} of the third parsing module, update the parameters based on the corresponding part Z3 in the gradient adjusted by the parameter weights, such as C11' = C11 + Z3 × ΔC11, obtaining the updated parameters {C11', C12', C13...}.

[0164] Through such iterative optimization processing, the dynamic update of the parameters of the optimization decision model is completed. The updated model can better adapt to the actual production situation. When predicting parameters according to the process feature set subsequently, it can generate a more accurate set of process control parameters, further improving the effect of intelligent control of coal preparation quality. After multiple such regulation and model update processes, the optimization decision model will be continuously optimized, making the coal preparation production process more stable and efficient, and the product quality more in line with the preset standards.

[0165] Figure 2 The figure shows a schematic diagram of exemplary hardware and software components of a management system 100 for intelligent control of coal preparation quality that can implement the idea of the present invention provided by some embodiments of the present invention. For example, the processor 120 can be used on the management system 100 for intelligent control of coal preparation quality and is used to execute the functions in the present invention.

[0166] The management system 100 for intelligent control of coal preparation quality can be a general-purpose server or a special-purpose server, both of which can be used to implement the management method for intelligent control of coal preparation quality of the present invention. Although only one server is shown in the present invention, for convenience, the functions described in the present invention can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0167] For example, the management system 100 for intelligent control of coal preparation quality may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as magnetic disks, ROM, or RAM, or any combination thereof. Exemplarily, the management system 100 for intelligent control of coal preparation quality may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present invention can be implemented according to the above program instructions. The management system 100 for intelligent control of coal preparation quality further includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0168] For ease of explanation, only one processor is described in the management system 100 for intelligent control of coal preparation quality. However, it should be noted that the management system 100 for intelligent control of coal preparation quality in the present invention may also include multiple processors. Therefore, the steps performed by one processor described in the present invention can also be jointly executed or separately executed by multiple processors. For example, if the processor of the management system 100 for intelligent control of coal preparation quality executes step A and step B, it should be understood that step A and step B can also be jointly executed by two different processors or separately executed in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0169] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned management method for intelligent control of coal preparation quality is implemented.

[0170] It should be noted that, in order to simplify the expression of the disclosure of the present invention and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.

Claims

1. A management method for intelligent control of coal preparation quality, characterized in that: The method comprises: Obtain the raw material quality index set, equipment operation index set and environmental status index set of the target production line; Dynamically correlating the raw material quality index set, the equipment operation index set and the environmental status index set to generate a process feature set; Calling a preset optimization decision model to perform parameter prediction on the process feature set to generate a process control parameter set; Generating a quality optimization instruction set based on the process control parameter set, and transmitting the quality optimization instruction set to the production control unit of the target production line to perform dynamic control operations; The parameter weight distribution of the optimization decision model is updated according to the quality feedback indicator set after the regulation is executed.

2. The management method for intelligent control of coal preparation quality according to claim 1 is characterized in that: The dynamically associating the raw material quality index set, the equipment operation index set and the environmental status index set to generate a process feature set includes: Performing time series fluctuation analysis on the raw material characteristic change data in the raw material quality index set to generate dynamic change characteristics of the raw materials; Performing coupling matching analysis on the equipment control parameters and the operation efficiency data in the equipment operation index set to generate dynamic characteristics of equipment efficiency; Performing correlation mapping processing on the environmental disturbance data in the environmental status indicator set and the process stability indicator to generate an environmental disturbance impact factor; Based on the dynamic change characteristics of the raw materials and the dynamic characteristics of the equipment efficiency, feature space fusion processing is performed to generate process adaptability characteristics; The process fitness characteristics and environmental disturbance influencing factors are dynamically weighted and superimposed to generate a process feature set including process deviation compensation characteristics, equipment load balancing characteristics and environmental anti-interference characteristics.

3. The management method for intelligent control of coal preparation quality according to claim 2 is characterized in that: The calling of the preset optimization decision model to perform parameter prediction on the process feature set to generate a process control parameter set includes: Inputting the process deviation compensation feature into the first analytical module of the optimization decision model to generate an initial adjustment value of the raw material sorting parameter; Inputting the equipment load balancing characteristics into the second analytical module of the optimization decision model to generate dynamic compensation values ​​of equipment operation parameters; Inputting the environmental anti-interference characteristics into the third analytical module of the optimization decision model to generate a set of process correction coefficients; Based on the a priori optimal control parameter set associated with the current operating condition label of the target production line, the initial adjustment value, the dynamic compensation value and the process correction coefficient set are jointly optimized to generate a process control parameter set that meets the multi-objective constraint conditions; the joint optimization process includes a hierarchical calibration operation based on process priority and a parameter boundary constraint operation based on the safe operating range of the equipment; The process control parameter set is associated with the real-time process status data and stored to form a process control knowledge base.

4. The management method for intelligent control of coal preparation quality according to claim 3 is characterized in that: The generating of a quality optimization instruction set based on the process control parameter set includes: Performing trend forecasting analysis on the initial adjustment values ​​of the raw material sorting parameters to generate sorting density gradient adjustment instructions; Performing load fluctuation analysis on the dynamic compensation value of the equipment operating parameter to generate equipment vibration frequency compensation instructions; Generate a sorting cycle optimization instruction according to the matching degree between the process correction coefficient set and the real-time process index; Performing strategy coding processing on the separation density gradient adjustment instruction, the equipment vibration frequency compensation instruction and the separation cycle optimization instruction to generate a quality optimization instruction set including an execution priority order; During the strategy coding process, real-time monitoring of the response status of the device execution unit is performed, and the execution timing parameters in the quality optimization instruction set are dynamically adjusted; The quality optimization instruction set is format-adapted to the device control protocol to generate an executable control signal stream.

5. The management method for intelligent control of coal preparation quality according to claim 3 is characterized in that: The updating of the parameter weight distribution of the optimization decision model according to the quality feedback indicator set after the control is executed includes: Collect product quality indicators, equipment energy consumption indicators and process stability indicators after dynamic control operations; Calculating the deviation between the product quality index and the preset quality standard to generate a quality error compensation factor; Compare and analyze the energy consumption index of the equipment with the historical optimal energy consumption data to generate energy consumption optimization correction amount; Performing dynamic trend prediction on the process stability index and generating process fluctuation warning parameters; Inputting the quality error compensation factor, energy consumption optimization correction amount and process fluctuation warning parameter into the weight updating module of the optimization decision model to generate a parameter weight adjustment gradient; The decision layer parameters of the optimization decision model are iteratively optimized based on the parameter weight adjustment gradient to complete the dynamic update of the model parameters, wherein the decision layer parameters include the parameters of the first analysis module, the second analysis module and the third analysis module.

6. The management method for intelligent control of coal preparation quality according to claim 2, characterized in that: The performing of time series fluctuation analysis on the raw material characteristic change data in the raw material quality index set to generate the raw material dynamic change characteristics includes: Performing standardization conversion and time series alignment processing on the raw material ash content fluctuation data, sulfur content distribution data and particle size change data in the raw material quality index set to generate a raw material multidimensional time series data set; Performing sliding window mean calculation on the raw material multidimensional time series data set to generate a smooth feature of the raw material characteristics; Performing fluctuation amplitude detection based on the smoothing characteristics of the raw material characteristics and generating a raw material abnormal fluctuation mark; Comparing the raw material abnormal fluctuation mark with a preset process tolerance threshold to generate a raw material quality stability assessment parameter; The raw material multidimensional time series data set is subjected to feature weighting processing according to the raw material quality stability assessment parameters to generate raw material dynamic change characteristics including fluctuation period characteristics, abnormal offset characteristics and trend prediction characteristics.

7. The management method for intelligent control of coal preparation quality according to claim 2, characterized in that: The coupling matching analysis of the equipment control parameters and the operation efficiency data in the equipment operation index set to generate the equipment efficiency dynamic characteristics includes: Extracting a device sorting medium density setting value, a device vibration frequency parameter, and a sorting time control parameter from the device operation index set as a device control parameter set; Collecting the actual sorting efficiency, energy consumption rate and vibration amplitude data of the equipment in the equipment operation index set as an operation efficiency data set; Performing time stamp alignment processing on the device control parameter set and the operation efficiency data set to generate a time series record of the device operation status; Performing standardized conversion and control parameter-performance response correlation analysis on the equipment operation status time series records in sequence to generate equipment response hysteresis characteristics and parameter sensitivity characteristics; Building a dynamic response model of the equipment based on the response hysteresis characteristics of the equipment and generating a control compensation coefficient of the equipment; The parameter sensitivity characteristics and the equipment control compensation coefficient are subjected to feature fusion processing to generate equipment performance dynamic characteristics including control delay compensation characteristics, performance fluctuation suppression characteristics and parameter optimization margin characteristics.

8. The management method for intelligent control of coal preparation quality according to claim 2, characterized in that: The step of associating and mapping the environmental disturbance data in the environmental state indicator set with the process stability indicator to generate an environmental disturbance impact factor includes: Collect environmental temperature change data, humidity fluctuation data and dust concentration data as environmental interference data sets; Obtain the standard deviation of sorting efficiency, product quality fluctuation rate and equipment failure frequency in historical process stability indicators; Performing a normalization operation adapted to multiple index dimensions on the environmental interference data set to generate an environmental interference comprehensive index; Perform regression analysis on the comprehensive environmental interference index and the standard deviation of sorting efficiency to generate a temperature-efficiency correlation coefficient and a humidity-quality correlation parameter; Generate environmental dust impact weights based on correlation analysis between the equipment failure frequency and dust concentration data; The temperature-efficiency correlation coefficient, humidity-quality correlation parameter and environmental dust impact weight are standardized and then dynamically weighted and summed to generate an environmental disturbance impact factor.

9. The management method for intelligent control of coal preparation quality according to claim 6, characterized in that: The feature space fusion processing based on the dynamic change characteristics of the raw materials and the dynamic characteristics of the equipment efficiency is performed to generate the process adaptability characteristics, including: Unifying the time units of the fluctuation period characteristics in the raw material dynamic change characteristics and the control delay compensation characteristics in the equipment efficiency dynamic characteristics, and then performing time domain matching analysis to generate process timing synchronization parameters; Performing spatial superposition processing on the abnormal deviation feature and the equipment performance fluctuation suppression feature to generate a process abnormality suppression factor; Performing probability distribution fitting on the trend prediction feature and the parameter optimization margin feature to generate a process optimization potential evaluation value; Building a process dynamic response model based on the process timing synchronization parameters to generate a process stability prediction index; The process anomaly suppression factor and the process optimization potential evaluation value are subjected to multi-dimensional space mapping processing to generate process fitness characteristics including real-time adjustment capability characteristics, anti-interference capability characteristics and optimization space characteristics.

10. A management system for intelligent control of coal preparation quality, characterized in that: It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the management method for intelligent control of coal preparation quality as described in any one of claims 1 to 9.

Images