Real-time prediction method for coal slurry concentration
Through the rolling conformal prediction method and data fusion technology, the iterative predictor is updated to solve the problems of linear relationship limitations and real-time detection difficulties in coal slurry concentration prediction, and real-time accurate prediction of coal slurry concentration is achieved.
Patent Information
- Application Number
- CN202410944227.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-15
AI Technical Summary
In the prior art, the coal slurry concentration prediction method has linear relationship limitations, poor generalization ability, and cannot realize real-time detection, resulting in limited prediction accuracy and reliability.
The rolling conformal prediction method is adopted to collect characteristic variables and coal slurry concentration data, train the predictor, and continuously update the iterative predictor using data fusion and learning-ready methods to achieve real-time prediction of coal slurry concentration.
With a small amount of label data, real-time accurate prediction of coal slurry concentration is achieved, which improves the predictor's prediction ability, and solves the problems of sparse label data and difficulty in real-time detection.
Smart Images

Figure CN118965290A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of advanced manufacturing and automation, and in particular to a real-time prediction method for coal slurry concentration. Background Art
[0002] Real-time and accurate prediction of coal slurry concentration is one of the key technologies for coal processing and utilization. It provides important indicators for the development and utilization of downstream tasks, and is of great significance in improving coal combustion efficiency, reducing production costs, and reducing environmental pollution. However, the traditional coal slurry concentration prediction method uses intermittent sampling for experimental determination. It has the disadvantages of high testing consumables and labor costs, lagging test results, and test results that are easily affected by the operating habits of testers. It cannot provide real-time, stable, and highly accurate coal slurry concentration values.
[0003] A coal slurry concentration detection method proposed in the prior art, the preset coal slurry concentration model proposed is a linear combination of the vertical pressure difference on the coal slurry pipeline, the horizontal pressure difference on the coal slurry pipeline, the coal slurry tank temperature, the temperature on the coal slurry pipeline, the water flow entering the coal mill, and the weight of coal entering the coal mill. The estimated value of the combination coefficient is obtained by training with historical labeled data, thereby obtaining a coal slurry concentration prediction method. However, this technology has two defects: 1. The linear relationship is very limited, it is difficult to characterize the coal slurry concentration, and the method has poor generalization ability; 2. The combination coefficient training is one-time, and the predictor is iteratively updated in a data fusion and learn-and-use manner, and the prediction accuracy and reliability are limited. Summary of the invention
[0004] The embodiment of the present invention provides a real-time prediction method for coal slurry concentration, so as to at least solve the technical problems that the label data containing coal slurry concentration is very scarce, it is difficult to train a predictor with accurate prediction ability, and it is impossible to realize real-time detection of coal slurry concentration.
[0005] According to one aspect of an embodiment of the present invention, a real-time prediction method for coal slurry concentration is provided. The method may include: collecting characteristic variables of the first time period of the target time interval, and based on the characteristic variables of the first time period, detecting the coal slurry concentration of the first time period corresponding to the characteristic variables of the first time period; training the characteristic variables of the first time period and the coal slurry concentration of the first time period through a target algorithm to obtain a target predictor, wherein the characteristic variables of the first time period and the coal slurry concentration of the first time period are labeled data sets; collecting characteristic variables of the second time period in real time, and predicting the coal slurry concentration of the second time period corresponding to the characteristic variables of the second time period based on the target predictor to obtain a predicted value of the coal slurry concentration of the second time period, wherein the characteristic variables of the second time period are used to characterize the characteristic variables of the coal slurry concentration of the second time period corresponding to the characteristic variables of the second time period that do not need to be detected, the characteristic variables of the second time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the second time period and the coal slurry concentration of the second time period predicted by the target predictor are the predicted values of the coal slurry concentration of the second time period. The degree is a labeled data set; based on the characteristic variables of the first time period, the characteristic variables of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period, a combined sample is obtained; the combined sample is input into the target predictor to update the target predictor; the characteristic variables of the third time period are collected in real time, and the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period is predicted based on the updated target predictor to obtain the predicted value of the coal slurry concentration of the third time period, wherein the characteristic variables of the third time period are used to characterize the characteristic variables of the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period that do not need to be detected, the characteristic variables of the third time period that have not been predicted by the target predictor are an unlabeled data set, and the characteristic variables of the third time period and the coal slurry concentration of the third time period that have been predicted by the target predictor are a labeled data set; the characteristic vectors of different time periods are repeatedly collected, the target predictor is iteratively updated, and the real-time coal slurry concentration is predicted.
[0006] Optionally, the characteristic variables of the first period and the coal slurry concentration of the first period are trained by a target algorithm to obtain a target predictor, including: training the characteristic variables of the first period and the coal slurry concentration of the first period by a conformal prediction algorithm of quantile regression to obtain a target predictor.
[0007] Optionally, a combined sample is obtained based on the characteristic variables of the first time period, the characteristic variables of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period, including: combining the characteristic variables of the first time period, the characteristic variables of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period to obtain a combined sample.
[0008] Optionally, the characteristic variables of the first time period, the second time period and the third time period are the same variables, and the characteristic variables of the first time period are the additive flow, the mill main motor current, the mill outlet tank agitator current, the mill outlet tank liquid level, the mill outlet tank liquid level, the speed, and the outlet pressure of the low-pressure coal slurry pump.
[0009] Beneficial effects of the present invention:
[0010] The present invention proposes a real-time prediction method for coal slurry concentration. According to the characteristic variables and the corresponding coal slurry concentrations observed based on a small number of laboratory test samples, a predictor is initially trained using a conformal prediction method; a specific step length is selected, and the coal slurry concentration corresponding to the characteristic variables under the step length is predicted based on the predictor; the laboratory test samples and the completed samples under the current step length are comprehensively used to train again using the conformal prediction method to update the predictor. In this way, the real-time prediction of coal slurry concentration is achieved with a small amount of laboratory complete labeled data, which solves the technical problem that the current labeled data containing coal slurry concentration is very scarce, it is difficult to train a predictor with accurate prediction capabilities, and it is impossible to achieve real-time detection of coal slurry concentration. The rolling conformal prediction method is used to continuously update the iterative predictor using data fusion and learn-and-use methods, and the prediction capability of the predictor is improved while achieving the technical effect of real-time prediction of coal slurry concentration. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0012] Figure 1 is a flow chart of a real-time prediction method for coal slurry concentration according to an embodiment of the present invention;
[0013] Figure 2 is a schematic diagram of a rolling conformal prediction algorithm result according to an embodiment of the present invention;
[0014] Figure 3 is a structural block diagram of a real-time prediction method for coal slurry concentration according to an embodiment of the present invention;
[0015] Figure 4 4 is a schematic diagram of a real-time prediction system for coal slurry concentration according to an embodiment of the present invention. DETAILED DESCRIPTION
[0016] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only embodiments of a part of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0017] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0018] Example 1
[0019] According to an embodiment of the present invention, a real-time prediction method for coal slurry concentration is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system comprising at least one set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.
[0020] Figure 1 is a flow chart of a real-time prediction method for coal slurry concentration according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps:
[0021] Step S101, collecting characteristic variables of a first period of a target time interval, and based on the characteristic variables of the first period of time, detecting the coal slurry concentration of the first period of time corresponding to the characteristic variables of the first period of time.
[0022] In the technical solution provided in the above step S101 of the present invention, the characteristic variables of the first time period are collected every 12 hours. The characteristic variables of the first time period in the first 12 hours, that is, at time 1, are: the additive flow rate is 435, the main motor current of the mill is 37, the agitator current of the mill outlet tank is 6.98, the liquid level of the mill outlet tank is 46.1, the liquid level of the mill outlet tank is 47.22, the rotation speed is 71.62, and the outlet pressure of the low-pressure coal slurry pump is 0.36. According to the characteristic variables of the first time period, the coal slurry concentration of the first time period corresponding to the characteristic variables of the first time period is obtained by chemical detection, and the coal slurry concentration of the first time period is 64.47.
[0023] Step S102, training the characteristic variables of the first period and the coal slurry concentration of the first period by a target algorithm to obtain a target predictor, wherein the characteristic variables of the first period and the coal slurry concentration of the first period are labeled data sets.
[0024] In the technical solution provided in the above step S102 of the present invention, the target algorithm trains the characteristic variables of the first period and the coal slurry concentration of the first period to obtain a target predictor, and determines the characteristic variables of the first period and the coal slurry concentration of the first period as a labeled data set.
[0025] Step S103, collect the characteristic variables of the second time period in real time, predict the coal slurry concentration of the second time period corresponding to the characteristic variables of the second time period based on the target predictor, and obtain the predicted value of the coal slurry concentration of the second time period, wherein the characteristic variables of the second time period are used to characterize the characteristic variables of the coal slurry concentration of the second time period that do not need to be detected corresponding to the characteristic variables of the second time period, the characteristic variables of the second time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the second time period and the coal slurry concentration of the second time period that have been predicted by the target predictor are labeled data sets.
[0026] In the technical solution provided in the above step S103 of the present invention, the characteristic variables of the second time period are collected in real time by the software, but the data set of the coal slurry concentration of the second time period is unknown. For example, one of the characteristic variables of the second time period can be: the additive flow rate is 430.25, the current of the main motor of the mill is 0, the current of the agitator of the mill outlet tank is 100, the liquid level of the mill outlet tank is 45.85, the liquid level of the mill outlet tank is -1.32, the speed is 0, and the outlet pressure of the low-pressure coal slurry pump is 0. The coal slurry concentration of the second time period corresponding to the characteristic variables of the second time period is predicted by the target predictor to obtain the predicted value of the coal slurry concentration of the second time period, wherein the coal slurry concentration of the second time period corresponds to the characteristic variables of the second time period one by one, wherein the characteristic variables of the second time period are unlabeled data sets, and the second time period can be 360.
[0027] Step S104, obtaining a combined sample based on the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period.
[0028] In the technical solution provided in the above step S104 of the present invention, a combined sample is obtained according to the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period.
[0029] Step S105: input the combined sample into the target predictor to update the target predictor.
[0030] In the technical solution provided in the above step S105 of the present invention, the combined samples are trained by the target predictor, and the target prediction predictor is trained again and iteratively updated according to the conformal prediction method of quantile regression, wherein the combined samples are labeled data sets.
[0031] Step S106, collect the characteristic variables of the third time period in real time, and predict the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period based on the updated target predictor to obtain the predicted value of the coal slurry concentration of the third time period, wherein the characteristic variables of the third time period are used to characterize the characteristic variables of the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period that do not need to be detected, the characteristic variables of the third time period that have not been predicted by the target predictor are an unlabeled data set, and the characteristic variables of the third time period and the coal slurry concentration of the third time period that have been predicted by the target predictor are a labeled data set.
[0032] In the technical solution provided in the above step S106 of the present invention, the characteristic variable of the third period is obtained, and the characteristic variable of the third period is input into the updated target predictor to obtain the predicted value of the coal slurry concentration of the third period corresponding to the characteristic variable of the third period.
[0033] Step S107, repeatedly collect feature vectors in different time periods, iteratively update the target predictor, and predict the real-time coal slurry concentration.
[0034] In the technical solution provided in step S107 of the present invention, the principle of data fusion and learning-as-you-go is adopted to repeatedly collect feature vectors of different step lengths, iteratively update the target predictor, and predict the real-time coal slurry concentration, that is, in the case of a small number of first period feature variables and first period coal slurry concentration, the real-time prediction of coal slurry concentration is achieved. The embodiment of the present invention proposes a schematic diagram of the rolling conformal prediction algorithm result, Figure 2 is a schematic diagram of a rolling conformal prediction algorithm result according to an embodiment of the present invention, such as Figure 2 As shown, the predicted coal slurry concentration fluctuates to a certain extent in the early stage of prediction and quickly reaches a stable value of 64.0927, which has good robustness.
[0035] The above method of this embodiment is further introduced below.
[0036] As an optional implementation method, step S102, training the characteristic variables of the first time period and the coal slurry concentration of the first time period through a target algorithm to obtain a target predictor, including: training the characteristic variables of the first time period and the coal slurry concentration of the first time period through a conformal prediction algorithm of quantile regression to obtain a target predictor.
[0037] In this embodiment, a conformal prediction algorithm of quantile regression is used to train the characteristic variables of the first period and the coal slurry concentration of the first period to obtain a target predictor.
[0038] As an optional implementation method, step S104 obtains a combined sample based on the characteristic variables of the first time period, the characteristic variables of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period, including: merging the characteristic variables of the first time period, the characteristic variables of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period to obtain a combined sample.
[0039] In this embodiment, the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period are combined to obtain a combined sample.
[0040] As an optional implementation method, the characteristic variables of the first time period, the characteristic variables of the second time period and the characteristic variables of the third time period are the same variables, and the characteristic variables of the first time period are the additive flow, the main motor current of the mill, the agitator current of the mill outlet tank, the liquid level of the mill outlet tank, the liquid level of the mill outlet tank, the rotation speed, and the outlet pressure of the low-pressure coal slurry pump.
[0041] In this embodiment, the first period characteristic variable, the second period characteristic variable and the third period characteristic variable are the same variable, wherein the data of the first period characteristic variable, the second period characteristic variable and the third period characteristic variable are different.
[0042] In an embodiment of the present invention, a first time period characteristic variable of a target time interval is collected, and based on the first time period characteristic variable, the first time period coal slurry concentration corresponding to the first time period characteristic variable is detected; the first time period characteristic variable and the first time period coal slurry concentration are trained by a target algorithm to obtain a target predictor, wherein the first time period characteristic variable and the first time period coal slurry concentration are labeled data sets; the second time period characteristic variable is collected in real time, and the second time period coal slurry concentration corresponding to the second time period characteristic variable is predicted based on the target predictor to obtain a predicted value of the second time period coal slurry concentration, wherein the second time period characteristic variable is used to characterize the characteristic variable of the second time period coal slurry concentration corresponding to the second time period characteristic variable that does not need to be detected, the second time period characteristic variable that has not been predicted by the target predictor is an unlabeled data set, and the second time period characteristic variable and the second time period coal slurry concentration that have been predicted by the target predictor are labeled data sets; a combined sample is obtained based on the first time period characteristic variable, the second time period characteristic variable, the first time period coal slurry concentration, and the second time period coal slurry concentration; the combined sample is input into the target In the target predictor, the target predictor is updated; the characteristic variables of the third time period are collected in real time, and the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period is predicted based on the updated target predictor to obtain the predicted value of the coal slurry concentration of the third time period, wherein the characteristic variables of the third time period are used to characterize the characteristic variables of the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period that do not need to be detected, the characteristic variables of the third time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the third time period and the coal slurry concentration of the third time period after being predicted by the target predictor are labeled data sets; the characteristic vectors of different time periods are repeatedly collected, the target predictor is iteratively updated, and the real-time coal slurry concentration is predicted, which solves the technical problems that the labeled data containing coal slurry concentration are very scarce, it is difficult to train a predictor with accurate prediction ability, and it is impossible to realize the real-time detection of coal slurry concentration, and achieves the technical effect of continuously updating the iterative predictor through the rolling conformal prediction method, using data fusion and learn-and-use methods to improve the prediction ability of the predictor while realizing the real-time prediction of coal slurry concentration.
[0043] Example 2
[0044] The technical solution of the embodiment of the present invention is illustrated below in conjunction with preferred implementation modes.
[0045] The embodiment of the present invention provides a structural block diagram of a real-time prediction method for coal slurry concentration. Figure 3 is a structural block diagram of a real-time prediction method for coal slurry concentration according to an embodiment of the present invention. Figure 3As shown, the manually collected label data in the structural flowchart are the characteristic variables sampled and collected by the laboratory every 12 hours and the data set of coal slurry concentration values obtained by laboratory testing, which are recorded as labeled data sets. The specific data value at time 1 is that when the additive flow rate is 435, the mill main motor current is 37, the mill outlet tank agitator current is 6.98, the mill outlet tank liquid level is 46.1, the mill outlet tank liquid level is 47.22, the speed is 71.62, and the outlet pressure of the low-pressure coal slurry pump is 0.36, the coal slurry concentration is 64.47.
[0046] The conformal prediction algorithm of quantile regression is used to train the obtained labeled data to obtain the target predictor f. The main idea of this conformal prediction method is to calculate the confidence score function, evaluate the uncertainty of the prediction result, and generate a prediction interval based on the uncertainty. The midpoint of the interval is used as the predicted value of the coal slurry concentration of the labeled data.
[0047] Historical unlabeled data collected by software X t-j ,.......,X t-1 , input into the training algorithm for training, update the predictor f, and obtain the coal slurry concentration value prediction Y t , where the historical unlabeled data can be the characteristic variables of the unlabeled data set with a step size of 360 data. One unlabeled data can be: the additive flow rate is 430.25, the mill main motor current is 0, the mill outlet tank agitator current is 100, the mill outlet tank liquid level is 45.85, the mill outlet tank liquid level is -1.32, the speed is 0, and the outlet pressure of the low-pressure coal slurry pump is 0. The coal slurry concentration value is predicted as Y t The data is written into a data server, wherein the data server interacts with the process production, the process production interacts with the data server by interacting with the process, and the data server transmits the algorithm result to the process production.
[0048] The manually collected labeled data and historical unlabeled data are combined into a new labeled data set. The conformal prediction method of quantile regression is used to train and iteratively update the predictor f again. The updated predictor f is used to predict the coal slurry concentration value corresponding to the characteristic variable of the next specific step.
[0049] Repeat the above steps step by step, adopt the principles of data fusion and learning-as-you-go, and realize the real-time prediction of coal slurry concentration with a small amount of manually collected complete labeled data.
[0050] In this embodiment, a predictor is initially trained using a conformal prediction method based on the characteristic variables observed by a small number of laboratory test samples and their corresponding coal slurry concentrations; a specific step length is selected, and the coal slurry concentration corresponding to the characteristic variables under the step length is predicted based on the predictor; the laboratory test samples and the completed samples under the current step length are comprehensively used to train again using the conformal prediction method to update the predictor. In this way, the real-time prediction of coal slurry concentration is achieved with a small amount of laboratory complete labeled data, which solves the technical problem that the current labeled data containing coal slurry concentration is very scarce, it is difficult to train a predictor with accurate prediction capabilities, and it is impossible to achieve real-time detection of coal slurry concentration. The rolling conformal prediction method is used to continuously update the iterative predictor using data fusion and learn-and-use methods, and the prediction capability of the predictor is improved while achieving the technical effect of real-time prediction of coal slurry concentration.
[0051] Example 3
[0052] According to an embodiment of the present invention, a real-time prediction system for coal slurry concentration is also provided. It should be noted that the real-time prediction system for coal slurry concentration can be used to execute the real-time prediction method for coal slurry concentration in Example 1.
[0053] Figure 4 Schematic diagram of a real-time prediction system for coal slurry concentration according to an embodiment of the present invention. Figure 4 As shown, a real-time prediction system for coal slurry concentration may include: a first processing unit, a training unit, a second processing unit, a obtaining unit, an updating unit, a third processing unit and a fourth processing unit.
[0054] The first processing unit is used for collecting the first time period characteristic variable of the target time interval, and based on the first time period characteristic variable, detecting the first time period coal slurry concentration corresponding to the first time period characteristic variable.
[0055] The training unit is used to train the characteristic variables of the first period and the coal slurry concentration of the first period through a target algorithm to obtain a target predictor, wherein the characteristic variables of the first period and the coal slurry concentration of the first period are labeled data sets.
[0056] The second processing unit is used to collect the characteristic variables of the second time period in real time, predict the coal slurry concentration of the second time period corresponding to the characteristic variables of the second time period based on the target predictor, and obtain the predicted value of the coal slurry concentration of the second time period, wherein the characteristic variables of the second time period are used to characterize the characteristic variables of the coal slurry concentration of the second time period that do not need to be detected corresponding to the characteristic variables of the second time period, the characteristic variables of the second time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the second time period and the coal slurry concentration of the second time period that have been predicted by the target predictor are labeled data sets.
[0057] The obtaining unit is used to obtain a combined sample based on the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period.
[0058] The updating unit is used to input the combined sample into the target predictor to update the target predictor.
[0059] The third processing unit is used to collect the characteristic variables of the third time period in real time, and predict the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period based on the updated target predictor to obtain the predicted value of the coal slurry concentration of the third time period, wherein the characteristic variables of the third time period are used to characterize the characteristic variables of the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period that do not need to be detected, the characteristic variables of the third time period that have not been predicted by the target predictor are an unlabeled data set, and the characteristic variables of the third time period and the coal slurry concentration of the third time period that have been predicted by the target predictor are a labeled data set.
[0060] The fourth processing unit is used to repeatedly collect feature vectors in different time periods, iteratively update the target predictor, and predict the real-time coal slurry concentration.
[0061] Optionally, the training unit further includes: a training module, used to train the characteristic variables of the first period and the coal slurry concentration of the first period by using a conformal prediction algorithm of quantile regression to obtain a target predictor.
[0062] Optionally, the obtaining unit further includes: a merging module for merging the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period to obtain a combined sample.
[0063] In this embodiment, a first processing unit is used to collect a first time period characteristic variable of a target time interval, and based on the first time period characteristic variable, a first time period coal slurry concentration corresponding to the first time period characteristic variable is detected; a training unit is used to train the first time period characteristic variable and the first time period coal slurry concentration through a target algorithm to obtain a target predictor, wherein the first time period characteristic variable and the first time period coal slurry concentration are labeled data sets; a second processing unit is used to collect a second time period characteristic variable in real time, and based on the target predictor, a prediction is made of the second time period coal slurry concentration corresponding to the second time period characteristic variable to obtain a predicted value of the second time period coal slurry concentration, wherein the second time period characteristic variable is used to characterize a characteristic variable of the second time period coal slurry concentration corresponding to the second time period characteristic variable that does not need to be detected, the second time period characteristic variable that has not been predicted by the target predictor is an unlabeled data set, and the second time period characteristic variable and the second time period coal slurry concentration that have been predicted by the target predictor are labeled data sets; an obtaining unit is used to obtain a combined sample based on the first time period characteristic variable, the second time period characteristic variable, the first time period coal slurry concentration, and the second time period coal slurry concentration; an updating unit is used to update the combined sample. The combined samples are input into the target predictor to update the target predictor; the third processing unit collects the characteristic variables of the third time period in real time, and predicts the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period based on the updated target predictor to obtain the predicted value of the coal slurry concentration of the third time period, wherein the characteristic variables of the third time period are used to characterize the characteristic variables of the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period that do not need to be detected, the characteristic variables of the third time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the third time period and the coal slurry concentration of the third time period after being predicted by the target predictor are labeled data sets; the fourth processing unit repeatedly collects characteristic vectors of different time periods, iteratively updates the target predictor, and predicts the real-time coal slurry concentration, which solves the technical problems that the labeled data containing coal slurry concentration is very scarce, it is difficult to train a predictor with accurate prediction ability, and it is impossible to realize the real-time detection of coal slurry concentration, and achieves the technical effect of continuously updating the iterative predictor through the rolling conformal prediction method, using data fusion and learn-and-use methods, improving the prediction ability of the predictor, and realizing the real-time prediction of coal slurry concentration.
[0064] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.
[0065] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0066] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.
[0067] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0068] In addition, each functional unit in each embodiment of the present invention may be integrated into a first processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0069] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A real-time prediction method for coal slurry concentration, characterized in that: include: Collecting a first period characteristic variable of a target time interval, and based on the first period characteristic variable, detecting a first period coal slurry concentration corresponding to the first period characteristic variable; The characteristic variables of the first time period and the coal slurry concentration of the first time period are trained by a target algorithm to obtain a target predictor, wherein the characteristic variables of the first time period and the coal slurry concentration of the first time period are labeled data sets; The characteristic variables of the second period are collected in real time, and the coal slurry concentration of the second period corresponding to the characteristic variables of the second period is predicted based on the target predictor to obtain the predicted value of the coal slurry concentration of the second period, wherein the characteristic variables of the second period are used to characterize the characteristic variables of the coal slurry concentration of the second period that do not need to be detected, and the characteristic variables of the second period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the second period and the coal slurry concentration of the second period that have been predicted by the target predictor are labeled data sets; Obtain a combined sample based on the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period; Inputting the combined sample into the target predictor to update the target predictor; The characteristic variables of the third period are collected in real time, and based on the updated target predictor, the coal slurry concentration of the third period corresponding to the characteristic variables of the third period is predicted to obtain the predicted value of the coal slurry concentration of the third period, wherein the characteristic variables of the third period are used to characterize the characteristic variables of the coal slurry concentration of the third period that do not need to be detected, and the characteristic variables of the third period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the third period and the coal slurry concentration of the third period that have been predicted by the target predictor are labeled data sets; The characteristic vectors of different time periods are repeatedly collected, and the target predictor is iteratively updated to predict the real-time coal slurry concentration.
2. The method according to claim 1, characterized in that The target algorithm is used to train the characteristic variables of the first time period and the coal slurry concentration of the first time period to obtain a target predictor, including: The characteristic variables of the first period and the coal slurry concentration of the first period are trained by a conformal prediction algorithm of quantile regression to obtain a target predictor.
3. The method according to claim 1, characterized in that The step of obtaining a combined sample based on the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period includes: The characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period and the coal slurry concentration of the second time period are combined to obtain the combined sample.
4. The method according to claim 1, characterized in that: The characteristic variables of the first time period, the second time period and the third time period are the same variables. The characteristic variables of the first time period are the additive flow, the current of the main motor of the mill, the agitator current of the mill outlet tank, the liquid level of the mill outlet tank, the liquid level of the mill outlet tank, the speed and the outlet pressure of the low-pressure coal slurry pump.
5. A real-time prediction system for coal slurry concentration, characterized in that: include: A first processing unit is used to collect a first period characteristic variable of a target time interval, and based on the first period characteristic variable, detect a first period coal slurry concentration corresponding to the first period characteristic variable; A training unit, configured to train the characteristic variables of the first time period and the coal slurry concentration of the first time period by a target algorithm to obtain a target predictor, wherein the characteristic variables of the first time period and the coal slurry concentration of the first time period are labeled data sets; A second processing unit is used to collect the characteristic variables of the second time period in real time, predict the coal slurry concentration of the second time period corresponding to the characteristic variables of the second time period based on the target predictor, and obtain a predicted value of the coal slurry concentration of the second time period, wherein the characteristic variables of the second time period are used to characterize the characteristic variables of the coal slurry concentration of the second time period that do not need to be detected corresponding to the characteristic variables of the second time period, the characteristic variables of the second time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the second time period and the coal slurry concentration of the second time period that have been predicted by the target predictor are labeled data sets; An obtaining unit, configured to obtain a combined sample based on the characteristic variable of the first time period, the characteristic variable of the second time period, the coal slurry concentration of the first time period, and the coal slurry concentration of the second time period; an updating unit, configured to input the combined sample into the target predictor to update the target predictor; A third processing unit is used to collect the characteristic variables of the third time period in real time, and predict the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period based on the updated target predictor to obtain the predicted value of the coal slurry concentration of the third time period, wherein the characteristic variables of the third time period are used to characterize the characteristic variables of the coal slurry concentration of the third time period corresponding to the characteristic variables of the third time period that do not need to be detected, the characteristic variables of the third time period that have not been predicted by the target predictor are unlabeled data sets, and the characteristic variables of the third time period and the coal slurry concentration of the third time period that have been predicted by the target predictor are labeled data sets; The fourth processing unit is used to repeatedly collect feature vectors in different time periods, iteratively update the target predictor, and predict the real-time coal slurry concentration.
6. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 4 when running.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 4.
