Decision control system and method based on glass production water supply

Abstract

The invention discloses a decision control system based on glass production water supply, and the system comprises an information collection module which is used for collecting the state information data of equipment; the information acquisition module is connected with the conversion module, and the conversion module is used for quantifying the received state information data according to a preset quantification standard to generate quantized data; the conversion module is connected with the analysis and decision module, the analysis and decision module can establish a learning algorithm model based on the quantitative data and the equipment starting condition, the starting rules of the equipment under different setting conditions are recognized through the learning algorithm model, and output data are generated. According to the decision-making control system and method based on glass production water supply, external state information of equipment can be collected, skilled operators are simulated, which method is adopted to adjust the state of each component is determined, unattended control is achieved, efficiency is improved, and meanwhile the production cost of a manufacturer is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of water supply control, and in particular to a decision-making control system and method based on water supply for glass production. Background Art

[0002] In the glass production process, the role of water is mainly reflected in the following aspects: first, cooling. The high-temperature glass liquid needs to be quickly cooled by water to maintain its shape and performance; second, cleaning. Water can wash away impurities and dust on the surface of the glass to improve product quality; in addition, water also participates in some chemical reactions to help form the specific structure and properties of the glass. For glass manufacturers, the rational use and management of water resources is not only related to the control of production costs, but also directly affects the quality and environmental performance of the products. Therefore, manufacturers need to pay close attention to the water supply of each equipment in the production process.

[0003] The water supply of various water pumps, cooling towers, emergency diesel pumps and other equipment is usually observed in real time by operators, who take different methods to adjust and control the status of various components in the equipment according to different situations. This method requires high proficiency and professionalism of operators, and has limited efficiency and is difficult to fully implement. Therefore, glass production water supply control system and manual auxiliary monitoring are used, but the combination of manual and water supply control system will increase the production cost of manufacturers. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned background technical problems and to propose a decision-making control system and method based on glass production water supply.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A decision control system based on glass production and water supply comprises: an information acquisition module, the information acquisition module is used to collect status information data of equipment; the information acquisition module is connected to a conversion module, the conversion module is used to quantify the received status information data according to a preset quantification standard to generate quantified data; the conversion module is connected to an analysis and decision module, the analysis and decision module can establish a learning algorithm model based on the quantified data and the equipment startup situation, and the learning algorithm model can be used to identify the startup rules of the equipment under different setting conditions and generate output data; the analysis and decision module is connected to a decompression module, the decompression module is used to quantify the output data and regress the output data into execution data that can be used to control the equipment; the decompression module is connected to a feedback module, the feedback module is connected to both the analysis and decision module and the conversion module, and the feedback module is used to compare the difference between the output data and the status information data, and respectively judge the accuracy of the execution data and the learning algorithm model.

[0007] As a further solution of the present invention: the status information data of the device includes independent variable status information and dependent variable status information, the independent variable status information includes: the operating status, frequency information, and controllable status of the control device, and the dependent variable status information includes: status information of pressure, liquid level, and temperature-related transmitter feedback.

[0008] A decision-making control method based on glass production water supply, comprising:

[0009] S1. Collect the status information data of the equipment and classify the status information into independent variable status information and dependent variable status information;

[0010] S2. Preset a quantization standard, first distinguish the state information data according to the quantization standard, and then quantize it into data that can be used by the algorithm, which is recorded as quantized data;

[0011] S3, establishing a learning algorithm model, identifying the startup rules of the device under different setting conditions through the learning algorithm model, and generating output data;

[0012] S4, quantifying the output data, so that the output data is regressed into execution data that can be used to control the device;

[0013] S5. Compare the difference between the output data and the status information data, determine the accuracy of the execution data and the learning algorithm model, and select data output or calibration model based on the determination result.

[0014] As a further solution of the present invention: in the step S2, the quantification standard includes: distinguishing the importance of the status information data by a weighting coefficient.

[0015] As a further solution of the present invention: in the step S2, the quantization standard includes: normalizing the state information data, and the normalized data range is [0, 1].

[0016] As a further solution of the present invention: in the step S3, when establishing the learning algorithm model, a mathematical model of the dependent variable information and the independent variable information is first established, and then a mathematical model of the dependent variable information, the independent variable information and the device startup state is established.

[0017] As a further solution of the present invention: in step S5, when judging whether the execution data is accurate, a difference prediction is performed based on the mathematical model of the dependent variable information and the independent variable information, and the predicted difference is recorded as E for , E for =Execution data-status information data;

[0018] When the independent variables do not change, and E for>T1, T1 is an empirical value, T1 is given to the transmitter;

[0019] When the independent variable is abnormal, the abnormal situation is fed back to step S2, the abnormal independent variable is closed, and the execution data is adopted.

[0020] As a further solution of the present invention: In step S5, when judging whether the learning algorithm model is accurate, if the independent variable changes normally and E for >T1, the learning algorithm model needs to be recalibrated.

[0021] Beneficial effects of the present invention:

[0022] In the present invention, a decision-making control system based on glass production water supply is combined with a method, which can collect external status information of the equipment, and by simulating skilled operators, decide what method to take to adjust the status of each component, thereby realizing unattended control, thereby improving efficiency and reducing the production cost of the manufacturer. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described below in conjunction with the accompanying drawings.

[0024] Figure 1 It is a structural schematic diagram of the decision-making control system based on glass production water supply of the present invention;

[0025] Figure 2 It is a flow chart of the decision-making control method based on glass production water supply of the present invention;

[0026] Figure 3 It is a block diagram of the learning algorithm model of the present invention.

[0027] In the figure:

[0028] 1. Input layer; 2. Hidden layer; 3. Output layer. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] See also Figure 1As shown, the present invention is a decision-making control system based on glass production water supply, including: an information acquisition module, the information acquisition module is used to collect equipment status information data, the equipment status information data includes independent variable status information and dependent variable status information, the independent variable status information includes: the operating status, frequency information, and controllable status of the control equipment, and the dependent variable status information includes: status information of pressure, liquid level, and temperature-related transmitter feedback.

[0031] The information acquisition module is connected to the conversion module, and the conversion module is used to quantify the received status information data according to a preset quantization standard to generate quantized data.

[0032] The conversion module is connected to the analysis and decision module, and the analysis and decision module can establish a learning algorithm model based on the quantitative data and the equipment startup status.

[0033] Specifically, the independent variable status information data of the equipment startup status and setting frequency can be collected, and the startup rules of the equipment under different setting conditions can be identified by using regression analysis or machine learning algorithms (such as decision trees and random forests), and output data can be generated. Not only can the operating frequency of the equipment be optimized, but the number of equipment startups under specific conditions can also be predicted, thereby improving the efficiency of the overall system.

[0034] The analysis and decision module is connected to the decompression module, and the decompression module is used to quantify the output data and regress the output data into execution data that can be used to control the device.

[0035] The decompression module is connected to the feedback module, and the feedback module is connected to the analysis and decision module and the conversion module. The feedback module is used to compare the differences between the output data and the state information data, and to judge the accuracy of the execution data and the learning algorithm model respectively.

[0036] See also Figure 1-Figure 2 As shown, a decision control method based on glass production water supply includes:

[0037] S1. Collect the status information data of each device through the information acquisition module, and classify the status information into independent variable status information and dependent variable status information.

[0038] The independent variable status information includes: the status information of controlling each water pump, cooling tower, and emergency diesel pump. The dependent variable status information includes: the status information of pressure, liquid level, and temperature-related transmitter feedback.

[0039] The independent variable status information is represented as X, and the dependent variable information is represented as Y.

[0040] In the independent variable status information, A 11 The running status of the 1# device, 0 means the device is not running, 1 means the device is running, and 2 means the device is faulty; A 12The frequency information fed back by device 1#. If the device does not have a frequency converter and is a direct start device, the data is 50.0; A 13 Is the controllable status of device 1#: 0 means the current device is uncontrollable, 1 means the current device is controllable, 2 means the current device is controllable but participates in PID control. 21 , A 22 , A 23 The status of the 2# device has the same meaning as the corresponding A 11 , A 12 , A 13 Similarly, the independent variable state information X can be expressed as:

[0041]

[0042] In the dependent variable status information, B1 is the information fed back by 1# transmitter, B2 is the information fed back by 2# transmitter, ... n is the information fed back by the n# transmitter. So the dependent variable state information Y can be expressed as:

[0043]

[0044] S2. Preset quantization standards. There are two quantization standards. One is to distinguish the importance of the state information data by weighting coefficients. The other is to normalize the state information data. The normalized data range is [0,1].

[0045] Since the importance of the state information data received by the conversion module is not consistent, it is necessary to distinguish the state information data according to the quantification standard. In addition, the state information data cannot be directly used in the analysis and decision-making module, and it needs to be quantified into data that can be used by the algorithm in the analysis and decision-making module. The quantified data is recorded as quantified data.

[0046] Specifically, when quantifying the independent variable state information data, the equipment operation state is divided into non-operation, operation, and failure. Among these three states, the failure state information is the most important, followed by operation, and finally non-operation. In order to distinguish the weights, it is necessary to introduce the weight coefficient λ, which is set to λ1=0 when not operating, λ1=0.3 when operating, and λ1=0.4 when failing; the frequency feedback is quantized to λ2=0.02; the controllable state is quantized to λ3=0.8, and X1 represents the independent variable quantization data.

[0047]

[0048] When the dependent variable state information data is quantified, the transmitter feedback state quantization coefficient η1, when the device is pressure data, η1 = 1, when the device is liquid level data, η1 = 0.1, when the device is temperature data, η1 = 0.01, after quantization, Y1 represents the dependent variable quantified data.

[0049]

[0050] S3. The analysis and decision-making module can establish a learning algorithm model, identify the startup rules of the equipment under different setting conditions through the learning algorithm model, and generate output data.

[0051] Specifically, quantitative data requires training mathematical models of dependent variables and independent variables through learning algorithms. By training dependent variables and equipment startup conditions through learning algorithms, the number of equipment startups and equipment setting frequency data under different setting conditions (operation, failure, frequency) can be established.

[0052] See also Figure 3 As shown, the general learning algorithm model of the learning algorithm is as follows:

[0053] Establish a learning algorithm model with n inputs and n outputs. By continuously learning the mathematical relationship between inputs and outputs, establish the mathematical model of its inputs and outputs. First, establish the mathematical model of the dependent variable information and the independent variable information, and then establish the mathematical model of the dependent variable information, the independent variable information and the equipment startup state. Thus, establish the decision.

[0054] Among them, the learning algorithm model is designed with input layer 1, hidden layer 2 and output layer 3. Input layer 1 is set with n neurons, hidden layer 2 is set with multiple layers, and each layer is set with multiple neurons. An activation function is selected, and output layer 3 is set with n neurons. When establishing a mathematical model of dependent variable information and independent variable information, it is necessary to train a multi-layer perceptron or regression model and optimize the loss function to capture the relationship between input and output. When establishing a mathematical model of dependent variable information, independent variable information and device startup status, the prediction of device status is added to output layer 3, and classification loss function is used for joint training. When making a decision, it is necessary to formulate a threshold or decision rule based on the model output to determine whether the device should be started or the status of each component should be adjusted.

[0055] S4. Since the output data generated by the analysis and decision module cannot be directly used in various devices for control, the output data needs to be quantized through the decompression module. Quantization is the process of decompressing the output data. The quantization method is the same as step S2, so that the output data can be regressed into execution data that can be used to control the device.

[0056] S5. Compare the difference between the output data and the status information data through the feedback module, judge the accuracy of the execution data and the learning algorithm model, and select data output or calibration model based on the judgment result.

[0057] When judging whether the execution data is accurate, the difference prediction is performed based on the mathematical model of the dependent variable information and the independent variable information. The predicted difference is recorded as Efor , E for =Execution data-status information data.

[0058] If the independent variables do not change, and E for >T1, T1 is an empirical value, T1 is given to the transmitter.

[0059] Among them, empirical data is not fixed data, which depends on the on-site conditions of the commissioning personnel and requires comprehensive consideration of on-site equipment and installation.

[0060] If the independent variable is abnormal, the abnormal situation is fed back to step S2, the abnormal independent variable is closed, and the execution data is adopted.

[0061] When judging whether the learning algorithm model is accurate, if the independent variable changes normally and E for >T1, it is judged that the learning algorithm model is not accurate enough. At this time, the feedback module feeds back the result to the analysis and decision module, and the analysis and decision module recalibrates the learning algorithm model. The model calibration method collects the most recent non-abnormal data for retraining, and then obtains a new learning algorithm model.

[0062] In this application, a decision-making control system and method based on glass production water supply are combined to collect external status information of the equipment, and by simulating skilled operators, decide what method to take to adjust the status of each component to achieve unmanned control, thereby improving efficiency and reducing the manufacturer's production costs.

[0063] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of application of the present invention should still fall within the scope of the claims of the present invention.

Claims

1. A decision-making control system based on glass production water supply, characterized in that: include: An information collection module, wherein the information collection module is used to collect status information data of the device; The information acquisition module is connected to the conversion module, and the conversion module is used to quantify the received status information data according to a preset quantization standard to generate quantized data; The conversion module is connected to the analysis and decision module, and the analysis and decision module can establish a learning algorithm model based on the quantitative data and the device startup situation, identify the startup rules of the device under different setting conditions through the learning algorithm model, and generate output data; The analysis and decision module is connected to a decompression module, and the decompression module is used to quantify the output data and return the output data to execution data that can be used to control the device; The decompression module is connected to the feedback module, and the feedback module is connected to the analysis and decision module and the conversion module. The feedback module is used to compare the difference between the output data and the state information data, and to judge the accuracy of the execution data and the learning algorithm model respectively.

2. A decision-making control system based on glass production water supply according to claim 1, characterized in that: The status information data of the device includes independent variable status information and dependent variable status information. The independent variable status information includes: operating status, frequency information, and controllable status of the control device; the dependent variable status information includes: status information of pressure, liquid level, and temperature-related transmitter feedback.

3. A method for using the decision control system based on glass production water supply according to any one of claims 1-2, characterized in that: include: S1. Collect the status information data of the equipment and classify the status information into independent variable status information and dependent variable status information; S2. Preset a quantization standard, first distinguish the state information data according to the quantization standard, and then quantize it into data that can be used by the algorithm, which is recorded as quantized data; S3, establishing a learning algorithm model, identifying the startup rules of the device under different setting conditions through the learning algorithm model, and generating output data; S4, quantifying the output data, so that the output data is regressed into execution data that can be used to control the device; S5. Compare the difference between the output data and the status information data, determine the accuracy of the execution data and the learning algorithm model, and select data output or calibration model based on the determination result.

4. The method according to claim 3, characterized in that: In the step S2, the quantification standard includes: distinguishing the importance of the status information data by weighting coefficients.

5. The method according to claim 4, characterized in that: In the step S2, the quantization standard includes: normalizing the state information data, and the normalized data range is [0, 1].

6. The method according to claim 3, characterized in that: In step S3, when establishing the learning algorithm model, a mathematical model of dependent variable information and independent variable information is first established, and then a mathematical model of dependent variable information, independent variable information and device startup status is established.

7. The method according to claim 6, characterized in that: In step S5, when judging whether the execution data is accurate, a difference prediction is performed based on the mathematical model of the dependent variable information and the independent variable information, and the predicted difference is recorded as E for , E for =Execution data-status information data; When the independent variables do not change, and E for >T1, T1 is an empirical value, T1 is given to the transmitter; When the independent variable is abnormal, the abnormal situation is fed back to step S2, the abnormal independent variable is closed, and the execution data is adopted.

8. The method according to claim 7, characterized in that: In step S5, when judging whether the learning algorithm model is accurate, if the independent variable changes normally and E for >T1, the learning algorithm model needs to be recalibrated.

Images