Industrial production path optimization system based on reinforcement learning

Abstract

The application discloses an industrial production path optimization system based on reinforcement learning and relates to the technical field of production path optimization. The method comprises the following steps: a preliminary production path module, obtaining blast furnace operation state monitoring data and production electricity price data, identifying the overlapping part of the high energy consumption period and the high electricity price period, and adjusting the production plan; a heating work path module, obtaining the characteristics of the processing raw materials, setting the heating work parameters to obtain an optimized heating work path; a cooling parameter adjustment module, obtaining the internal state data of the blast furnace and the cold water state data, and adjusting the cooling work parameters; a cooling work path module, dynamically adjusting the cooling work intensity to obtain an optimized cooling work path; and an optimized production path module, integrating the two optimized work paths to obtain an optimized production path. The application reduces the energy consumption in the production process, helps to improve the quality of the products, and meets the higher demands of customers.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of production path optimization, and in particular to an industrial production path optimization system based on reinforcement learning. BACKGROUND

[0002] The industrial production path refers to the combination of a series of production activities and technical means from raw materials to finished products in the industrial production process. Specifically, it mainly includes research and design, raw material preparation, production process control, and quality control and inspection. These links jointly constitute the path of industrial production. The production process control link is to monitor and adjust the parameters of each link, use automatic production lines, and use computer control systems and other technical means for real-time monitoring and control to achieve efficient and accurate production and ensure the stability and consistency of product quality.

[0003] In the related art, for the automatic control of the industrial production process, parameters are usually controlled or adjusted based on static rules or fixed models. This method not only relies heavily on human experience and has low production efficiency, but also cannot cope with complex and variable industrial production environments, cannot guarantee the stability and sustainability of industrial production, and needs to be improved. SUMMARY

[0004] The purpose of the present application is to provide an industrial production path optimization system based on reinforcement learning to solve the problems raised in the background.

[0005] The industrial production path optimization system based on reinforcement learning provided by the present application adopts the following technical solution:

[0006] The preliminary production path module acquires blast furnace operation state monitoring data and production electricity price data, predicts the production work demand of the blast furnace based on the operation state monitoring data, matches the production work demand with the production electricity price data, identifies the overlapping part of the high energy consumption period and the high electricity price period, adjusts the production plan according to the overlapping part, and obtains a preliminary production path.

[0007] The heating work path module is signal connected with the preliminary production path module, divides the production work demand of the blast furnace into heating work demand and cooling work demand, acquires the processing raw material characteristics according to the heating work demand, divides the heating work demand into stages, obtains the heating stage of the blast furnace, sets the heating work parameters based on the heating stage and the processing raw material characteristics, and obtains an optimized heating work path.

[0008] The cooling parameter adjustment module is signal connected with the preliminary production path module, acquires blast furnace internal state data and cold water state data, constructs a cooling demand prediction model based on the internal state data and the cooling water state data, and adjusts the cooling work parameters.

[0009] The cooling work path module is connected with the cooling parameter adjustment module, and is used for dividing the cooling work demand into stages, dynamically adjusting the cooling work intensity, and obtaining an optimized cooling work path.

[0010] The optimized production path module is connected with the heating work path module and the cooling work path module, and is used for integrating the optimized heating work path and the optimized cooling work path to obtain an optimized production path for industrial production.

[0011] Preferably, the preliminary production path module is used for obtaining blast furnace running state monitoring data and production electricity price data, predicting the production work demand of the blast furnace based on the running state monitoring data, matching the production work demand with the production electricity price data, identifying an overlapping part of a high energy consumption period and a high electricity price period, and adjusting a production plan according to the overlapping part to obtain a preliminary production path.

[0012] The blast furnace running state monitoring data is obtained, and a production demand prediction model is constructed by using reinforcement learning based on the running state monitoring data, so as to predict the production work demand of the blast furnace and obtain production work demand data.

[0013] A standard energy consumption threshold is set, and a period higher than the standard energy consumption threshold is marked as a high energy consumption period according to the production work demand data.

[0014] The production electricity price data is obtained, the production work demand data is matched with the production electricity price data according to a time sequence, a standard electricity price threshold is set, and a period higher than the standard electricity price threshold is marked as a high electricity price period according to the production electricity price data.

[0015] The high energy consumption period and the high electricity price period are compared, and an overlapping part of the high energy consumption period and the high electricity price period is identified.

[0016] The overlapping part is quantified, the overlapping degree is evaluated according to the quantification result of the overlapping part, and a production plan is adjusted based on the overlapping degree to obtain a preliminary production path.

[0017] Preferably, the step of quantifying the overlapping part, evaluating the overlapping degree according to the quantification result of the overlapping part, and adjusting the production plan based on the overlapping degree to obtain the preliminary production path is specifically as follows:

[0018] The overlapping part is quantified, the percentage of the energy consumption amount of the overlapping part in the total energy consumption amount is calculated to obtain an overlapping energy consumption percentage, and the percentage of the electricity price cost of the overlapping part in the total electricity price cost is calculated to obtain an overlapping cost percentage.

[0019] The overlapping degree level is preliminarily determined based on the overlapping energy consumption percentage to obtain a preliminary overlapping degree.

[0020] setting an initial overlap cost ratio threshold, if the overlap cost ratio is less than the initial overlap cost ratio threshold or the overlap cost ratio is equal to the initial overlap cost ratio threshold, marking the preliminary overlap degree as the first overlap degree, and no adjustment is needed for the production work in the period of the first overlap degree;

[0021] if the overlap cost ratio is greater than the initial overlap cost ratio threshold, upgrading the preliminary overlap degree to obtain the second overlap degree, and adjusting the production work in the period of the second overlap degree;

[0022] determining a low electricity price period according to production electricity price data, adjusting the production work in the period of the second overlap degree to the low electricity price period to obtain a preliminary production path.

[0023] Preferably, the heating work path module is in signal connection with the preliminary production path module, divides the production work demand of the blast furnace into heating work demand and cooling work demand, acquires processing raw material characteristics according to the heating work demand, divides the heating work demand into stages to obtain a heating stage of the blast furnace, sets heating work parameters based on the heating stage and the processing raw material characteristics to obtain the steps of optimizing the heating work path, specifically:

[0024] defining heating curve parameters including temperature range, temperature rising rate and duration according to the heating work demand, and constructing a heating curve model according to the temperature rising rate, the temperature range and the duration;

[0025] acquiring processing raw material characteristics, the processing raw material characteristics including thermal conductivity, water content and impurity content;

[0026] dividing the heating work demand into stages to obtain a heating stage of the processing raw material, the heating stage including a preheating stage, a main heating stage and a holding stage;

[0027] when the processing raw material is in the preheating stage, comparing and judging the thermal conductivity of the processing raw material with a preset thermal conductivity, if the thermal conductivity of the processing raw material is lower than the preset thermal conductivity, increasing the temperature rising rate to obtain a standard temperature rising rate;

[0028] when the processing raw material is in the main heating stage, comparing and judging the impurity content of the processing raw material with a preset impurity content, if the impurity content of the processing raw material is higher than the preset impurity content and the reaction speed of the processing raw material is reduced, increasing the temperature range to obtain a standard temperature range;

[0029] when the processing raw material is in the holding stage, comparing and judging the water content of the processing raw material with a preset water content, if the water content of the processing raw material is higher than the preset water content, prolonging the holding time to obtain a standard duration;

[0030] According to the standard heating rate, the standard temperature range, the standard duration, the parameter of the heating curve model is adjusted to obtain a standard heating curve model;

[0031] Based on the standard heating curve model, a dynamic adjustment rule is formulated, and the heating working parameters are set according to the dynamic adjustment rule to obtain an optimized heating working path.

[0032] Preferably, based on the standard heating curve model, a dynamic adjustment rule is formulated, and the heating working parameters are set according to the dynamic adjustment rule to obtain an optimized heating working path, and the step is specifically:

[0033] The external environment information of the blast furnace is obtained, and the dynamic adjustment rule of the heating curve model is formulated according to the external environment information;

[0034] The external environment information includes environment temperature data and environment humidity data, when the environment temperature data decreases by one set environment temperature interval, the preheating temperature range is increased by one set preheating temperature interval, and a first dynamic adjustment signal is generated;

[0035] When the environment humidity data increases by one set environment humidity interval, the preheating duration is extended by one set duration interval, and a second dynamic adjustment signal is generated;

[0036] The first dynamic adjustment signal and the second dynamic adjustment information are integrated to generate a dynamic adjustment rule, the dynamic adjustment rule is input into the standard heating curve model, the heating working parameters are set, and an optimized heating working path is obtained.

[0037] Preferably, the cooling parameter adjustment module is connected with the initial production path module signal, the internal state data of the blast furnace and the cooling water state data are obtained, the cooling demand prediction model is constructed based on the internal state data and the cooling water state data, and the step of adjusting the cooling working parameters is specifically:

[0038] The internal state data of the blast furnace is obtained, and the internal temperature change rate in a set time period is obtained;

[0039] The cooling water state data is obtained, and the cooling water state data includes the flow rate change rate and the inlet and outlet temperature difference in a set time period;

[0040] Based on the internal temperature change rate, the flow rate change rate and the inlet and outlet temperature difference, a cooling demand prediction model is constructed, and the predicted cooling working demand in the future N hours is obtained according to the cooling demand prediction model;

[0041] According to the predicted cooling working demand, the cooling working parameters are adjusted.

[0042] Preferably, the step of adjusting the cooling work parameters according to the predicted cooling work requirement comprises:

[0043] According to the predicted cooling work requirement, a standard internal temperature range threshold is generated, real-time internal temperature data of the blast furnace is obtained, and the real-time internal temperature data is compared with the standard internal temperature range threshold.

[0044] When the real-time internal temperature data is lower than the standard internal temperature range threshold, the cooling water flow rate is reduced and the cooling water temperature is increased, while ensuring that the cooling water flow rate is not lower than a preset lower limit and the cooling water temperature is not higher than a preset upper limit.

[0045] When the real-time internal temperature data belongs to the standard internal temperature range threshold, the cooling water flow rate and the cooling water temperature are maintained unchanged.

[0046] When the real-time internal temperature data is higher than the standard internal temperature range threshold, the cooling water flow rate is increased and the cooling water temperature is reduced, while ensuring that the cooling water flow rate is not higher than a preset upper limit and the cooling water temperature is not lower than a preset lower limit.

[0047] Preferably, the cooling work path module is connected with the cooling parameter adjustment module, the cooling work requirement is divided into stages, the cooling work intensity is dynamically adjusted, and the optimized cooling work path is obtained, and the step comprises:

[0048] Real-time internal temperature change data of the blast furnace in a preset time period is obtained, and a real-time cooling rate of the cooling work is calculated based on the real-time internal temperature change data.

[0049] The cooling work requirement is divided into stages, the cooling stage in which the blast furnace is located is obtained, a target cooling rate is set for each cooling stage, and a cooling rate target range of each cooling stage is set according to the target cooling rate.

[0050] The real-time cooling rate is compared with the target cooling rate.

[0051] If the real-time cooling rate is lower than the target cooling rate, a deviation between the real-time cooling rate and the cooling rate target range is calculated to obtain first cooling rate deviation data, the cooling water flow rate is increased or the cooling water temperature is reduced according to the first cooling rate deviation data in proportion, while ensuring that the cooling water flow rate is not higher than a preset upper limit and the cooling water temperature is not lower than a preset lower limit, and a first standard cooling rate adjustment signal is generated.

[0052] If the real-time cooling rate is higher than the target cooling rate, a deviation of the real-time cooling rate from the target cooling rate range is calculated to obtain second cooling rate deviation data, and the cooling water flow rate is proportionally reduced or the cooling water temperature is increased according to the second cooling rate deviation data, while ensuring that the cooling water flow rate is not lower than a preset lower limit and the cooling water temperature is not higher than a preset upper limit, to generate a second standard cooling rate adjustment signal;

[0053] The first standard cooling rate adjustment signal and the second standard cooling rate adjustment signal are integrated to obtain a cooling rate adjustment rule, and the cooling work intensity is dynamically adjusted according to the cooling rate adjustment rule to obtain an optimized cooling work path.

[0054] In summary, the present application includes at least one of the following beneficial technical effects:

[0055] 1. By comparing production demand and electricity price data, the overlapping part of high energy consumption period and high electricity price period can be identified, and the production plan can be flexibly adjusted according to the identification result of the overlapping part to reduce production cost and ensure the maximization of production benefit. According to the heating demand and the characteristics of raw materials, the heating process is divided into stages, and appropriate heating parameters are set for each stage to ensure the uniformity and efficiency of the heating process. By optimizing the heating parameters and stage division, a more efficient and energy-saving heating work path can be obtained. The prediction model can predict future cooling demand, and the flow, temperature and other parameters of the cooling water can be dynamically adjusted according to the prediction result to ensure that the temperature inside the blast furnace remains within an appropriate range. The cooling work demand is divided into stages, and the cooling work intensity is dynamically adjusted according to the change of cooling demand to ensure the stability and efficiency of the cooling process. The optimized paths of heating and cooling processes are integrated to form a complete and efficient production optimization system, which can reduce energy consumption and waste in the production process, improve production efficiency and reduce production cost, and at the same time help to improve product quality, meet customers' higher demands and improve product competitiveness.

[0056] 2. By real-time monitoring and adjusting the internal temperature of the blast furnace, temperature abnormalities can be discovered and corrected in time to prevent the blast furnace from malfunctioning due to overheating or overcooling, thereby improving the running stability of the blast furnace. Reasonable cooling parameter settings can help reduce thermal stress inside the blast furnace and prevent the furnace body from being damaged due to large temperature changes, thereby prolonging the service life of the blast furnace and helping to maintain the continuity and stability of the production process, thereby improving production efficiency and product quality. By precisely controlling the flow and temperature of the cooling water, unnecessary energy waste can be avoided, and the energy utilization efficiency of the cooling system can be improved.

[0057] 3. Calculate the cooling rate, adjust according to the target cooling rate, ensure the efficiency and stability of the cooling process. Through stage division and setting of target cooling rate, different cooling strategies can be formulated for different cooling stages, realizing the optimization of cooling work path and improving the cooling efficiency. By dynamically adjusting the cooling water flow and temperature, the cooling intensity can always be kept at the best state, avoiding the situation of excessive cooling or insufficient cooling. Reasonable cooling work path can effectively reduce the thermal stress inside the blast furnace, reduce the damage of furnace shell and refractory material, thereby prolonging the service life of the blast furnace; at the same time, it can improve the production efficiency of the blast furnace, and reduce the risk of safety accidents caused by improper cooling. BRIEF DESCRIPTION OF DRAWINGS

[0058] Fig. 1 is the module connection diagram of the industrial production path optimization system based on reinforcement learning of the present application.

[0059] Fig. 2 is the specific flow step diagram of the industrial production path optimization system based on reinforcement learning of the present application.

[0060] Marked with: 1, preliminary production path module. 2, heating work path module. 3, cooling parameter adjustment module. 4, cooling work path module. 5, optimized production path module. DETAILED DESCRIPTION

[0061] The present application will be further described in detail below in conjunction with the embodiments and Figs. 1-2 The embodiments of the present application are not limited to this.

[0062] The present application discloses an industrial production path optimization system based on reinforcement learning, specifically including the following steps:

[0063] The preliminary production path module 1 acquires blast furnace running state monitoring data and production electricity price data, predicts the production work demand of the blast furnace based on the running state monitoring data, matches the production work demand with the production electricity price data, identifies the overlapping part of the high energy consumption period and the high electricity price period, adjusts the production plan according to the overlapping part, and obtains the preliminary production path;

[0064] The heating work path module 2 is signal connected with the preliminary production path module 1, divides the production work demand of the blast furnace into heating work demand and cooling work demand, acquires the processing raw material characteristics according to the heating work demand, divides the heating work demand into stages, obtains the heating stage of the blast furnace, sets the heating work parameters based on the heating stage and the processing raw material characteristics, and obtains the optimized heating work path;

[0065] The cooling parameter adjustment module 3 is connected with the primary production path module 1, obtains the internal state data of the blast furnace and the cooling water state data, constructs a cooling demand prediction model based on the internal state data and the cooling water state data, and adjusts the cooling work parameters;

[0066] The cooling work path module 4 is connected with the cooling parameter adjustment module 3, divides the cooling work demand into stages, dynamically adjusts the cooling work intensity, and obtains an optimized cooling work path;

[0067] The optimized production path module 5 is connected with the heating work path module 2 and the cooling work path module 4, integrates the optimized heating work path and the optimized cooling work path, and obtains an optimized production path for industrial production.

[0068] In actual application, the blast furnace operation state monitoring data is obtained, which is the basis for optimizing production. Through real-time monitoring of the operation state of the blast furnace, the stability and safety of the production process can be ensured. Based on the monitoring data, the production demand of the blast furnace at different time periods can be predicted to provide a basis for the formulation of production plans. By comparing the production demand with the electricity price data, the overlapping part of the high energy consumption period and the high electricity price period can be identified, so as to avoid high energy consumption production in the high electricity price period and reduce production cost. According to the identification result of the overlapping part, the production plan is flexibly adjusted to ensure the maximization of production benefit. The heating work demand and the cooling work demand are divided, and the heating and cooling demands are clearly defined, so that the parameter setting can be more targeted. The processing raw material characteristics are obtained, which have an important influence on the heating process and are helpful to develop a more reasonable heating strategy. According to the heating demand and the raw material characteristics, the heating process is divided into stages, and appropriate heating parameters are set for each stage to ensure the uniformity and efficiency of the heating process. Through optimization of the heating parameters and stage division, a more efficient and energy-saving heating work path can be obtained. The internal state data of the blast furnace and the cooling water state data are obtained to understand the temperature change of the blast furnace and the circulation state of the cooling water. Based on the collected data, a prediction model is constructed to predict future cooling demand. According to the prediction result, the flow, temperature and other parameters of the cooling water are dynamically adjusted to ensure that the temperature inside the blast furnace remains within a suitable range. The cooling work demand is divided into stages to control the cooling process more finely. According to the change of the cooling demand, the cooling work intensity is dynamically adjusted to ensure the stability and efficiency of the cooling process. Through optimization of the cooling parameters and stage division, a more energy-saving and efficient cooling work path can be obtained. The optimized paths of the heating and cooling processes are integrated to obtain a more complete and efficient production path, which constitutes a complete and efficient production optimization system. This path can reduce energy consumption and waste in the production process, improve production efficiency and reduce production cost. Meanwhile, the stable heating and cooling processes help to improve the quality of products, meet the higher demands of customers, and enhance the competitiveness of products.

[0069] The preliminary production path module 1 acquires blast furnace operation state monitoring data and production electricity price data, predicts production work demand of the blast furnace based on the operation state monitoring data, matches the production work demand with the production electricity price data, identifies an overlapping part of a high energy consumption period and a high electricity price period, adjusts a production plan according to the overlapping part, and obtains a step of the preliminary production path, specifically as follows:

[0070] S11. Acquire blast furnace operation state monitoring data, use reinforcement learning to construct a production demand prediction model based on the operation state monitoring data, predict production work demand of the blast furnace, and obtain production work demand data;

[0071] S12. Set a standard energy consumption threshold, and mark a period higher than the standard energy consumption threshold as a high energy consumption period according to the production work demand data;

[0072] S13. Acquire production electricity price data, align and match the production work demand data with the production electricity price data according to time sequence, set a standard electricity price threshold, and mark a period higher than the standard electricity price threshold as a high electricity price period according to the production electricity price data;

[0073] S14. Compare the high energy consumption period and the high electricity price period, and identify an overlapping part of the high energy consumption period and the high electricity price period;

[0074] S15. Quantify the overlapping part, evaluate an overlapping degree according to the quantification result of the overlapping part, adjust a production plan based on the overlapping degree, and obtain a preliminary production path.

[0075] In actual application, a prediction model is constructed by reinforcement learning using blast furnace operation state monitoring data to accurately predict production work demand of the blast furnace. The prediction model can dynamically adjust prediction results according to real-time data to adapt to changes in the operation state of the blast furnace, and provide reliable data support for subsequent energy consumption management and production plan optimization. A standard energy consumption threshold is set to identify a high energy consumption period, a standard electricity price threshold is set to identify a high electricity price period, and an overlapping part of the high energy consumption period and the high electricity price period is identified by comparison, so as to clearly define double peaks of energy consumption and energy cost, thereby targetedly optimizing production plan and energy use, and avoiding high energy consumption operation in a high electricity price period. Energy consumption and electricity price cost of the overlapping part are quantified to evaluate an overlapping degree, a production plan is dynamically adjusted according to the overlapping degree, energy consumption and energy cost are reduced, production efficiency is improved, carbon emission is reduced, the operation path of the blast furnace is optimized, and economic benefits of blast furnace operation and support for green production are realized.

[0076] The overlapping part is quantified, the overlapping degree is evaluated according to the overlapping part quantification result, the production plan is adjusted based on the overlapping degree, and the step of obtaining the preliminary production path is specific to:

[0077] S151. The overlapping part is quantified, the percentage of the overlapping part energy consumption in the total energy consumption is calculated, the overlapping energy consumption ratio is obtained, the percentage of the overlapping part electricity price cost in the total electricity price cost is calculated, and the overlapping cost ratio is obtained;

[0078] S152. Based on the overlapping energy consumption ratio, the overlapping degree level is preliminarily determined, and the preliminary overlapping degree is obtained;

[0079] S153. Set the initial overlapping cost ratio threshold, if the overlapping cost ratio is less than the initial overlapping cost ratio threshold or the overlapping cost ratio is equal to the initial overlapping cost ratio threshold, mark the preliminary overlapping degree as the first overlapping degree, and no adjustment is needed for the production work in the first overlapping degree period;

[0080] S154. If the overlapping cost ratio is greater than the initial overlapping cost ratio threshold, upgrade the preliminary overlapping degree to obtain the second overlapping degree, and adjust the production work in the second overlapping degree period;

[0081] S155. According to the production electricity price data, determine the low electricity price period, adjust the production work in the second overlapping degree period to the low electricity price period, and obtain the preliminary production path.

[0082] In actual application, the total energy consumption of the overlapping period is divided by the total energy consumption of all periods to obtain the overlapping energy consumption ratio, and the electricity price cost of the overlapping period is divided by the total electricity price cost of all periods to obtain the overlapping cost ratio. Set the overlapping energy consumption ratio < 20% as the low overlapping degree, i.e. the preliminary overlapping degree is the low overlapping degree; 20% ≤ overlapping energy consumption ratio ≤ 50% is the medium overlapping degree, i.e. the preliminary overlapping degree is the medium overlapping degree; the overlapping energy consumption ratio > 50% is the high overlapping degree, i.e. the preliminary overlapping degree is the high overlapping degree. Set the initial overlapping cost ratio threshold to 30%, if the overlapping cost ratio ≤ 30%, the preliminary overlapping degree is the first overlapping degree, and no adjustment is needed for the production work; if the overlapping cost ratio > 30%, upgrade the preliminary overlapping degree, for example, the preliminary overlapping degree is the low overlapping degree, then upgrade to the medium overlapping degree; the preliminary overlapping degree is the medium overlapping degree, then upgrade to the high overlapping degree; the preliminary overlapping degree is the high overlapping degree, then upgrade to the super high overlapping degree; and thus the second overlapping degree is obtained. Adjusting the production work in the second overlapping degree period to the low electricity price period can reduce the electricity price cost, improve the overall economic benefit, and realize the sustainable development of industrial production work.

[0083] The heating work path module 2 is in signal connection with the preliminary production path module 1, divides the production work demand of the blast furnace into heating work demand and cooling work demand, acquires the processing raw material characteristics according to the heating work demand, performs stage division on the heating work demand, obtains the heating stage in which the blast furnace is located, sets the heating work parameters based on the heating stage and the processing raw material characteristics, and obtains the steps of optimizing the heating work path, specifically:

[0084] S21. According to the heating work demand, define the heating curve parameters, including temperature range, heating rate, and duration, and construct a heating curve model according to the heating rate, temperature range, and duration.

[0085] S22. Acquire the processing raw material characteristics, including thermal conductivity, water content, and impurity content.

[0086] S23. Perform stage division on the heating work demand to obtain the heating stage in which the processing raw material is located, including the preheating stage, the main heating stage, and the holding stage.

[0087] S24. When the processing raw material is in the preheating stage, compare the thermal conductivity of the processing raw material with the preset thermal conductivity, if the thermal conductivity of the processing raw material is lower than the preset thermal conductivity, increase the heating rate to obtain the standard heating rate.

[0088] S25. When the processing raw material is in the main heating stage, compare the impurity content of the processing raw material with the preset impurity content, if the impurity content of the processing raw material is higher than the preset impurity content, the reaction speed of the processing raw material is reduced, then increase the temperature range to obtain the standard temperature range.

[0089] S26. When the processing raw material is in the holding stage, compare the water content of the processing raw material with the preset water content, if the water content of the processing raw material is higher than the preset water content, then extend the holding time to obtain the standard duration.

[0090] S27. According to the standard heating rate, standard temperature range, and standard duration, adjust the parameters of the heating curve model to obtain the standard heating curve model.

[0091] S28. Based on the standard heating curve model, develop a dynamic adjustment rule, set the heating work parameters according to the dynamic adjustment rule, and obtain the optimized heating work path.

[0092] In actual application, in the preheating stage, the raw material is gradually heated to prepare for the subsequent main heating stage, if the heat conductivity of the raw material is low, the heating rate is increased to speed up the preheating process; in the main heating stage, if the impurity content of the raw material is high, the reaction speed may be reduced, so the temperature range needs to be increased to promote the reaction; in the holding stage, if the water content of the raw material is high, the holding time may need to be extended to ensure that the moisture is fully evaporated or reacted. Taking a certain metal material heated in a high-temperature muffle furnace as an example, the temperature range of the heating curve is from room temperature to 1000℃, the initial heating rate is set to 10℃ / min, the preheating stage lasts for 1h, the main heating stage lasts for 2h, and the holding stage lasts for 1h. The heat conductivity of the metal material is medium, the water content is extremely low and can be ignored, and the impurity content is slightly higher than the standard value. The heating work requirements are divided into stages: the preheating stage gradually heats the metal material to 400℃, the main heating stage heats from 400℃ to 1000℃, and the holding stage keeps at 1000℃ for 1h. For the preheating stage: since the heat conductivity is medium, the heating rate remains unchanged; for the main heating stage: since the impurity content is slightly higher than the standard value, the heating rate is increased to 12℃ / min to promote the chemical reaction inside the metal; for the holding stage: since the water content is extremely low, the holding time remains unchanged. According to the above adjustment, the heating rate is modified to 12℃ / min (main heating stage), and other parameters remain unchanged, to obtain a new heating curve model, i.e. a standard heating curve model. According to the standard heating curve model, a series of dynamic adjustment rules are developed to deal with various situations that may occur during the heating process, thereby improving the heating efficiency and product quality.

[0093] Based on the standard heating curve model, a dynamic adjustment rule is developed, and the heating work parameters are set according to the dynamic adjustment rule to obtain an optimized heating work path, specifically:

[0094] S281. Obtain external environment information of the blast furnace, and develop a dynamic adjustment rule for the heating curve model according to the external environment information;

[0095] S282. The external environment information includes environmental temperature data and environmental humidity data, when the environmental temperature data decreases by one set environmental temperature interval, the preheating temperature range is increased by one set preheating temperature interval, and a first dynamic adjustment signal is generated;

[0096] S283. When the environmental humidity data increases by one set environmental humidity interval, the preheating duration is extended by one set duration interval, and a second dynamic adjustment signal is generated;

[0097] S284. Integrate the first dynamic adjustment signal and the second dynamic adjustment information to generate a dynamic adjustment rule, input the dynamic adjustment rule into the standard heating curve model, set the heating work parameters, and obtain an optimized heating work path.

[0098] In actual application, the decrease of the ambient temperature can result in the need for a higher temperature in the preheating stage to achieve the desired heating effect; and the increase of the ambient humidity can prolong the time required for preheating to ensure that the moisture inside the raw materials is fully evaporated or reaches the required thermal reaction condition. For example, a set ambient temperature interval is [0, 5], a set preheating temperature interval is [0, 10], that is, when the ambient temperature data decreases by 5℃, the preheating temperature range is increased by 10℃, and a first dynamic adjustment signal is generated; assuming that a set ambient humidity interval is [0, 0.1], a set duration interval is [0, 5], that is, when the ambient humidity data increases by 10%, the preheating duration is extended by 5 minutes, and a second dynamic adjustment signal is generated. The first dynamic adjustment signal and the second dynamic adjustment signal are integrated to obtain a dynamic adjustment rule. Through the dynamic adjustment rule, the heating curve model can be corrected in real time according to the changes of the external environment, so as to obtain an optimized heating working path that is more in line with the actual situation. The flexibility and adaptability of the heating process are improved, so that the heating process can better cope with changes in the external environment, and reduce the poor heating effect or product quality problems caused by environmental changes. At the same time, through the application of the dynamic adjustment rule, the heating efficiency and energy utilization rate can be further improved, and the production cost is reduced.

[0099] The cooling parameter adjustment module 3 is connected with the initial production path module 1, obtains the internal state data of the blast furnace and the cooling water state data, constructs a cooling demand prediction model based on the internal state data and the cooling water state data, and adjusts the cooling working parameters. The steps are as follows:

[0100] S31. Obtain the internal state data of the blast furnace, and the internal state data is the internal temperature change rate in a set time period;

[0101] S32. Obtain the cooling water state data, and the cooling water state data includes the flow rate change rate and the inlet and outlet temperature difference in a set time period;

[0102] S33. Construct a cooling demand prediction model based on the internal temperature change rate, the flow rate change rate, and the inlet and outlet temperature difference, and obtain the predicted cooling working demand in the next N hours according to the cooling demand prediction model;

[0103] S34. Adjust the cooling working parameters according to the predicted cooling working demand.

[0104] In actual application, the internal state data is the internal temperature change rate in the set time period, which can reflect the change of the thermal state of the blast furnace and is an important basis for predicting the cooling demand. By monitoring the temperature change rate, temperature abnormalities in the blast furnace can be found in time, so that cooling measures can be taken in advance to avoid overheating or damage of the blast furnace. The cooling water state data includes the flow rate change rate and the inlet and outlet temperature difference in the set time period, which reflects the circulation state and cooling effect of the cooling water. The flow of the cooling water in the blast furnace and the cooling capacity can be understood, so as to optimize the use efficiency of the cooling water. Based on the internal temperature change rate, the flow rate change rate and the inlet and outlet temperature difference, a cooling demand prediction model is constructed. Through the cooling demand prediction model, the cooling work demand in the next N hours can be predicted, the cooling demand of the blast furnace can be known in advance, and the cooling work parameters can be adjusted, so as to reasonably arrange the supply of cooling water and adjust the cooling work parameters, so as to ensure that the temperature in the blast furnace is maintained within a suitable range, thereby prolonging the service life of the blast furnace and improving the production efficiency.

[0105] According to the step of adjusting the cooling work parameters according to the predicted cooling work demand, specifically:

[0106] S341. According to the predicted cooling work demand, a standard internal temperature range threshold is generated, real-time internal temperature data of the blast furnace is obtained, and the real-time internal temperature data is compared with the standard internal temperature range threshold;

[0107] S342. When the real-time internal temperature data is lower than the standard internal temperature range threshold, the cooling water flow rate is reduced and the cooling water temperature is increased, while ensuring that the cooling water flow rate is not lower than the preset lower limit of the flow rate and the cooling water temperature is not higher than the preset upper limit of the temperature;

[0108] S343. When the real-time internal temperature data belongs to the standard internal temperature range threshold, the cooling water flow rate and the cooling water temperature are maintained unchanged;

[0109] S344. When the real-time internal temperature data is higher than the standard internal temperature range threshold, the cooling water flow rate is increased and the cooling water temperature is reduced, while ensuring that the cooling water flow rate is not higher than the preset upper limit of the flow rate and the cooling water temperature is not lower than the preset lower limit of the temperature.

[0110] In practical application, by monitoring and adjusting the internal temperature of the blast furnace in real time, temperature abnormalities can be discovered and corrected in a timely manner, preventing the blast furnace from malfunctioning due to overheating or overcooling, thereby improving the operational stability of the blast furnace. Reasonable cooling parameter settings help reduce internal thermal stress in the blast furnace, prevent the furnace body from being damaged due to excessive temperature changes, thereby extending the service life of the blast furnace, while helping to maintain the continuity and stability of the production process, thereby improving production efficiency and product quality. By precisely controlling the flow and temperature of cooling water, unnecessary energy waste can be avoided, and the energy utilization efficiency of the cooling system can be improved. Taking a blast furnace cooling system as an example, according to the operational experience and process requirements of the blast furnace, the standard internal temperature range threshold is set to 500℃±50℃. That is, the internal temperature of the blast furnace should be maintained between 450℃ and 550℃. The internal temperature of the blast furnace is monitored in real time by a temperature sensor inside the blast furnace, and the data is transmitted to the cooling system controller. The cooling system controller compares the real-time internal temperature with the set standard internal temperature range threshold. When the real-time internal temperature is below 450℃, the cooling system controller will reduce the cooling water flow and appropriately increase the cooling water temperature, while ensuring that the cooling water flow does not fall below the pre-set lower limit (such as 50m³ / h) and the cooling water temperature does not exceed the pre-set upper limit (such as 40℃). When the real-time internal temperature is between 450℃ and 550℃, the cooling system controller will maintain the current cooling water flow and temperature unchanged. When the real-time internal temperature is higher than 550℃, the cooling system controller will increase the cooling water flow and reduce the cooling water temperature, while ensuring that the cooling water flow does not exceed the pre-set upper limit (such as 100m³ / h) and the cooling water temperature does not fall below the pre-set lower limit (such as 20℃). The internal temperature of the blast furnace is effectively controlled, improving the operational stability and service life of the blast furnace. At the same time, due to the precise control of the cooling parameters, the energy utilization efficiency is also improved, and the production cost is reduced.

[0111] The cooling work path module 4 is in signal connection with the cooling parameter adjustment module 3, and the steps of dividing the cooling work demand into stages, dynamically adjusting the cooling work intensity, and obtaining an optimized cooling work path are as follows:

[0112] S41. Obtain real-time internal temperature change data of the blast furnace in a preset time period, and calculate a real-time cooling rate of the cooling work based on the real-time internal temperature change data;

[0113] S42. Divide the cooling work demand into stages to obtain the cooling stage in which the blast furnace is located, set a target cooling rate for each cooling stage, and set a cooling rate target range for each cooling stage according to the target cooling rate;

[0114] S43. Compare the real-time cooling rate with the target cooling rate;

[0115] S44. If the real-time cooling rate is lower than the target cooling rate, a deviation of the real-time cooling rate from the target cooling rate range is calculated to obtain first cooling rate deviation data, and the cooling water flow rate is proportionally increased or the cooling water temperature is proportionally decreased according to the first cooling rate deviation data, while ensuring that the cooling water flow rate does not exceed a preset upper flow limit and the cooling water temperature does not fall below a preset lower temperature limit, to generate a first standard cooling rate adjustment signal;

[0116] S45. If the real-time cooling rate is higher than the target cooling rate, a deviation of the real-time cooling rate from the target cooling rate range is calculated to obtain second cooling rate deviation data, and the cooling water flow rate is proportionally decreased or the cooling water temperature is proportionally increased according to the second cooling rate deviation data, while ensuring that the cooling water flow rate does not fall below a preset lower flow limit and the cooling water temperature does not rise above a preset upper temperature limit, to generate a second standard cooling rate adjustment signal;

[0117] S46. The first standard cooling rate adjustment signal and the second standard cooling rate adjustment signal are integrated to obtain a cooling rate adjustment rule, and the cooling work intensity is dynamically adjusted according to the cooling rate adjustment rule to obtain an optimized cooling work path.

[0118] In practical application, by real-time monitoring of the temperature change inside the blast furnace, the cooling rate can be accurately calculated, and adjustments can be made according to the target cooling rate to ensure the efficiency and stability of the cooling process. Through stage division and target cooling rate setting, different cooling strategies can be developed for different cooling stages to optimize the cooling work path and improve cooling efficiency. According to the comparison results of real-time cooling rate and target cooling rate, the cooling water flow and temperature are dynamically adjusted to ensure that the cooling intensity always remains at the optimal state, avoiding excessive cooling or insufficient cooling. A reasonable cooling work path can effectively reduce the thermal stress inside the blast furnace, reduce the damage to the furnace shell and refractory materials, and thus prolong the service life of the blast furnace; at the same time, it can improve the production efficiency of the blast furnace and reduce the risk of safety accidents caused by improper cooling. Taking the blast furnace body cooling system as an example, the system usually adopts a soft water closed loop circulation cooling method. In actual application, temperature sensors installed inside the blast furnace are used to obtain real-time temperature change data inside the blast furnace. Based on the real-time temperature change data, the real-time cooling rate of the cooling work is calculated. According to the cooling requirements of the blast furnace, the cooling process is divided into different stages, and target cooling rates and cooling rate target ranges are set for each stage. The real-time cooling rate is compared with the target cooling rate, and the cooling water flow and temperature are adjusted according to the comparison results. For example, if the real-time cooling rate is lower than the target cooling rate during the cooling process of a certain blast furnace, the cooling rate can be increased by increasing the cooling water flow; conversely, the cooling rate can be reduced by reducing the cooling water flow. The adjustment signals are integrated to form a cooling rate adjustment rule, and the cooling work intensity is dynamically adjusted according to the rule. By continuously optimizing the cooling work path, the stable operation and efficient production of the blast furnace are ensured, effectively improving the production efficiency and safety of the blast furnace, and prolonging the service life of the blast furnace.

[0119] The above are preferred embodiments of the present application, and do not limit the protection scope of the present application. Therefore, any equivalent changes made on the basis of the structure, shape, and principle of the present application should be covered within the protection scope of the present application.

Claims

1. A system for optimizing an industrial production path based on reinforcement learning, characterized in that, The method comprises the following steps: The preliminary production path module (1) obtains blast furnace operation state monitoring data and production electricity price data, predicts the production work demand of the blast furnace based on the operation state monitoring data, matches the production work demand with the production electricity price data, identifies the overlapping part of the high energy consumption period and the high electricity price period, adjusts the production plan according to the overlapping part, and obtains a preliminary production path; The heating work path module (2) is signal connected with the preliminary production path module (1), divides the production work demand of the blast furnace into heating work demand and cooling work demand, obtains the processing raw material characteristics according to the heating work demand, divides the heating work demand into stages, obtains the heating stage of the blast furnace, sets the heating work parameters based on the heating stage and the processing raw material characteristics, and obtains an optimized heating work path; The cooling parameter adjustment module (3) is signal connected with the preliminary production path module (1), obtains blast furnace internal state data and cold water state data, constructs a cooling demand prediction model based on the internal state data and the cooling water state data, and adjusts the cooling work parameters; The cooling work path module (4) is signal connected with the cooling parameter adjustment module (3), divides the cooling work demand into stages, dynamically adjusts the cooling work intensity, and obtains an optimized cooling work path; The optimized production path module (5) is signal connected with the heating work path module (2) and the cooling work path module (4), integrates the optimized heating work path and the optimized cooling work path, obtains an optimized production path, and is used for industrial production.

2. The reinforcement learning-based industrial production path optimization system of claim 1, wherein, The preliminary production path module (1) obtains blast furnace operation state monitoring data and production electricity price data, predicts the production work demand of the blast furnace based on the operation state monitoring data, matches the production work demand with the production electricity price data, identifies the overlapping part of the high energy consumption period and the high electricity price period, adjusts the production plan according to the overlapping part, and obtains a preliminary production path. The steps are as follows: Obtain blast furnace operation state monitoring data, use reinforcement learning to construct a production demand prediction model based on the operation state monitoring data, predict the production work demand of the blast furnace, and obtain production work demand data; Set a standard energy consumption threshold, and mark the period higher than the standard energy consumption threshold as a high energy consumption period according to the production work demand data; Obtain production electricity price data, align and match the production work demand data with the production electricity price data according to the time sequence, set a standard electricity price threshold, and mark the period higher than the standard electricity price threshold as a high electricity price period according to the production electricity price data; Compare the high energy consumption period and the high electricity price period, and identify the overlapping part of the high energy consumption period and the high electricity price period; Quantify the overlapping part, evaluate the overlapping degree according to the quantification result of the overlapping part, adjust the production plan based on the overlapping degree, and obtain a preliminary production path.

3. The reinforcement learning-based industrial production path optimization system of claim 2, wherein, The steps of quantifying the overlapping part, evaluating the overlapping degree according to the quantification result of the overlapping part, and obtaining a preliminary production path based on the overlapping degree are as follows: Quantifying the overlapping part, calculating the percentage of the energy consumption of the overlapping part in the total energy consumption, obtaining the overlapping energy consumption ratio, calculating the percentage of the electricity price cost of the overlapping part in the total electricity price cost, obtaining the overlapping cost ratio; Based on the overlapping energy consumption ratio, the preliminary overlapping degree is determined, and the preliminary overlapping degree is obtained; Set the initial overlapping cost ratio threshold, if the overlapping cost ratio is less than the initial overlapping cost ratio threshold or the overlapping cost ratio is equal to the initial overlapping cost ratio threshold, the preliminary overlapping degree is marked as the first overlapping degree, and the production work in the first overlapping degree period does not need to be adjusted; If the overlapping cost ratio is greater than the initial overlapping cost ratio threshold, the preliminary overlapping degree is upgraded to obtain the second overlapping degree, and the production work in the second overlapping degree period is adjusted in the period; According to the production electricity price data, the low electricity price period is determined, the production work in the second overlapping degree period is adjusted to the low electricity price period, and the preliminary production path is obtained.

4. The reinforcement learning-based industrial production path optimization system of claim 1, wherein, The heating work path module (2) is connected with the preliminary production path module (1) signal, the production work demand of the blast furnace is divided into heating work demand and cooling work demand, the processing raw material characteristics are obtained according to the heating work demand, the heating work demand is divided into stages, the heating stage of the blast furnace is obtained, the heating work parameters are set based on the heating stage and the processing raw material characteristics, and the steps of obtaining the optimized heating work path are as follows: According to the heating work demand, define the heating curve parameters, including temperature range, heating rate, and duration, and construct the heating curve model according to the heating rate, temperature range and duration; Obtain the processing raw material characteristics, including thermal conductivity, water content and impurity content; The heating work demand is divided into stages, and the heating stage of the processing raw material is obtained, including the preheating stage, the main heating stage and the holding stage; When the processing raw material is in the preheating stage, compare the thermal conductivity of the processing raw material with the preset thermal conductivity, if the thermal conductivity of the processing raw material is lower than the preset thermal conductivity, the heating rate is increased to obtain the standard heating rate; When the processing raw material is in the main heating stage, compare the impurity content of the processing raw material with the preset impurity content, if the impurity content of the processing raw material is higher than the preset impurity content, the reaction speed of the processing raw material is reduced, and the temperature range is increased to obtain the standard temperature range; When the processing raw material is in the holding stage, compare the water content of the processing raw material with the preset water content, if the water content of the processing raw material is higher than the preset water content, the holding time is prolonged to obtain the standard duration; According to the standard heating rate, the standard temperature range and the standard duration, the parameters of the heating curve model are adjusted to obtain the standard heating curve model; Based on the standard heating curve model, the dynamic adjustment rule is formulated, and the heating work parameters are set according to the dynamic adjustment rule to obtain the optimized heating work path.

5. The reinforcement learning-based industrial production path optimization system of claim 4, wherein, Based on the standard heating curve model, the dynamic adjustment rule is formulated, and the heating work parameters are set according to the dynamic adjustment rule to obtain the optimized heating work path. Acquire external environment information of the blast furnace, and formulate a dynamic adjustment rule of a heating curve model according to the external environment information; The external environment information includes environment temperature data and environment humidity data, when the environment temperature data decreases by a set environment temperature interval, the preheating temperature range is increased by a set preheating temperature interval, and a first dynamic adjustment signal is generated; When the environment humidity data increases by a set environment humidity interval, the preheating duration is extended by a set duration interval, and a second dynamic adjustment signal is generated; Integrate the first dynamic adjustment signal and the second dynamic adjustment information to generate a dynamic adjustment rule, input the dynamic adjustment rule into a standard heating curve model, set heating working parameters, and obtain an optimized heating working path.

6. The reinforcement learning-based industrial production path optimization system of claim 1, wherein, The cooling parameter adjustment module (3) is connected with the preliminary production path module (1) and acquires internal state data of the blast furnace and cooling water state data, constructs a cooling demand prediction model based on the internal state data and the cooling water state data, and adjusts the cooling working parameters, specifically as follows: Acquire internal state data, and the internal state data is an internal temperature change rate in a set time period; Acquire cooling water state data, and the cooling water state data includes a flow rate change rate and an inlet and outlet temperature difference in a set time period; Construct a cooling demand prediction model based on the internal temperature change rate, the flow rate change rate and the inlet and outlet temperature difference, and obtain a predicted cooling working demand in the next N hours according to the cooling demand prediction model; Adjust the cooling working parameters according to the predicted cooling working demand.

7. The reinforcement learning-based industrial production path optimization system of claim 6, wherein, The step of adjusting the cooling working parameters according to the predicted cooling working demand is specifically as follows: Generate a standard internal temperature range threshold value according to the predicted cooling working demand, acquire real-time internal temperature data of the blast furnace, and compare the real-time internal temperature data with the standard internal temperature range threshold value; When the real-time internal temperature data is lower than the standard internal temperature range threshold value, reduce the cooling water flow rate and increase the cooling water temperature, while ensuring that the cooling water flow rate is not lower than a preset lower limit of the flow rate and the cooling water temperature is not higher than a preset upper limit of the temperature; When the real-time internal temperature data belongs to the standard internal temperature range threshold value, maintain the cooling water flow rate and the cooling water temperature unchanged; When the real-time internal temperature data is higher than the standard internal temperature range threshold value, increase the cooling water flow rate and decrease the cooling water temperature, while ensuring that the cooling water flow rate is not higher than a preset upper limit of the flow rate and the cooling water temperature is not lower than a preset lower limit of the temperature.

8. The reinforcement learning-based industrial production path optimization system of claim 7, wherein, The cooling working path module (4) is connected with the cooling parameter adjustment module (3) and divides the cooling working demand into stages, dynamically adjusts the cooling working intensity, and obtains an optimized cooling working path, specifically as follows: Acquire real-time internal temperature change data of the blast furnace in a preset time period, and calculate a real-time cooling rate of the cooling working based on the real-time internal temperature change data; Divide the cooling working demand into stages, obtain a cooling stage of the blast furnace, set a target cooling rate for each cooling stage, and set a cooling rate target range for each cooling stage according to the target cooling rate; comparing the real-time cooling rate with a target cooling rate; if the real-time cooling rate is lower than the target cooling rate, calculating a deviation of the real-time cooling rate from a target range of cooling rates to obtain first cooling rate deviation data, proportionally increasing a cooling water flow rate or decreasing a cooling water temperature according to the first cooling rate deviation data while ensuring that the cooling water flow rate does not exceed a preset upper limit of the flow rate and the cooling water temperature is not lower than a preset lower limit of the temperature, and generating a first standard cooling rate adjustment signal; if the real-time cooling rate is higher than the target cooling rate, calculating a deviation of the real-time cooling rate from the target range of cooling rates to obtain second cooling rate deviation data, proportionally decreasing the cooling water flow rate or increasing the cooling water temperature according to the second cooling rate deviation data while ensuring that the cooling water flow rate is not lower than a preset lower limit of the flow rate and the cooling water temperature is not higher than a preset upper limit of the temperature, and generating a second standard cooling rate adjustment signal; integrating the first standard cooling rate adjustment signal and the second standard cooling rate adjustment signal to obtain a cooling rate adjustment rule, and dynamically adjusting a cooling work intensity according to the cooling rate adjustment rule to obtain an optimized cooling work path.

Images