An Optimization Strategy for Enterprise Production Control and Management Considering Environmental Carrying Capacity

Abstract

The present invention discloses an optimization strategy for enterprise production control considering environmental carrying capacity, belonging to the technical field of enterprise production control. The dilution capacity of the water environment is evaluated through the MIKE11 water quality model, and the capacity of the atmospheric environment is calculated by combining the multi-source model method to ensure that enterprise production activities do not exceed the environmental carrying capacity. By analyzing the production requirements and resource usage of the enterprise, a Markov decision process model is established to dynamically adjust the production strategy, and a real-time monitoring and early warning mechanism is used to ensure that the environmental pollution concentration is within the safe state range. By adopting the above optimization strategy for enterprise production control considering environmental carrying capacity, the present invention can achieve the optimization of enterprise production control on the premise of meeting the environmental carrying capacity and can conduct real-time monitoring and early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise production control, and particularly to an optimized strategy for enterprise production control considering environmental carrying capacity. Background Art

[0002] The environmental carrying capacity assessment technology is a method used to evaluate the capacity and tolerance of a specific area or ecosystem to various resources (such as water, soil, air, etc.). Its purpose is to help decision-makers, policymakers, and enterprise managers understand and manage the potential impacts of human activities on the environment, so as to ensure the realization of the goals of sustainable development and environmental protection. Previously, the traditional production control technology in China focused on the preparation of production plans and the arrangement of scheduling, paid attention to the management and optimization of inventory, emphasized the improvement of production efficiency and cost control, and centered on process optimization, resource management, and quality control, aiming to improve production efficiency, reduce costs, and ensure product quality. However, with the increasing prominence of environmental problems and the growing importance of sustainable development, modern enterprises not only focus on economic benefits but also increasingly attach importance to environmental impacts and resource sustainability. Therefore, considering environmental carrying capacity has become one of the important factors that many enterprises and governments must consider when making decisions.

[0003] Real-time monitoring and big data analysis technologies play a crucial role in environmental carrying capacity assessment. Through a widely deployed sensor network, enterprises can monitor and record key environmental parameters in real time, such as air quality, water quality status, and soil pollution level. At the same time, these systems can also instantaneously collect data on production activities, including energy consumption, raw material usage, product productivity, and waste and chemical emissions. The large amount of real-time and historical data collected is stored, analyzed, and mined through advanced big data processing technologies to identify potential correlations and patterns between production activities and environmental impacts. Based on the results of these data analyses, enterprises can establish early warning systems to promptly detect and respond to situations that exceed environmental regulatory standards, such as excessive emissions or low resource utilization efficiency. In addition, by dynamically adjusting production plans and operation strategies, enterprises can maximize resource utilization efficiency and reduce negative impacts on the surrounding environment. The continuous data-driven decision support system helps enterprises manage and optimize their environmental management strategies to ensure the long-term goal of environmental sustainable development while continuously improving production efficiency. Enterprises face multiple challenges in implementing real-time sewage monitoring. At the technical level, it includes the selection of high-precision sensors and data quality assurance, as well as the complexity of big data processing and transmission; economically, they need to cope with the high costs of equipment and systems and the pressure of increased operating expenses; the complex environmental conditions and the need to analyze a large amount of real-time data at the operation level are also important problems that need to be overcome during the implementation process.

[0004] The application of the Intelligent Decision Support System (IDSS) in modern enterprise production control marks the deep integration and progress of technology and management concepts. Traditional production control methods often rely on static data and empirical judgments. Enterprises usually formulate production plans and environmental management strategies based on historical data and experience. However, this method is limited by the data collection cycle and the limitations of static environmental assessment reports. Dependent on periodic data reports and static environmental impact assessment reports, it fails to effectively capture the actual impacts of production activities under different conditions, making it difficult to cope with the dynamic changes and complexities of environmental impacts, and unable to respond in a timely manner to challenges brought about by emergencies or environmental changes. With the rapid development of information technology and data science, the intelligent decision support system not only provides real-time data collection and processing capabilities, but also enables enterprises to more comprehensively and scientifically analyze and predict the environmental impacts of their production activities through advanced algorithms and models, thereby better managing environmental risks and achieving the goal of sustainable development. Existing intelligent decision support systems face challenges such as low data quality and lack of integrity management in enterprise production control, and need to effectively integrate and clean multi-source data, and optimize algorithms and models to cope with complex production environments and high-demand real-time decision-making requirements. In terms of environmental carrying capacity assessment, the intelligent decision support system plays an important role. The system can not only consider single environmental factors such as air quality, water quality, and soil health, but also comprehensively consider the combined impacts of multiple factors on the environment. Through comprehensive data analysis and model establishment, the system can evaluate the load of production activities on the overall environmental carrying capacity. The intelligent decision support system has real-time data monitoring and analysis capabilities, and can promptly detect and respond to changes in environmental impacts that may be brought about by production activities, thereby reducing potential environmental risks and problems. Through data analysis and model prediction technologies, the system can predict production scenarios that may lead to excessive environmental loads or environmental problems, enabling enterprises to take preventive measures before problems occur and protecting the health of the ecosystem. The introduction of the intelligent decision support system marks a major innovation in enterprise production control technology and is a hot topic among domestic universities and research institutions.

[0005] Existing enterprise production control systems are outstanding in improving efficiency and economic benefits, but often ignore the impacts of environmental sustainability and dynamic change factors. This has led to the overconsumption of environmental resources and the inefficiency of production systems in the face of market fluctuations or supply chain disruptions. Traditional systems are restricted by static data and subjective judgments, lacking intelligence and adaptability, which limits the scientific nature and flexibility of production decisions. However, by introducing environmental carrying capacity assessment models, artificial intelligence technologies, and real-time big data analysis, more sustainable and intelligent production control optimization can be achieved, ensuring that production activities operate efficiently within the environmental tolerance range, adapting to rapidly changing market and environmental conditions, and promoting the production system towards a more advanced development direction. Summary of the Invention

[0006] The object of the present invention is to provide an optimized enterprise production control strategy considering environmental carrying capacity to solve the problems existing in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides an optimized enterprise production control strategy considering environmental carrying capacity, including the following steps:

[0008] S1. Evaluation of environmental carrying capacity, which is used to provide the environmental background for the safe operation of the enterprise during the production process and provide important data support for the subsequent enterprise production plan and resource analysis;

[0009] S2. Analysis of enterprise production requirements and resources. The enterprise needs to analyze its own production requirements and resource usage in detail, including the analysis of the demand for raw materials, energy, water resources, etc. and the consumption patterns; at the same time, analyze the enterprise's production plan and production capacity arrangement to ensure the matching with the environmental carrying capacity and avoid production activities that exceed the environmental carrying capacity.

[0010] S3. Construction of the reinforcement learning environment and model;

[0011] S4. Real-time monitoring and early warning. Through the real-time monitoring and data collection functions of the production control system, the real-time monitoring and data recording of the environmental pollution concentration are realized.

[0012] Preferably, the evaluation of environmental carrying capacity in step S1 includes the evaluation of water environmental capacity and the evaluation of atmospheric environmental capacity.

[0013] Preferably, the evaluation of water environmental capacity is specifically as follows: The indicators for measuring water environmental capacity include the self-purification ability and dilution ability of the water body to pollutants. The river water quality is simulated by the MIKE11 water quality model. Since the MIKE11 water quality model has taken into account the attenuation processes such as microbial degradation, plant interception, and plant absorption of pollutants during the river simulation process, only the dilution ability of the water body is considered in the calculation of water environmental carrying capacity. The dilution ability of the water body is calculated according to the MIKE11 water quality model to determine the water environmental capacity, and the calculation formula is

[0014]

[0015] where Q is the calculated flow rate, with the unit of m 3 / s; q is the calculated unit, i.e., the sewage discharge in the study area, with the unit of m 3 / s; C0 is the pollutant concentration in the upstream water simulated by the MIKE11 water quality model, with the unit of mg / L; C s is the pollutant water quality target concentration in the calculated unit, i.e., the study area, with the unit of mg / L; C1 is the sewage discharge concentration of the river section, with the unit of mg / L; W iFor the computing unit, i.e., the water environmental capacity of the research area, the unit is g / s;

[0016] Let the dilution flow ratio Then W i The derivation is as follows:

[0017]

[0018] Preferably, the atmospheric environment assessment calculates the atmospheric environmental capacity through the multi-source model method. Specifically: The multi-source model method is the basic and main method for calculating the actual environmental capacity. First, the emissions of different pollution sources in the region should be simulated to calculate the ground pollutant concentration; according to the pollutant concentration standard of the control point, the allowable emissions of each pollution source are inversely deduced to determine the atmospheric environmental capacity; the formula for calculating the regional atmospheric environmental capacity is

[0019]

[0020] Among them, Q a is the atmospheric environmental capacity, and the unit is t·a -1 ; C0 is the pollutant concentration limit, and the unit is mg·m -3 ; C s is the background concentration of regional pollutants, and the unit is mg·m-3; C 关心点 is the predicted concentration value of the control point, and the unit is mg·m-3; Q 现有 is the source strength of existing projects in the control area, and the unit is t·a -1 .

[0021] Preferably, the goal of the reinforcement learning in step S3 is to find the optimal policy π * , so that the expected value of the discounted cumulative return is maximized. The enterprise production control is abstracted into the basic Markov decision process of reinforcement learning, which is composed of the five-tuple S, A, P, r, γ. The state space S is defined as the set of the current operating states of the production line (such as equipment operating states, production rates, etc.), the order completion situation, and the raw material inventory level states, describing all different states that the system may be in at each time; the action space A is defined as the decisions such as adjusting the production line rate, dispatching workers, and purchasing raw materials, representing the set of all possible actions available in each state; the state transition function P(S, A, S’) is defined as the uncertainty in the production process, such as state transitions caused by equipment failures, raw material quality changes, etc.; the reward function r(S, A) represents the immediate reward or penalty obtained by the agent after executing the action space A in the S state. According to the optimization goal of the enterprise, the immediate reward is defined as the profit from order completion, the improvement of production efficiency, and the cost reduction of inventory; the degree of emphasis on future rewards is defined as the discount factor γ, and γ will affect the balance between long-term planning and short-term benefits; the reward function and the discount factor together constitute the discounted cumulative return G t, and its formula is:

[0022]

[0023] Among them, K represents the time used for a complete cycle of production control by the enterprise. Starting from the initial state (such as a certain production line state, order completion situation, and raw material inventory level), the agent selects actions according to the current state (adjusting the production line rate, scheduling workers, purchasing raw materials, etc.) and updates the state according to the state transition function until a certain termination condition is reached (such as reaching the specified number of time steps); r t represents the immediate reward or penalty obtained by the agent according to the current state space S and action space A after performing an action at a specific time step t.

[0024] Preferably, the optimal policy π * is obtained as follows:

[0025] Define the expected discounted cumulative return G t as the value function V(s), and define the long-term return expectation brought by taking different decisions (such as scheduling, resource allocation, etc.) in a specific state as the action-value function Q(s,a). The expressions are as follows respectively

[0026]

[0027] Among them, S t represents the state at time step t; A t represents the action selected by the agent at time step t; s and a respectively represent the values of the state space and action space in the MDP; π represents the policy, which describes the probability distribution of the agent selecting each action in a given state;

[0028] By calculating the value function V(s) of the state and the action-value function Q(s,a), the optimal policy π * is obtained. By establishing the above MDP model, the reinforcement learning method can be used to find the optimal production control strategy, enabling the enterprise to maximize benefits and comply with the environmental carrying capacity in a dynamically changing environment.

[0029] Preferably, step S4 is specifically as follows:

[0030] Set the safety state threshold of the environmental carrying capacity, judge the degree of environmental pollution by comparing the calculation results at each moment with the safety state threshold; divide the confidence region for monitoring the degree of environmental pollution to determine the range of the safety state threshold. The formula is

[0031] P[θ1≤θ≤θ2]=1 - a

[0032] Wherein, P is the confidence range for monitoring the environmental pollution status; a is the significance level; θ is the confidence level; θ1 is the minimum value of the confidence level; θ2 is the maximum value of the confidence level;

[0033] When an abnormal situation is detected or the parameter exceeds the safety status threshold, the system automatically issues a warning and notifies the relevant personnel for handling, timely discovers potential problems of the equipment, and prevents pollutant emissions from exceeding the environmental carrying capacity.

[0034] Therefore, by adopting the above-mentioned optimization strategy for enterprise production control considering the environmental carrying capacity, the present invention has the following beneficial effects:

[0035] (1) An environmental carrying capacity assessment model is added to the traditional production control system, improving the accuracy of measuring the use of environmental resources, avoiding overuse of environmental resources leading to exceeding the environmental carrying limit, and reducing the damage to the ecosystem;

[0036] (2) By introducing artificial intelligence and machine learning technologies, an intelligent decision support system is established, providing data-driven optimized decisions while improving the scientificity and accuracy of decisions, making the decision-making scheme no longer affected by subjective factors such as manual experience;

[0037] (3) By applying real-time monitoring and big data analysis technologies, the production plan is dynamically adjusted to optimize resource allocation and production rhythm, ensuring that production activities can be carried out within the environmental carrying capacity, and solving the problem that the production control system based on static data cannot respond to changes in the environment and production conditions in a timely manner.

[0038] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0039] Figure 1 It is a schematic diagram for evaluating the water environment capacity in the present invention;

[0040] Figure 2 It is a schematic diagram for evaluating the atmospheric environment capacity in the present invention;

[0041] Figure 3 It is a schematic diagram of the Markov decision process in the present invention;

[0042] Figure 4 It is an interaction process diagram between the environment and the reinforcement learning agent in the present invention;

[0043] Figure 5 It is a flow chart of the present invention. Detailed Embodiment

[0044] Embodiment

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Generally, the components of the embodiments of the present invention described and illustrated in the drawings here can be arranged and designed in various different configurations.

[0046] As Figures 1-5 shown, an optimization strategy for enterprise production control considering environmental carrying capacity includes the following steps:

[0047] S1. Environmental carrying capacity assessment, which is used to provide an environmental background for the safe operation of an enterprise during production and provide important data support for subsequent enterprise production planning and resource analysis;

[0048] S11. Water environment capacity assessment

[0049] The indicators for measuring the water environment capacity include the self-purification ability and dilution ability of water bodies to pollutants. The water quality of rivers is simulated through the MIKE11 water quality model. Since the MIKE11 water quality model has taken into account the attenuation processes such as microbial degradation, plant interception, and plant absorption of pollutants during the river simulation process, only the water body dilution ability is considered in the calculation of the water environment carrying capacity. According to the MIKE11 water quality model, the dilution ability of the water body is calculated to determine the water environment capacity. The calculation formula is

[0050]

[0051] where Q is the calculated flow rate, with the unit of m 3 / s; q is the sewage discharge in the calculation unit, that is, the sewage discharge in the study area, with the unit of m 3 / s; C0 is the pollutant concentration in the upstream water simulated by the MIKE11 water quality model, with the unit of mg / L; C s is the pollutant water quality target concentration in the calculation unit, that is, the study area, with the unit of mg / L; C1 is the sewage discharge concentration of the river section, with the unit of mg / L; W i is the water environment capacity of the calculation unit, that is, the study area, with the unit of g / s;

[0052] Let the dilution flow ratio Then W i is deduced as:

[0053]

[0054] S12. Atmospheric environment capacity assessment

[0055] The assessment of the atmospheric environmental capacity calculates the atmospheric environmental capacity through the multi-source model method. Specifically: The multi-source model method is the basic and main method for calculating the actual environmental capacity. First, the emissions of different pollution sources in the region should be simulated, and the ground pollutant concentration should be calculated; according to the pollutant concentration standard of the control points, the allowable emissions of each pollution source should be deduced backwards to determine the atmospheric environmental capacity; the formula for calculating the regional atmospheric environmental capacity is

[0056]

[0057] Among them, Q a is the atmospheric environmental capacity, with the unit of t·a -1 ; C0 is the pollutant concentration limit, with the unit of mg·m -3 ; C s is the background concentration of regional pollutants, with the unit of mg·m-3; C 关心点 is the predicted concentration value of the control points, with the unit of mg·m-3; Q 现有 is the source strength of existing projects in the control area, with the unit of t·a -1

[0058] S2. Analysis of enterprise production demand and resources. The enterprise needs to analyze its own production demand and resource usage in detail, including the analysis of the demand for raw materials, energy, water resources, etc. and the consumption pattern; at the same time, analyze the enterprise's production plan and production capacity arrangement to ensure matching with the environmental carrying capacity and avoid production activities that exceed the environmental carrying capacity.

[0059] S3. Construction of the reinforcement learning environment and model;

[0060] The goal of reinforcement learning is to find the optimal policy π *, maximizing the expected discounted cumulative return, abstract the enterprise production control as the basic Markov decision process of reinforcement learning, which is composed of the five-tuple S, A, P, r, γ. Define the state space S as the set of the current operating states of the production line (such as equipment operating state, production rate, etc.), order completion status, and raw material inventory levels, describing all different states that the system may be in at each time; define the action space A as the decisions such as adjusting the production line rate, scheduling workers, purchasing raw materials, etc., representing the set of all possible actions available in each state; define the state transition function P(S, A, S') as the uncertainties in the production process, such as state transitions caused by equipment failures, changes in raw material quality, etc.; the reward function r(S, A) represents the immediate reward or punishment obtained by the agent after executing the action space A in the S state. According to the optimization objectives of the enterprise, define the immediate reward as the profit from completing orders, the improvement of production efficiency, and the cost reduction of inventory; define the degree of emphasis on future rewards as the discount factor γ, and γ will affect the balance between long-term planning and short-term benefits; the reward function and the discount factor together constitute the discounted cumulative return G t , and its formula is:

[0061]

[0062] where K represents the time taken for a complete cycle of the enterprise's production control. Starting from the initial state (such as a certain production line state, order completion status, and raw material inventory level), the agent selects actions (adjusting the production line rate, scheduling workers, purchasing raw materials, etc.) according to the current state, and updates the state according to the state transition function until a certain termination condition is reached (such as reaching the specified number of time steps); r t represents the immediate reward or punishment obtained by the agent according to the current state space S and action space A after executing the action at a specific time step t.

[0063] The optimal policy π * is obtained as follows:

[0064] Define the expected discounted cumulative return G t as the value function V(s), and define the expected long-term return brought by taking different decisions (such as scheduling, resource allocation, etc.) in a specific state as the action-value function Q(s, a). The expressions are as follows

[0065]

[0066] where S t represents the state at time step t; A tDenote the action selected by the agent at time step t; s and a represent the values in the state space and action space of the MDP respectively; π represents the policy, which describes the probability distribution of the agent selecting each action given a state.

[0067] By calculating the value function V(s) of the state and the action-value function Q(s,a), the optimal policy π can be obtained. * By establishing the above MDP model, reinforcement learning methods can be used to find the optimal production control strategy, enabling the enterprise to maximize benefits and comply with the environmental carrying capacity in a dynamically changing environment.

[0068] S4. Real-time monitoring and early warning. Through the real-time monitoring and data collection functions of the production control system, real-time monitoring and data recording of environmental pollution concentration are achieved.

[0069] Set the safety state threshold of the environmental carrying capacity. By comparing the calculation results at each moment with the safety state threshold, the degree of environmental pollution is judged; divide the confidence region of the environmental pollution degree monitoring to determine the range of the safety state threshold. The formula is

[0070] P[θ1≤θ≤θ2]=1 - a

[0071] where P is the confidence range of the environmental pollution state monitoring; a is the significance level; θ is the confidence level; θ1 is the minimum value of the confidence level; θ2 is the maximum value of the confidence level.

[0072] When abnormal situations or parameters exceeding the safety state threshold are detected, the system automatically issues an early warning and notifies relevant personnel for handling, timely discovers potential problems of equipment, and prevents pollutant emissions from exceeding the environmental carrying capacity.

[0073] The following is illustrated with a specific example.

[0074] A chemical enterprise is located in an area rich in water resources but environmentally sensitive. It is necessary to evaluate its environmental carrying capacity and implement a real-time monitoring system. The production activities of this enterprise mainly consume water resources and discharge wastewater containing chemicals into the nearby river. First, we need to evaluate the water environment capacity: Use the MIKE11 water quality model to predict the impact of wastewater discharge on the river water quality. Assume that the enterprise discharges 1000 cubic meters of wastewater per day, and the water body needs to dilute it to below the safe concentration; Assume that according to the model calculation, the river can effectively dilute about 1200 cubic meters of wastewater per day. Therefore, the water environment capacity of the enterprise is to discharge at most 1000 cubic meters of wastewater per day. Then evaluate the atmospheric environment capacity: Assume that the enterprise discharges 1000 tons of sulfur dioxide and 500 tons of nitrogen oxides per year. According to the local atmospheric environmental quality standard, assume that the annual average concentration of sulfur dioxide and nitrogen oxides shall not exceed a certain preset value; Through the atmospheric diffusion model and the emission data of emission sources, calculate the annual allowable emissions of the enterprise to ensure that it does not exceed the carrying capacity of the atmospheric environment.

[0075] Analyze the production plan of the enterprise and the resource consumption pattern, and use the Markov decision process (MDP) model to optimize the production control of the enterprise; Let the current production configuration and environmental pollution level be the state space S, and the decisions such as adjusting the production plan, optimizing resource allocation or increasing pollution control investment be the action space A, the uncertainty in the production process be the state transition function P, and the maximum profit of keeping environmental pollution within an acceptable range be the reward function r(s,a). The degree of emphasis on future rewards is the discount factor γ. Select a suitable reinforcement learning algorithm for model learning and optimization, and decide on a plan that can maximize the enterprise's benefits and comply with the environmental carrying capacity.

[0076] Use the real-time monitoring system to monitor the real-time wastewater discharge and the concentration of atmospheric pollutants of the enterprise. If it is detected that the wastewater concentration exceeds the preset threshold or the concentration of atmospheric pollutants exceeds the allowable value, the system will immediately send an alarm to the environmental protection department and the enterprise management. The real-time monitoring system analyzes the monitored data in real time, identifies potential production optimization or improvement opportunities, and feeds them back for continuous improvement of the production process and the early warning system itself to ensure that the system can more effectively identify and respond to potential problems in the future.

[0077] Therefore, the present invention adopts the above-mentioned enterprise production control optimization strategy considering environmental carrying capacity, which can realize the optimization of enterprise production control on the premise of meeting the environmental carrying capacity, and can conduct real-time monitoring and early warning, which helps the enterprise to protect the environment, reduce pollution risks while achieving economic benefits, and promote sustainable development and the long-term stable operation of the enterprise.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements do not cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An enterprise production control optimization strategy taking into account environmental carrying capacity, characterized in that: The following steps are involved: S1. Environmental carrying capacity assessment is used to provide an environmental background for safe operation during the production process and provide data support for subsequent enterprise production planning and resource analysis; S2. Enterprise production demand and resource analysis; S3, reinforcement learning environment and model construction; S4. Real-time monitoring and early warning: through the real-time monitoring and data collection functions of the production control system, the real-time monitoring and data recording of environmental pollution concentration can be achieved; The goal of reinforcement learning in step S3 is to find the optimal strategy π * , the production control of the enterprise is abstracted as the basic Markov decision process of reinforcement learning, which consists of a five-tuple S, A, P, r, γ. The state space S is defined as the current operating state of the production line, the completion of orders, and the state set of raw material inventory levels, describing all the different states of the system at each time; the action space A is defined as the set of all actions available for selection in each state, including the decision to adjust the production line rate, schedule workers, and purchase raw materials; the state transition function P(S, A, S') is defined as the uncertainty in the production process; the reward function r(S, A) represents the immediate reward or penalty obtained by the agent after executing the action space A in the S state. According to the optimization goal of the enterprise, the immediate reward is defined as the profit of completing the order, the improvement of production efficiency, and the cost of reducing inventory; the importance of future rewards is defined as the discount factor γ, which will affect the balance between long-term planning and short-term benefits; the reward function and the discount factor together constitute the discounted cumulative return G t , the formula is: Among them, K represents the time taken by the enterprise to carry out a complete cycle of production control, starting from the initial state, through the agent to select actions according to the current state, and update the state according to the state transfer function until the termination condition is reached; r t Represents the immediate reward or penalty obtained by the agent based on the current state space S and action space A after performing an action at a specific time step t; Optimal strategy π * The acquisition process is as follows: The expected discounted cumulative return G t Defined as the value function V(s), the expected long-term return brought by taking different decisions in a specific state is defined as the action-value function Q(s,a), and the expressions are as follows Among them, S t represents the state at time step t; A t represents the action selected by the agent at time step t; s and a represent the values ​​of the state space and action space in the MDP respectively; π represents the strategy, which describes the probability distribution of each action selected by the agent in a given state; By calculating the state value function V(s) and the action-value function Q(s,a), we can get the optimal strategy π * .

2. According to claim 1, the enterprise production control optimization strategy taking into account the environmental carrying capacity is characterized by: The environmental carrying capacity assessment in step S1 includes water environment capacity assessment and atmospheric environment capacity assessment.

3. According to claim 2, the enterprise production control optimization strategy taking into account the environmental carrying capacity is characterized in that: The specific water environment capacity assessment is as follows: The indicators for measuring water environment capacity include the self-purification and dilution capacity of water bodies for pollutants. The dilution capacity of water bodies is calculated through the MIKE11 model to determine the water environment capacity. The calculation formula is: Where Q is the calculated flow rate, in m 3 / s; q is the sewage discharge volume in the calculation unit, in m 3 / s; C0 is the pollutant concentration in the upstream water simulated by the MIKE11 model, in mg / L; C s is the target concentration of pollutants in the calculation unit, in mg / L; C1 is the sewage discharge concentration in the river section, in mg / L; W i To calculate the unit water environment capacity, the unit is g / s; Dilution flow ratio Then W i It is derived as follows:

4. According to claim 2, the enterprise production control optimization strategy taking into account the environmental carrying capacity is characterized in that: Atmospheric environmental assessment calculates atmospheric environmental capacity through the multi-source model method, specifically: simulate the emissions of different pollution sources in the region and calculate the ground pollutant concentration; according to the pollutant concentration standard of the control point, invert the allowable emission of each pollution source to determine the atmospheric environmental capacity; the regional atmospheric environmental capacity calculation formula is: Among them, Q a is the atmospheric environmental capacity, in t·a -1 ; C0 is the pollutant concentration limit, in mg·m -3 ; C s is the background concentration of regional pollutants, in mg·m-3; C 关心点 is the predicted concentration value of the control point, in mg·m-3; Q 现有 It is the source strength of existing projects in the control area, the unit is t·a -1 .

5. According to claim 1, the enterprise production control optimization strategy taking into account the environmental carrying capacity is characterized in that: Step S4 is specifically as follows: Set the safety state threshold of the environmental carrying capacity, and judge the degree of environmental pollution by comparing the calculation results at each moment with the safety state threshold; divide the confidence area for environmental pollution monitoring and determine the safety state threshold range. The formula is: P[θ1≤θ≤θ2]=1-a Among them, P is the confidence range of environmental pollution status monitoring; a is the significance level; θ is the confidence level; θ1 is the minimum value of the confidence level; θ2 is the maximum value of the confidence level; When an abnormal situation is detected or parameters exceed the safety status threshold, the system automatically issues an early warning and notifies personnel to handle the situation.

Images