Ai-based high-speed composting system for livestock manure

Abstract

The present invention provides an artificial intelligence-based high-speed livestock manure composting system comprising: a livestock manure composting facility equipped with an internal sensor module and an external sensor module that operate to compost livestock manure stored internally and measure internal state information and external environment information, and an environmental control device that controls the internal environment; and a control unit that analyzes the internal state information and external environment information obtained from the internal sensor module and the external sensor module through an artificial intelligence model and controls the operation of the environmental control device according to the analysis results, wherein the control unit comprises: an input processing unit that collects the internal state information and external environment information and converts them into time-series data; an artificial intelligence analysis unit that analyzes the time-series data transmitted from the input processing unit through an artificial intelligence model to predict internal state information at a future point in time, estimates the degree of maturation of the current compost, and generates operating conditions for the environmental control device to increase the composting speed by considering the prediction and estimation results; and an environmental control control unit that controls the operation of the environmental control device according to the operating conditions generated through the artificial intelligence analysis unit.

Description

Technology Field

[0001] The present invention relates to an artificial intelligence-based high-speed composting system for livestock manure. More specifically, the invention relates to an artificial intelligence-based high-speed composting system for livestock manure that can shorten composting time and improve composting efficiency and compost quality by using an artificial intelligence model to predict the state and estimate the degree of maturation during the composting process of livestock manure, thereby setting optimal environmental conditions, and controlling an environmental control device based on this. Background Technology

[0002] Livestock manure is one of the major types of organic waste generated globally, and improper disposal can lead to serious environmental problems, including water and soil pollution, odor generation, and greenhouse gas emissions. To address these issues and create a sustainable agricultural environment, the efficient and stable treatment of livestock manure is essential. Composting is an eco-friendly method that contributes to soil improvement and increased agricultural productivity by converting organic waste into stable organic fertilizer; during this process, organic matter within the manure is decomposed by aerobic microorganisms and transformed into compost in the form of humus.

[0003] However, conventional composting methods have several limitations, such as long processing times, odor generation, and the need for large land areas. In particular, if insufficiently matured compost is applied to agricultural land, the ammonia gas produced during the decomposition process causes severe odors and serious hindrances to crop growth. Due to these issues, strict management of maturity levels is legally mandated to ensure that only fully matured compost is distributed.

[0004] Against this backdrop, 'high-speed composting' technology is emerging as an alternative that drastically shortens the composting period and improves compost quality. High-speed composting is a technology that accelerates the decomposition of organic matter by optimizing the activity of aerobic microorganisms, and it requires precise and dynamic control of key environmental variables such as temperature, humidity (moisture content), and oxygen concentration inside the compost. However, since each of these environmental variables is interdependent, the introduction of artificial intelligence (AI) technology is essential to efficiently manage these complex biochemical reaction processes. Prior art literature

[0005] Korean Patent Publication No. 10-2022-0137356 The problem to be solved

[0006] The present invention was developed to solve the problems of the prior art. The objective of the present invention is to provide an AI-based high-speed composting system for livestock manure that can shorten composting time and improve quality by inferring changes in the state of the fermentation process and the maturation state through an AI model based on time-series data of various environmental variables in a composting facility, setting optimal environmental conditions to maintain the activity of fermentation microorganisms at a maximum level, and continuously controlling an environmental control device according to the set environmental conditions. means of solving the problem

[0007] The present invention provides an artificial intelligence-based high-speed livestock manure composting system comprising: a livestock manure composting facility equipped with an internal sensor module and an external sensor module that operate to compost livestock manure stored internally and measure internal state information and external environment information, and an environmental control device that controls the internal environment; and a control unit that analyzes the internal state information and external environment information obtained from the internal sensor module and the external sensor module through an artificial intelligence model and controls the operation of the environmental control device according to the analysis results, wherein the control unit comprises: an input processing unit that collects the internal state information and external environment information and converts them into time-series data; an artificial intelligence analysis unit that analyzes the time-series data transmitted from the input processing unit through an artificial intelligence model to predict internal state information at a future point in time, estimates the degree of maturation of the current compost, and generates operating conditions for the environmental control device to increase the composting speed by considering the prediction and estimation results; and an environmental control control unit that controls the operation of the environmental control device according to the operating conditions generated through the artificial intelligence analysis unit.

[0008] At this time, the internal state information may include at least one of temperature, humidity, oxygen concentration, acidity (pH), ammonia concentration, carbon dioxide concentration, carbon / nitrogen concentration ratio, and electrical conductivity for the internal space or compost of the livestock manure composting facility.

[0009] Additionally, the artificial intelligence analysis unit may include: an environment change prediction unit including an LSTM model that analyzes time series data transmitted from the input processing unit to predict internal state information for a future point in time; a maturity degree estimation unit including an XGBoost model that estimates the current degree of maturity of compost based on gas concentration among the internal state information; an environment information calculation unit including a DDPG reinforcement learning model that calculates target environment information capable of shortening the composting time based on the prediction result of the environment change prediction unit and the estimation result of the maturity degree estimation unit; and an operation condition generation unit that generates operation conditions for the environment control device according to the target environment information calculated through the environment information calculation unit.

[0010] In addition, the above-mentioned maturation state estimation unit can estimate the degree of maturation of the compost by analyzing the change in the gas generation amount and mutual generation ratio of carbon dioxide and ammonia among the above-mentioned internal state information through the XGBoost model.

[0011] In addition, the target environment information calculated by the environment information calculation unit may include information regarding temperature control values, moisture control values, and oxygen supply control values.

[0012] In addition, the above DDPG reinforcement learning model can be trained using a reward function calculated based on the ammonia / carbon dioxide ratio, carbon / nitrogen ratio, and composting time.

[0013] In addition, the DDPG reinforcement learning model can be retrained to reflect the fermentation characteristics of each type of livestock manure by updating the policy for determining the target environment information using time series data transmitted from the input processing unit for each type of livestock manure.

[0014] In addition, the control unit further includes an operating state analysis unit that collects operating data of the environment control device, analyzes the operating state of the environment control device, and transmits the operating data and analysis results to the artificial intelligence analysis unit through the input processing unit. The artificial intelligence analysis unit calculates an error between the actual operating state of the environment control device and the operating conditions based on the operating data and analysis results, and can correct the operating conditions by reflecting the calculated error. Effects of the invention

[0015] According to the present invention, in the process of composting livestock manure, optimal environmental conditions are set by predicting the state and estimating the degree of maturation through an artificial intelligence model, and by controlling an environmental control device based on this, the composting time can be shortened and the composting efficiency and quality of the compost can be improved.

[0016] In addition, unlike traditional manual control methods or rule-based automatic control methods, by setting and controlling optimal environmental conditions according to the environment and maturity status, it is possible to perform an optimized composting process for various types of compost. Brief explanation of the drawing

[0017] FIG. 1 is a conceptual diagram illustrating the overall configuration of an artificial intelligence-based high-speed livestock manure composting system according to one embodiment of the present invention. FIG. 2 is a block diagram illustrating the configuration of a control unit according to an embodiment of the present invention, classified according to function. FIG. 3 is a diagram illustrating, in an exemplary manner, the configuration of a neural network architecture of an LSTM model according to one embodiment of the present invention. FIG. 4 is a conceptual diagram illustrating the process of estimating compost maturity through the XGBoost model according to one embodiment of the present invention. FIG. 5 is a conceptual diagram illustrating the process of calculating target environment information through a reinforcement learning model according to one embodiment of the present invention. Specific details for implementing the invention

[0018] Hereinafter, specific embodiments for implementing the present invention will be described in detail with reference to the drawings.

[0019] First, it should be noted that when assigning reference numerals to the components of each drawing, the same components are assigned the same reference numeral whenever possible, even if they are shown in different drawings. Furthermore, in describing the present invention, if it is determined that a detailed description of related known components or functions could obscure the essence of the invention, such detailed description is omitted.

[0020] Furthermore, when it is stated that one component is 'connected,' 'supported,' 'connected,' 'supplied,' 'transmitted,' or 'contacted' with another component, it should be understood that while the connection, support, connection, supply, transmission, or contact may be direct to that other component, there may also be other components present in between.

[0021] The terms used in this specification are used merely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0022] Furthermore, it should be noted in advance that expressions such as "upper side," "lower side," and "side" in this specification are described based on the drawings, and may be expressed differently if the orientation of the object changes. For the same reason, some components in the attached drawings may be exaggerated, omitted, or schematically depicted, and the size of each component does not entirely reflect its actual size.

[0023] Additionally, terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but such components are not limited by such terms. These terms are used solely for the purpose of distinguishing one component from another.

[0024] The meaning of "comprising" as used in the specification is to specify certain characteristics, regions, integers, steps, actions, elements, and / or components, and does not exclude the existence or addition of other specific characteristics, regions, integers, steps, actions, elements, components, and / or groups.

[0025] FIG. 1 is a conceptual diagram illustrating the overall configuration of an artificial intelligence-based high-speed livestock manure composting system according to one embodiment of the present invention.

[0026] An artificial intelligence-based high-speed composting system for livestock manure according to one embodiment of the present invention may be configured to include a livestock manure composting facility (100) and a control unit (200) that controls the operation of the livestock manure composting facility (100) under specific operating conditions.

[0027] The livestock manure composting facility (100) is a device that converts livestock manure into compost by decomposing organic matter through a microbial fermentation reaction while storing livestock manure inside. During the fermentation process, the temperature of the internal environment rises, oxygen consumption, moisture evaporation, microbial activity increases, and gases such as carbon dioxide (CO2) and ammonia (NH3) are generated dynamically over time. In particular, since the physical properties and fermentation reaction rate of livestock manure may vary by batch due to factors such as feed composition, livestock species characteristics, moisture content, and seasonal influences, if the internal environmental conditions are not properly controlled, quality degradation problems such as delayed fermentation reaction, odor generation, excessive heat generation, and the occurrence of immature compost may occur.

[0028] Accordingly, in one embodiment of the present invention, a sensor module (110, 120) is provided to monitor the internal state and external environmental state of the composting facility (100) in real time. The external sensor module (110) can measure ambient environmental information such as ambient temperature, ambient humidity, ambient CO₂ concentration, and atmospheric pressure, and since such external environmental information can affect the fermentation reaction and heat and moisture exchange behavior inside the facility, it can be utilized in establishing control policies in the future.

[0029] The internal sensor module (120) can measure internal state information such as temperature, humidity, oxygen concentration, acidity (pH), ammonia concentration, carbon dioxide concentration, carbon / nitrogen (C / N) ratio, and electrical conductivity for the space inside the composting facility or the livestock manure itself. For example, in the initial stage of fermentation, the internal temperature rises rapidly and the oxygen concentration decreases as microbial activity increases, whereas after the middle stage of fermentation, changes in CO₂ and NH₃ production and a decrease in the C / N ratio can be observed. This information can be usefully utilized for determining the fermentation stage and estimating the degree of maturity.

[0030] The environmental control device (130) is configured to control the internal environment of the composting facility and may include actuators capable of supplying oxygen and discharging exhaust gas using a blower module, controlling moisture through a moisture supply module, controlling the temperature rise through a heating module, and improving air permeability and mixing uniformity through a stirring module. For example, if the internal temperature rises excessively, the temperature can be lowered by introducing outside air through the blower module, and conversely, if the temperature does not rise sufficiently during the initial fermentation stage in winter, the heating module can be activated to promote microbial activity. Additionally, control can be achieved by adjusting the amount of spray through the moisture supply module if the moisture content is insufficient, and by increasing the amount of airflow if the oxygen supply is insufficient.

[0031] The operating status measuring unit (140) can measure data reflecting the actual operating status of the environmental control device (130). The measured data may consist of current, voltage, rotational speed (RPM), airflow rate, heat amount, moisture supply amount, operating time, load, and power consumption, and these data are useful for determining whether the environmental control device (130) executes commands and whether there is a change in actuator performance. In particular, in livestock farming sites, device performance degradation, such as contamination of the device or a decrease in airflow rate due to increased load, may frequently occur; therefore, as in one embodiment of the present invention, feedback of the operating status data can be reflected in the AI ​​control policy to ensure quality stability of the composting process and prevent failure of the device.

[0032] The control unit (200) is configured to analyze internal state information and external environment information from the internal sensor module (120) and the external sensor module (110) through an artificial intelligence model, and to control the operation of the environment control device (130) according to the analysis results. Below, the control unit (200) will be examined in detail with reference to FIG. 2.

[0033] FIG. 2 is a block diagram illustrating the configuration of a control unit according to an embodiment of the present invention, classified according to function.

[0034] The control unit (200) may be configured to include an input processing unit (210) that collects internal state information and external environment information and converts them into time series data; an artificial intelligence analysis unit (220) that analyzes the time series data transmitted from the input processing unit (210) through an artificial intelligence model to predict internal state information at a future point in time, estimates the degree of maturation of the current compost, and generates operating conditions for an environment control device (130) to increase the composting speed by considering the prediction and estimation results; an environment control control unit (230) that controls the operation of the environment control device (130) according to the operating conditions generated through the artificial intelligence analysis unit (220); and an operation state analysis unit (240) that collects operation data of the environment control device (130), analyzes the operation state of the environment control device (130), and transmits the operation data and analysis results to the artificial intelligence analysis unit (220) through the input processing unit (210).

[0035] The input processing unit (210) can collect internal state information and external environment information collected through the external sensor module (110) and the internal sensor module (120), convert them into a time-series data format aligned along the time axis, and then transmit them to the artificial intelligence analysis unit (220). During this conversion process, the input processing unit (210) may apply a time stamp-based synchronization technique to correct for differences in collection cycles per sensor, communication delays, or missing data occurrences, and may be configured to maintain the continuity of the time-series data by performing interpolation or compensation operations as needed.

[0036] Additionally, the input processing unit (210) can perform preprocessing processes such as normalization, standardization, or range scaling (min-max scaling) to resolve scale discrepancies in input data that may occur due to differences in measurement units and ranges per sensor. These preprocessing processes can contribute to improving the learning efficiency of the LSTM model and reinforcement learning model of the artificial intelligence analysis unit (220).

[0037] The input processing unit (210) can also map the operation data provided through the operation status analysis unit (240) to time-series data on the same time axis. At this time, the input processing unit (210) can adjust the internal state information and the operation status information of the livestock manure composting facility (100) to be compared or analyzed at the same point in time by taking into account the measurement cycle and response delay difference of the operation data, and through this, the artificial intelligence analysis unit (220) can model the relationship between the control command and the actual operation of the environment control device (130).

[0038] Additionally, the input processing unit (210) can analyze the reaction sensitivity of the fermentation process by reflecting external environmental data such as ambient temperature, ambient humidity, and atmospheric pressure collected from the external sensor module (110) as needed, and can preprocess data so that an accurate control policy can be generated even under conditions where changes in the external environment have a significant impact on the fermentation speed, such as winter when the temperature has a large influence or summer when the moisture evaporation rate increases.

[0039] This input processing unit (210) performs time-series refinement and alignment functions of sensor-based composting data, thereby enabling the artificial intelligence analysis unit (220) to reliably generate input data for predicting the internal state at future time points and estimating the degree of maturation.

[0040] The artificial intelligence analysis unit (220) is configured to analyze changes in the internal state of the composting facility (100) over time and generate control policies necessary for the operation of the environmental control device (130), and may include an environmental change prediction unit (221), a maturity degree estimation unit (222), an environmental information calculation unit (223), and an operation condition generation unit (224).

[0041] The environment change prediction unit (221) can predict internal state parameters at a future point in time by analyzing the time series changes of internal state information and external environment information provided through the input processing unit (210).

[0042] In one embodiment of the present invention, a Long Short-Term Memory (LSTM) model is used to predict dynamic changes such as internal temperature, humidity, oxygen concentration, CO₂, and NH₃ generation, in order to account for the characteristic that the fermentation reaction appears with a time delay due to microbial activity.

[0043] Generally, during the initial stage of composting, internal temperature and CO₂ concentration rise rapidly due to increased microbial activity; subsequently, during the mid-stage of fermentation, a pattern is observed in which oxygen consumption and gas production decrease while moisture evaporation increases. Since these dynamic responses can be significantly influenced by external environmental conditions (e.g., ambient temperature), predicting future states can be utilized for preemptive intervention in future control policies.

[0044] The prediction information produced by the environment change prediction unit (221) can be reflected in the process of calculating rewards and selecting actions in the reinforcement learning-based control policy described later.

[0045] An LSTM neural network architecture according to one embodiment of the present invention may be composed of 9 sequential feature inputs and 1 task label as shown in FIG. 3. The LSTM layer may have 10 hidden states, the shared layer may be composed of 16 and 32 neurons respectively, and the multihead layer may be composed of 16 and 8 neurons respectively.

[0046] The degree of maturity estimation unit (222) can estimate the degree of maturity of the current compost based on the amount of gas generated among internal state information. For example, the degree of maturity of the current compost can be estimated by analyzing the change in the amount of carbon dioxide and ammonia gas generated and the mutual generation ratio, and additionally, the degree of maturity of the compost can be quantitatively estimated by analyzing the amount of oxygen consumption and the change and rate of change of the carbon / nitrogen (C / N) ratio.

[0047] The composting process using livestock manure is a reaction system in which temperature, gas production, water evaporation, and nutrient changes dynamically over time due to microbial activity. In the initial fermentation stage, the decomposition of proteins and organic matter proceeds actively, resulting in increased production of ammonia and carbon dioxide and increased oxygen consumption. In contrast, in the middle and later stages, protein decomposition decreases and carbon decomposition reactions become relatively dominant, leading to a decrease in gas production and a pattern in which the C / N ratio gradually stabilizes.

[0048] In this embodiment, the XGBoost (Extreme Gradient Boosting) model may be applied to the degree of ripening estimation unit (222) to learn these time-series response patterns and model the correlation between non-linear variables. The XGBoost model has the advantage of being able to effectively handle noise, measurement deviations, flow rate fluctuations, and the influence of external environmental factors in sensor data occurring during the ripening process, and to automatically model the interaction between variables.

[0049] For example, since ammonia concentration reflects the state of protein and urea decomposition, carbon dioxide concentration and increase in generation reflect microbial respiration activity, and the C / N ratio reflects nutrient balance and the progress of maturation, the maturation stage can be estimated by using the rate of change (d / dt) or moving average (filtered signal) of these values ​​together. In addition, since the fermentation efficiency may decrease and the ammonia / carbon dioxide ratio may be distorted if the oxygen supply decreases or a moisture shortage occurs during the fermentation process, the maturation degree estimation unit (222) can improve reliability by also considering the operating status data of the environmental control device (130). Fig. 4 conceptually illustrates the process of estimating the degree of compost maturity through this XGBoost model.

[0050] The degree of maturity estimation unit (222) can classify the stages of compost maturity into multiple stages, such as “initial fermentation stage, high-temperature fermentation stage, intermediate fermentation stage, stabilization stage, and post-maturation stage,” and the results of these stage classifications are transmitted to the environmental information output unit (223) and can be used as state variables for a reinforcement learning-based control policy. At this time, since the maturity state of each stage is a key indicator for judging fermentation efficiency, the target environmental information generated by the reinforcement learning policy can be automatically adjusted with the goal of maximizing maturity efficiency.

[0051] Additionally, the degree of decomposition estimation unit (222) can utilize accumulated training data to model the state of variation between batches caused by the type of livestock manure (e.g., pig / cattle / poultry), seasonal temperature fluctuations, changes in feed components, and differences in moisture content per batch unit, and can improve the estimation accuracy by performing online updates or additional training as needed.

[0052] The environmental information output unit (223) receives as input the future state prediction result generated by the environmental change prediction unit (221) and the current maturation stage information calculated by the maturation degree estimation unit (222), and outputs target environmental information to maximize the efficiency of the composting process and shorten the total composting time. The target environmental information output at this time may be composed of control variables (action space) that are a mixture of continuous and discontinuous, such as temperature control values, oxygen supply amounts, and water supply amounts, and the control target at a specific point in time may be determined by comprehensively considering the microbial activity stage, fermentation heat generation amount, water evaporation amount, and changes in exhaust gas composition.

[0053] In this embodiment, the Deep Deterministic Policy Gradient (DDPG) algorithm can be applied as a reinforcement learning-based policy learning model. DDPG is a model capable of handling continuous control variables, and unlike general discrete action-based Q-learning or DQN series models, action values ​​can be calculated directly in the real domain; thus, it can directly generate real-valued control commands such as airflow, spray volume, and heating volume. This is advantageous considering that the composting process is highly sensitive to fine-tuning of temperature and moisture, and is a physical process in which the controlled quantity changes continuously.

[0054] The state used in the policy learning process of the environmental information output unit (223) may include not only real-time state information measured through the internal sensor module (120) and the external sensor module (110), but also future state changes predicted through the environmental change prediction unit (221). For example, if it is predicted that the internal temperature will continuously rise due to microbial fever, the model may select a temperature suppression control policy in advance. In addition, information on the degree of maturation provided by the degree of maturation estimation unit (222) can be used as a goal-based signal to reduce the process time within the policy.

[0055] Referring to FIG. 5, the reinforcement learning model used in the environmental information output unit (223) can define the problem of controlling the composting environment of livestock manure as an optimization problem of continuous control variables, and for this purpose, physics-chemical models of the composting process can be embedded in the simulator environment. This is because microbial activity, heat generation, gas generation, and nutrient changes occurring in the fermentation process are difficult to control accurately using only simple sensory estimation or rule-based control.

[0056] The simulator may include process key parameters such as temperature (T), moisture (H), oxygen concentration, ammonia concentration, carbon dioxide concentration, and C / N ratio as state variables, which may be composed of state variables combined with input data based on sensor modules (110, 120) and output values ​​of the degree of maturation estimation unit (222). The reinforcement learning model may generate operation commands for continuous control variables such as ΔT (temperature adjustment amount), ΔO₂ (oxygen supply adjustment amount), and ΔH (moisture supply adjustment amount) as actions, which may correspond one-to-one with physical control actions of the environment control device (130).

[0057] Additionally, the DDPG reinforcement learning model may use a compensation function calculated based on the ammonia / carbon dioxide ratio, carbon / nitrogen (C / N) ratio and the time required for composting to learn a policy for controlling the composting process of the composting facility (100). This compensation function is constructed based on key process indicators that affect the maturity quality and composting speed of the compost, which allows the composting process to be interpreted as an optimal control problem based on environmental response.

[0058] Meanwhile, in the composting process, the activity (μ) of fermentation microorganisms is influenced by fermentation environment variables as shown in Equation 1 below.

[0059]

[0060] According to the microbial activity model in Equation 1, the fermentation process exhibits maximum activity under optimal conditions (temperature, moisture, oxygen concentration, and acidity), and as conditions deviate from the optimal state, microbial activity decreases, leading to a reduction in the maturation rate.

[0061] The DDPG reinforcement learning model can simulate state changes according to actions (control inputs) using the above microbial activity model, and the results of the state changes can be reflected in the calculation of the reward function.

[0062] In addition, the DDPG reinforcement learning model can update its policy in a self-improving manner by accumulating batch-by-batch experience through an experience replay buffer, thereby responding to inter-batch variations caused by the type of livestock manure (e.g., cattle / pigs / poultry), moisture variation, differences in feed composition, and seasonal variations, and through this, it can be retrained to reflect the batch-by-batch or type-by-type fermentation characteristics of livestock manure.

[0063] For example, to solve the problem of batch-to-batch variation caused by the type of livestock manure, changes in feed composition, seasonal ambient temperature influence, differences in moisture content, differences in microbial communities and differences in the amount of heat generated during initial fermentation, the policy can be repeatedly relearned through time series data for each processing unit. Accordingly, the environmental information output unit (223) can operate in a control method in which control strategies are accumulated and improved based on experience, rather than simple rule-based control.

[0064] The policy update process can be performed in an online or offline manner. In the online method, the reinforcement learning model can directly utilize real-time feedback data from the composting facility (100) to update the policy, while in the offline method, the policy can be further improved using accumulated time-series data records. Additionally, the policy experience of a completed batch can be stored in the form of a replay buffer and subsequently recycled for control optimization of a new batch, thereby contributing to ensuring consistency in process quality and improving compost production efficiency.

[0065] The operation condition generation unit (224) is configured to generate operation conditions for the environment control device (130) according to the target environment information calculated through the environment information calculation unit (223), and can perform mapping so that the target environment information provided through the environment information calculation unit (223) is converted into control commands that can be applied to each component of the environment control device (130).

[0066] For example, control commands can be mapped in a manner such as a control command to increase the blower module RPM when the target environmental information is an increase in the target oxygen supply, and a control command to operate the heating module when the target temperature is an increase.

[0067] The operation condition generation unit (224) can generate commands by considering the physical response limit of the actuator, power load, operation delay time, and activation order, thereby extending the lifespan of the device and preventing malfunction.

[0068] Meanwhile, the operating state analysis unit (240) can analyze the actual operating state of the environment control device (130) by collecting operating data such as current, voltage, rotational speed (RPM), airflow volume, injection volume, or heat volume provided through the operating state measurement unit (140), and the analysis result can be transmitted to the artificial intelligence analysis unit (220) through the input processing unit (210).

[0069] The artificial intelligence analysis unit (220) can calculate the error between the operating conditions determined by the environmental information calculation unit (223) and the actual operating conditions using the operating data and analysis results provided through the operating condition analysis unit (240). This error may occur due to performance degradation of the environmental control device (130), increased load, reduced flow due to pollution, or seasonal influences, and the artificial intelligence analysis unit (220) can control the composting process so that it can be stably performed under target conditions by correcting the operating conditions to be applied at the next time point by reflecting the error value.

[0070] Functions related to artificial intelligence according to the present disclosure are operated through a processor and memory. The processor may be composed of one or more processors. In this case, the one or more processors may be general-purpose processors such as CPUs, APs, and DSPs (Digital Signal Processors), graphics-dedicated processors such as GPUs and VPUs (Vision Processing Units), or artificial intelligence-dedicated processors such as NPUs. The one or more processors control the processing of input data according to predefined operation rules or artificial intelligence models stored in memory. Alternatively, if the one or more processors are artificial intelligence-dedicated processors, the artificial intelligence-dedicated processors may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0071] The predefined rules of operation or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that a basic artificial intelligence model is trained using a number of training data by a learning algorithm, thereby creating predefined rules of operation or artificial intelligence models configured to perform desired characteristics (or objectives). Such learning may be performed on the device itself where the artificial intelligence according to the present disclosure is executed, or it may be performed through a separate server and / or system. Examples of learning algorithms include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but are not limited to the examples described above.

[0072] The configuration of the artificial intelligence-based high-speed livestock manure composting system described herein may be realized by digital electronic circuits, integrated circuits, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementation by one or more computer programs executable on a programmable system. The programmable system comprises a storage system, at least one input device, and at least one programmable processor (which may be a special-purpose processor or a general-purpose processor) coupled to receive data and commands from at least one output device and to transmit data and commands to them. Computer programs (which are also known as programs, software, software applications, or code) include instructions for the programmable processor and are stored on a "computer-readable recording medium."

[0073] Computer-readable recording media include all types of recording devices in which data that can be read by a computer system is stored. Such computer-readable recording media may further include non-volatile or non-transitory media such as ROM, CD-ROM, magnetic tape, floppy disk, memory card, hard disk, magneto-optical disk, and storage device, or transitory media such as data transmission media. Additionally, computer-readable recording media may be distributed across networked computer systems, and computer-readable code may be stored and executed in a distributed manner.

[0074] Various embodiments of the methods described herein may be implemented by a programmable computer. Here, the computer includes a programmable processor, a data storage system (including volatile memory, non-volatile memory, or other types of storage systems, or a combination thereof), and at least one communication interface. For example, the programmable computer may be one of a server, a network device, a set-top box, an embedded device, a computer expansion module, a personal computer, a laptop, a PDA (Personal Data Assistant), a cloud computing system, or a mobile device.

[0075] The foregoing description is merely an illustrative explanation of the technical concept of the present invention, and those skilled in the art to which the present invention pertains will be able to make various modifications and variations within the scope of the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are intended to explain, not limit, the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by these embodiments. The scope of protection of the present invention shall be interpreted by the claims below, and all technical concepts within an equivalent scope shall be interpreted as being included within the scope of rights of the present invention. Explanation of the symbols

[0076] 100: Livestock manure composting facility 110: External sensor module 120: Internal sensor module 130: Environmental control device 140: Operating status measuring unit 200: Control unit 210: Input processing unit 220: Artificial Intelligence Analysis Department 221: Environmental Change Prediction Section 222: Estimated degree of ripeness section 223: Environmental Information Output Unit 224: Operation condition generation unit 230: Environmental control unit 240: Operating Status Analysis Unit

Claims

Claim 1 A livestock manure composting facility equipped with an internal sensor module and an external sensor module that measure internal state information and external environmental information, and an environmental control device that controls the internal environment, and which operates to compost livestock manure stored internally; and a control unit that analyzes the internal state information and external environmental information obtained from the internal sensor module and the external sensor module through an artificial intelligence model and controls the operation of the environmental control device according to the analysis results, wherein the control unit includes an input processing unit that collects the internal state information and external environmental information and converts them into time-series data; and an artificial intelligence analysis unit that analyzes the time-series data transmitted from the input processing unit through an artificial intelligence model to predict internal state information at a future point in time, estimates the degree of maturation of the current compost, and generates operating conditions for the environmental control device to increase the composting speed by considering the prediction and estimation results. The system includes an environmental control unit that controls the operation of the environmental control device according to operating conditions generated through the artificial intelligence analysis unit, wherein the internal state information includes at least one of temperature, humidity, oxygen concentration, acidity (pH), ammonia concentration, carbon dioxide concentration, carbon / nitrogen concentration ratio, and electrical conductivity for the internal space or compost of the livestock manure composting facility, and the artificial intelligence analysis unit includes an environmental change prediction unit comprising an LSTM model that analyzes time series data transmitted from the input processing unit to predict internal state information for a future point in time; a maturity degree estimation unit comprising an XGBoost model that estimates the current degree of maturity of the compost based on gas concentration among the internal state information; and an environmental information calculation unit comprising a DDPG reinforcement learning model that calculates target environmental information capable of shortening the composting time based on the prediction result of the environmental change prediction unit and the estimation result of the maturity degree estimation unit.An artificial intelligence-based high-speed composting system for livestock manure, comprising an operation condition generation unit that generates operation conditions for the environment control device according to target environment information calculated through the environment information calculation unit. Claim 2 delete Claim 3 delete Claim 4 In claim 1, the degree of maturation estimation unit estimates the degree of maturation of the compost by analyzing the change state regarding the amount of carbon dioxide and ammonia gas generated and the mutual generation ratio among the internal state information through the XGBoost model, an artificial intelligence-based high-speed composting system for livestock manure. Claim 5 An artificial intelligence-based high-speed livestock manure composting system according to claim 1, wherein the target environmental information calculated by the environmental information calculation unit includes information on temperature control values, moisture control values, and oxygen supply control values. Claim 6 In claim 1, the AI-based high-speed livestock manure composting system, wherein the DDPG reinforcement learning model is learned using a reward function calculated based on the ammonia / carbon dioxide ratio, the carbon / nitrogen ratio, and the time required for composting. Claim 7 In claim 1, the AI-based high-speed composting system for livestock manure, wherein the DDPG reinforcement learning model is retrained to reflect the fermentation characteristics of each type of livestock manure by updating a policy for determining the target environment information using time series data transmitted from the input processing unit for each type of livestock manure. Claim 8 In claim 1, the control unit further includes an operation status analysis unit that collects operation data of the environment control device, analyzes the operation status of the environment control device, and transmits the operation data and analysis results to the artificial intelligence analysis unit through the input processing unit, and the artificial intelligence analysis unit calculates an error between the actual operation status of the environment control device and the operation conditions based on the operation data and analysis results, and corrects the operation conditions by reflecting the calculated error, an artificial intelligence-based high-speed composting system for livestock manure.

Images