Aquaponics Integrated System

Abstract

This technology recovers waste heat from liquid-cooled server clusters and supplies it to hydroponic cultivation tanks and aquaculture tanks, thereby maintaining a stable temperature throughout the entire aquaponics system. [Solution] The aquaponics integrated system comprises a heat acquisition unit that receives waste heat from the coolant of a liquid-cooled server, a state acquisition unit that acquires computational load information of the server group and temperature information of the hydroponic cultivation tank and the aquaculture tank, a heat supply control unit that controls the flow rate of the heat exchanger valve and circulation pump to maintain the temperature of the circulating water within a predetermined range based on the acquired information, and an aquaponics circulation system that circulates the aquaculture tank and the hydroponic cultivation tank.

Description

Technical Field

[0001] The present invention relates to an aquaponics integrated system.

Background Art

[0002] In recent years, in AI processing servers, blockchain mining servers, and high-density information processing facilities including these, a large amount of heat is generated during operation, so forced cooling is required for stable operation. In particular, in a group of servers adopting a liquid cooling method, since heat continuously moves to the cooling medium, a large amount of waste heat is generated. Conventionally, this waste heat has generally been dissipated into the outside air through air-cooling equipment or cooling towers. However, there is a problem that a large amount of power is consumed at that time, and the waste heat itself is discarded into the environment without being effectively utilized.

[0003] On the other hand, in the agricultural field, in hydroponic cultivation facilities and plant factories, heat sources such as boilers and electric heaters are required to maintain a temperature environment suitable for the growth of crops. In the aquaculture industry, appropriate energy is also consumed for water temperature management of fish and the like. The heating energy in these agricultural and aquaculture fields depends on fossil fuels and commercial electricity, and an increase in environmental load and an increase in operation cost have become problems.

[0004] In recent years, from the viewpoints of countermeasures against global warming and sustainable food production, there is a demand for a technology that reduces the heating energy for agriculture and aquaculture and enables energy recycling by effectively utilizing industrial waste heat that has conventionally been discarded. An aquaponics system can efficiently utilize nutrient resources by mutually circulating hydroponic cultivation and aquaculture. However, conventionally, temperature control depending on the external environment has been a major problem, and there has been a problem that additional heating energy is required to maintain the temperature especially in winter and when sunlight is insufficient.

[0005] Against this backdrop, there is a need for technology that enables year-round operation of aquaponics facilities by supplying stable thermal energy, while also reducing external energy consumption. Patent Document 1 proposes an aquaponics system using a solar greenhouse, but natural energy is easily affected by weather conditions, making it difficult to ensure a stable heat supply. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Patent No. 7536145 [Overview of the project] [Problems that the invention aims to solve]

[0007] This invention has been made in view of the above problems, and aims to provide a technology that can stably maintain the temperature of the entire aquaponics system by actively recovering waste heat generated from liquid-cooled server groups such as AI processing servers and mining servers as a heat source and supplying it to hydroponic cultivation tanks and aquaculture tanks via a heat exchange unit. Furthermore, this invention aims to achieve zero emissions by circulating the server cooling water and the aquaponics water system, and by reusing excess heat and excess wastewater with thermoelectric conversion devices and filtration devices. [Means for solving the problem]

[0008] The main invention of the present invention for solving the above problems is an aquaponics integrated system comprising: a heat acquisition unit that receives waste heat from the coolant of a liquid-cooled server; a state acquisition unit that acquires computational load information of the server group and temperature information of hydroponic cultivation tanks and aquaculture tanks; a heat supply control unit that maintains the temperature of the circulating water within a predetermined range by controlling the flow rate of the valves and circulation pumps of the heat exchanger based on the acquired information; and an aquaponics circulation system that circulates and connects the aquaculture tank and the hydroponic cultivation tank.

[0009] Further issues and solutions disclosed in this application will be made clear in the section on embodiments of the invention and in the drawings. [Effects of the Invention]

[0010] According to the aquaponics integrated system of the present invention, waste heat from a liquid-cooled server is efficiently recovered in the heat acquisition unit, and the heat exchanger and circulation pump are automatically controlled based on the computational load and temperature information of each tank, thereby stably maintaining the temperature of the circulating water within a predetermined range. This allows for optimal maintenance of the environment in hydroponic cultivation tanks and aquaculture tanks without relying on external energy, improving the production efficiency and energy saving of the aquaponics circulation system. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows an example of the overall configuration of the aquaponics integrated system of this embodiment. [Figure 2] This figure shows an example of the hardware configuration of a control device. [Figure 3] This figure shows an example of the software configuration for a control device. [Figure 4] This diagram illustrates the processing flow in an integrated aquaponics system. [Modes for carrying out the invention]

[0012] <System Overview> The following describes an integrated aquaponics system according to one embodiment of the present invention. The integrated aquaponics system of this embodiment is a system that integrates hydroponics and aquaculture by effectively utilizing the waste heat discharged by a liquid-cooled server. This system supplies heat received from the server's cooling liquid to the hydroponic cultivation tank and aquaculture tank via a heat exchanger, maintaining the water temperature in these tanks within an appropriate range. Furthermore, by constructing an aquaponics circulation system that utilizes nutrients generated in the aquaculture tank in the hydroponic cultivation tank, water quality maintenance and wastewater reduction are achieved. In addition, by acquiring the server's computing load and temperature information of each tank in real time and dynamically controlling the flow rate of the heat exchanger valve and circulation pump, stable and efficient heat supply is achieved.

[0013] <System Configuration> Figure 1 shows an example of the overall configuration of the aquaponics integrated system of this embodiment. The aquaponics integrated system 100 of this embodiment consists of a liquid-cooled server group 10, a control device 20, a heat exchange unit 30, a hydroponic cultivation tank 40, aquaculture tank 50, a circulation pump 60, a valve group 70, and a sensor group 80. The control device 20 is connected to the liquid-cooled server group 10 and the sensor group 80 so as to be able to communicate via a communication network. The communication network is, for example, a local area network (LAN) and is constructed by wired or wireless means.

[0014] The liquid-cooled server group 10 consists of multiple servers that perform computational processing such as AI (Artificial Intelligence) processing and cryptocurrency mining. The liquid-cooled server group 10 removes the heat generated during computational processing using a coolant and sends the coolant to the heat exchange unit 30. The heat exchange unit 30 extracts heat from the high-temperature coolant received from the liquid-cooled server group 10 and heats the circulating water supplied to the hydroponic cultivation tank 40 and the aquaculture tank 50. The hydroponic cultivation tank 40 and the aquaculture tank 50 are connected by piping, forming an aquaponics circulation system in which water containing nutrients discharged from the aquaculture tank 50 is supplied to the hydroponic cultivation tank 40, and the water purified in the hydroponic cultivation tank 40 is returned to the aquaculture tank 50.

[0015] The circulation pump 60 is a pump that circulates water in the aquaponics circulation system, and its flow rate is adjusted based on commands from the control device 20. The valve group 70 consists of multiple valves installed in the hot water supply path from the heat exchange unit 30 to the hydroponic cultivation tank 40 and the aquaculture tank 50, and their opening degree is adjusted based on commands from the control device 20. The sensor group 80 consists of multiple sensors that measure the temperature, pH (hydrogen ion concentration), EC (electrical conductivity), DO (dissolved oxygen), etc., of the hydroponic cultivation tank 40 and the aquaculture tank 50.

[0016] The control device 20 acquires computational load information from the liquid-cooled server group 10 and measurement information from the sensor group 80, and controls the opening degree of the valve group 70 and the flow rate of the circulation pump 60 based on this information. The control device 20 may be a general-purpose computer such as a workstation or personal computer, or it may be logically implemented by cloud computing.

[0017] <Control device 20> FIG. 2 is a diagram showing an example of the hardware configuration of the control device 20. The illustrated configuration is an example, and it may have other configurations. The control device 20 includes a CPU 201, a memory 202, a storage device 203, a communication interface 204, an input device 205, and an output device 206. The storage device 203 stores various data and programs, such as a hard disk drive, a solid state drive, a flash memory, etc. The communication interface 204 is an interface for connecting to a communication network, such as an adapter for connecting to Ethernet (registered trademark), a wireless communication device for performing wireless communication, a USB (Universal Serial Bus) connector for serial communication, etc. The input device 205 inputs data, such as a keyboard, a mouse, a touch panel, buttons, etc. The output device 206 outputs data, such as a display, a printer, a speaker, etc. Each functional unit of the control device 20 described later is realized by the CPU 201 reading a program stored in the storage device 203 into the memory 202 and executing it, and each storage unit of the control device 20 is realized as a part of the storage area provided by the memory 202 and the storage device 203.

[0018] <Software Configuration> FIG. 3 is a diagram showing an example of the software configuration of the control device 20. The control device 20 includes a heat acquisition unit 211, a state acquisition unit 212, a heat supply control unit 213, an external data acquisition unit 214, a heat demand prediction unit 215, a model predictive control unit 216, an explanation generation unit 217, a digital twin unit 218, a load coordination unit 219, a continuous monitoring unit 221, a feedback learning unit 222, an improvement plan generation unit 223, an adaptive control unit 224, a water quality optimization unit 225, a preventive maintenance unit 226, a carbon accounting unit 227, a thermoelectric conversion unit 228, a microgrid management unit 229, an operation load information storage unit 231, a temperature information storage unit 232, a water quality information storage unit 233, an operation history storage unit 234, a model storage unit 235, and a set value storage unit 236.

[0019] <Control Device 20> Hereinafter, the functional units of the control device 20 will be described.

[0020] The operation load information storage unit 231 stores the operation load information acquired from the liquid-cooled server group 10. The operation load information includes the CPU usage rate, memory usage rate, number of jobs in execution, job types, job scheduled execution times, etc. of each server. The operation load information storage unit 231 accumulates the operation load information as time-series data.

[0021] The temperature information storage unit 232 stores the temperature information of the hydroponic cultivation tank 40 and the aquaculture tank 50 acquired from the sensor group 80. The temperature information includes the water temperature of each tank, the inlet and outlet temperatures of the heat exchange unit 30, the temperature of the coolant of the liquid-cooled server group 10, etc. The temperature information storage unit 232 accumulates the temperature information as time-series data.

[0022] The water quality information storage unit 233 stores the water quality information of the hydroponic cultivation tank 40 and the aquaculture tank 50 acquired from the sensor group 80. The water quality information includes pH, EC, DO, ammonia concentration, nitrate concentration, phosphate concentration, etc. The water quality information storage unit 233 accumulates the water quality information as time-series data.

[0023] The operation history storage unit 234 stores the operation history of the aquaponics integrated system 100. The operation history includes the opening degree history of the valve group 70, the flow rate history of the circulation pump 60, the heat supply amount to each tank, the energy balance, the types and growth status of cultivated crops, the types and growth status of cultured fish, the operation time of the equipment, etc. The operation history storage unit 234 accumulates the operation history as time-series data.

[0024] The model storage unit 235 stores various models used for control. The model storage unit 235 stores a machine learning model for estimating the heat dissipation amount from the operation load information, a prediction model for predicting the heat demand, a control model for optimizing control parameters, a water quality prediction model for predicting water quality fluctuations, a degradation prediction model for estimating the remaining life of the equipment, etc. These models are appropriately updated by the feedback learning unit 222.

[0025] The setting value storage unit 236 stores various setting values ​​used for control. The setting value storage unit 236 stores the target temperature range for the hydroponic cultivation tank 40 and the aquaculture tank 50, the target residence time for each tank, the flow rate distribution ratio between tanks, the control cycle, constraints, weighting coefficients, etc. These setting values ​​are updated as appropriate by the adaptive control unit 224.

[0026] The heat acquisition unit 211 manages the function of receiving waste heat from the coolant of the liquid-cooled server group 10. The heat acquisition unit 211 monitors the temperature and flow rate of the coolant sent from the liquid-cooled server group 10 to the heat exchange unit 30 and calculates the amount of heat that can be received. The heat acquisition unit 211 provides information on the amount of heat received to the heat supply control unit 213.

[0027] The heat acquisition unit 211 may include a multi-stage heat exchange unit. The multi-stage heat exchange unit has multiple heat exchange stages with different temperature settings. For example, the first stage can be set to a high temperature to supply heat to aquaculture fish species or cultivated crops that require high temperatures, the second stage can be set to a medium temperature to supply heat to fish species or crops that require medium temperatures, and the third stage can be set to a low temperature to supply heat to fish species or crops that require low temperatures. The heat acquisition unit 211 dynamically adjusts the heat distribution ratio to each stage according to the type of crop being cultivated and the type of fish being farmed. This makes it possible to provide a suitable temperature environment for each, even when cultivating or raising multiple varieties of crops and fish species simultaneously.

[0028] The status acquisition unit 212 acquires computational load information from the liquid-cooled server group 10, temperature information from the hydroponic cultivation tank 40 and the aquaculture tank 50, and water quality information. The status acquisition unit 212 communicates with the liquid-cooled server group 10 via a communication network and periodically acquires computational load information from each server. The status acquisition unit 212 also periodically acquires temperature information and water quality information from the sensor group 80. The status acquisition unit 212 stores the acquired information in the computational load information storage unit 231, the temperature information storage unit 232, and the water quality information storage unit 233, respectively.

[0029] The status acquisition unit 212 acquires pH, EC, and DO measurements from the water quality sensor group included in the sensor group 80. If the status acquisition unit 212 detects an abnormal value in the measurement, it corrects the measurement or outputs a warning. When the same physical quantity is measured from multiple sensors, the status acquisition unit 212 can use the average or median value of the measurement.

[0030] The heat supply control unit 213 maintains the temperature of the circulating water within a predetermined range by controlling the flow rate of the heat exchanger valves and circulation pump based on the computation load information and temperature information acquired by the state acquisition unit 212. The heat supply control unit 213 compares the current temperature of the hydroponic cultivation tank 40 and the aquaculture tank 50 with the target temperature range, and if the temperature is outside the target range, it adjusts the opening of the valve group 70 to increase or decrease the amount of heat supplied. The heat supply control unit 213 controls the amount of circulating water supplied to each tank by adjusting the flow rate of the circulation pump 60.

[0031] The heat supply control unit 213 can maintain the number of times the circulating water is replaced by optimizing the target residence time for each tank or the flow rate distribution between tanks. The target residence time is the time the circulating water remains in the hydroponic cultivation tank 40 or the aquaculture tank 50, and is set according to the type of crop or fish being cultivated. The heat supply control unit 213 reads the target residence time from the setting value storage unit 236 and controls the flow rate of the circulation pump 60 to achieve the target residence time. By adjusting the flow rate distribution ratio between tanks, the heat supply control unit 213 can control the rate of nutrient transfer between the hydroponic cultivation tank 40 and the aquaculture tank 50. This makes it possible to achieve both water quality stabilization and promotion of crop / fish growth.

[0032] The heat supply control unit 213 can estimate the target heat flux of the heat exchanger using a machine learning model that estimates the amount of heat dissipated from the computational load information of the liquid-cooled server group 10. The machine learning model is constructed using supervised learning algorithms such as neural networks, support vector machines, random forests, and gradient boosting. By learning the relationship between past computational load information and the measured amount of heat dissipated, the machine learning model can estimate the amount of heat dissipated from the computational load information with high accuracy. Based on the estimated amount of heat dissipated, the heat supply control unit 213 calculates the target heat flux and controls the opening degree of the valve group 70 to achieve the target heat flux. This allows the amount of heat supplied to be dynamically adjusted in response to fluctuations in the computational load of the servers.

[0033] The external data acquisition unit 214 acquires weather forecasts and electricity prices. The external data acquisition unit 214 accesses a weather information provider server via a communication network such as the Internet and acquires weather forecast data. The weather forecast data includes temperature, humidity, solar radiation, wind speed, probability of precipitation, etc. The external data acquisition unit 214 accesses a server that provides electricity market price information and acquires electricity price data. The electricity price data includes electricity unit prices by time of day, demand response signals, power generation forecasts, etc.

[0034] The external data acquisition unit 214 can acquire demand response signals or electricity market prices. Demand response signals are signals issued for supply and demand adjustment in the power grid and include requests for reductions or increases in electricity demand. When the external data acquisition unit 214 receives a demand response signal, it provides the signal to the model prediction control unit 216. The model prediction control unit 216 can simultaneously optimize the output target of the thermoelectric conversion unit 228 and the purchase and sale of electricity from the grid in response to the demand response signal.

[0035] The heat demand forecasting unit 215 estimates the future heat demand of the circulation system based on crop and fish species growth plans, external data, and computation load information. The heat demand forecasting unit 215 reads the crop and fish species growth plans from the set value storage unit 236. The growth plans include the type of crop to be cultivated, the start date of cultivation, the planned harvest date, the type of fish to be farmed, the input date, the planned shipping date, etc. The heat demand forecasting unit 215 identifies the required temperature range for each period from the growth plans.

[0036] The heat demand forecasting unit 215 uses weather forecast data acquired by the external data acquisition unit 214 to estimate the amount of heat loss due to fluctuations in outside temperature. The heat demand forecasting unit 215 predicts that if the outside temperature decreases, the amount of heat loss will increase, and therefore the required amount of heat supply will increase. The heat demand forecasting unit 215 reads past computation load information from the computation load information storage unit 231 and predicts the future computation load. The heat demand forecasting unit 215 estimates the amount of waste heat that can be used from the predicted computation load. The heat demand forecasting unit 215 compares the required amount of heat supply with the amount of waste heat that can be used and predicts whether there will be a surplus or shortage of heat supply.

[0037] The heat demand forecasting unit 215 can predict future heat demand using a time series forecasting model. Examples of time series forecasting models include the ARIMA (AutoRegressive Integrated Moving Average) model, state-space models, recurrent neural networks, and LSTM (Long Short-Term Memory). By learning past time series data of heat demand, the time series forecasting model can predict future heat demand with high accuracy.

[0038] The model prediction control unit 216 controls the valves and pumps by coordinating the optimization of heat exchange and water circulation in the time direction based on the predictions of the heat demand prediction unit 215. The model prediction control unit 216 solves an optimization problem to minimize the objective function, using the heat supply amount and water circulation flow rate at each time point within the prediction horizon as decision variables. The objective function is, for example, a weighted sum of the sum of squares of temperature deviations, energy consumption, and fluctuations in control inputs.

[0039] The model prediction control unit 216 solves the optimization problem while satisfying constraints. These constraints include the temperature range of each tank, the opening range of the valve group 70, the flow rate range of the circulation pump 60, and the allowable range of water quality parameters. The model prediction control unit 216 outputs the control input at the current time from the control input sequence obtained by solving the optimization problem to the valve group 70 and the circulation pump 60. In the next control cycle, the model prediction control unit 216 solves the optimization problem again and updates the control input. This allows for continuous optimal control while correcting the error between prediction and actual measurement.

[0040] The model prediction control unit 216 can simultaneously optimize the output target of the thermoelectric conversion unit 228 and the purchase and sale of electricity from the grid. The model prediction control unit 216 uses electricity price data acquired by the external data acquisition unit 214 to calculate the purchase cost and sales revenue of electricity at each time point. The model prediction control unit 216 defines an objective function that integrates heat supply and power supply, and solves an optimization problem with the heat exchanger valve, circulation pump flow rate, output of the thermoelectric conversion unit 228, amount of electricity purchased from the grid, and amount of electricity sold to the grid as decision variables. This makes it possible to achieve the necessary heat and power supply while minimizing energy costs.

[0041] The explanation generation unit 217 outputs an explanation of the reason and basis for the control. The explanation generation unit 217 acquires the content of the control performed by the model prediction control unit 216 and generates a text that explains the reason and basis for performing the control. For example, the explanation generation unit 217 generates an explanation such as, "Due to the decrease in outside temperature, heat loss increased, so the valve opening was increased by 10%." The explanation generation unit 217 outputs the generated explanation to the output device 206. This allows the operator to understand the intent of the control.

[0042] The explanation generation unit 217 can output a summary of the control policy and a basis score for the data sources used, using a generating AI (Artificial Intelligence). The generating AI is, for example, a natural language generation system using a Large Language Model (LLM). The explanation generation unit 217 inputs the data used for control (computation load information, temperature information, weather forecast data, etc.) into the generating AI and generates a summary of the control policy. The explanation generation unit 217 calculates a basis score indicating the reliability or importance of each data source and outputs it along with the summary. The basis score is calculated based on, for example, the freshness of the data, the accuracy of the sensor, the accuracy of the prediction model, etc. This allows the operator to quantitatively evaluate the basis for control decisions.

[0043] The digital twin unit 218 evaluates the results of virtual operation under assumed server load, weather, and cultivation conditions. The digital twin unit 218 simulates the operation of the system using a physical model or data-driven model of the aquaponics integrated system 100. The digital twin unit 218 uses weather forecast data acquired by the external data acquisition unit 214, predicted values ​​of the computation load stored in the computation load information storage unit 231, and cultivation conditions stored in the setting value storage unit 236 as inputs for the simulation. As simulation results, the digital twin unit 218 calculates temperature fluctuations, water quality fluctuations, energy consumption, crop and fish growth, etc. for each tank.

[0044] The digital twin unit 218 evaluates the simulation results and updates the constraints of the model predictive control unit 216 based on the evaluation results. If the simulation results do not meet predetermined criteria, the digital twin unit 218 relaxes or strengthens the constraints. For example, if the water quality deviates from the criteria in the simulation results, the digital twin unit 218 strengthens the constraints on the water quality parameters. The digital twin unit 218 provides the updated constraints to the model predictive control unit 216. This allows for risk assessment before actual operation and improves the safety of the control system.

[0045] The load coordination unit 219 works in conjunction with the job scheduler of the liquid-cooled server group 10 to shape the waste heat supply by adjusting the allocation of computing load or execution time according to the heat demand. Here, "shaping the waste heat supply" means adjusting the allocation of computing load or execution time so that the amount of waste heat supplied increases during periods of high heat demand, thereby adapting the temporal pattern of waste heat supply to the demand side. The load coordination unit 219 obtains future heat demand forecasts from the heat demand forecasting unit 215. The load coordination unit 219 requests the job scheduler to concentrate the computing load during periods of high heat demand in order to increase the amount of waste heat supplied during those periods. The load coordination unit 219 communicates with the job scheduler to make adjustments such as shifting the execution time of jobs or distributing jobs across multiple servers.

[0046] The load coordination unit 219 adjusts the computation load to minimize the mismatch between heat demand and waste heat supply. The load coordination unit 219 reduces the computation load during periods of low heat demand and increases it during periods of high heat demand. The load coordination unit 219 obtains the adjustable range of the computation load from the job scheduler and determines the optimal computation load distribution within that range. This enables efficient utilization of waste heat.

[0047] The continuous monitoring unit 221 continuously monitors and stores time-series data on computation load information, temperature information, water quality information, energy balance, and equipment status. The continuous monitoring unit 221 records information acquired from the status acquisition unit 212 in the operation history storage unit 234. The continuous monitoring unit 221 monitors the opening degree of the valve group 70, the flow rate of the circulation pump 60, the heat flux of the heat exchange unit 30, the output of the thermoelectric conversion unit 228, etc., and records these in the operation history storage unit 234. The continuous monitoring unit 221 also monitors equipment status information such as operating time, vibration, pressure, and temperature, and records this information in the operation history storage unit 234.

[0048] The continuous monitoring unit 221 outputs a warning if it detects an abnormality in the monitored data. The continuous monitoring unit 221 determines that an abnormality exists if parameters such as temperature, water quality, and vibration exceed a predetermined threshold or fall outside a predetermined range. When the continuous monitoring unit 221 detects an abnormality, it displays a warning on the output device 206 or sends a notification to the administrator.

[0049] The feedback learning unit 222 updates the prediction model and control model online based on the data stored in the operation history storage unit 234. The feedback learning unit 222 reads past computation load information, temperature information, water quality information, control inputs, energy balance, etc. from the operation history storage unit 234 and uses them as learning data. The feedback learning unit 222 executes a machine learning algorithm using the learning data and updates the parameters of the prediction model and control model.

[0050] The feedback learning unit 222 updates a machine learning model that estimates heat dissipation from computational load information, for example. The feedback learning unit 222 calculates the error between the measured heat dissipation and the predicted value, and updates the model parameters to minimize this error. The feedback learning unit 222 can update the model using online learning algorithms, such as stochastic gradient descent, recurrent learning, or reinforcement learning. The feedback learning unit 222 stores the updated model in the model storage unit 235.

[0051] The feedback learning unit 222 can adaptively update the model in response to environmental changes such as seasonal fluctuations, equipment deterioration, and changes in crop and fish species. The feedback learning unit 222 analyzes long-term operating data and learns seasonal heat demand patterns, trends in equipment performance deterioration, etc. By reflecting the learned trends in the prediction model, the feedback learning unit 222 can improve prediction accuracy.

[0052] The improvement plan generation unit 223 generates an operation plan or equipment upgrade proposal, taking into account seasonal fluctuations and equipment deterioration. The improvement plan generation unit 223 acquires the seasonal fluctuation patterns and equipment deterioration trends learned by the feedback learning unit 222. The improvement plan generation unit 223 predicts future heat demand and equipment performance and formulates an optimal operation plan. The operation plan includes setting target temperatures for each season, setting circulation flow rates, and cultivation plans for crops and fish species.

[0053] The improvement plan generation unit 223 generates an equipment replacement proposal when the remaining lifespan of the equipment falls below a predetermined threshold. The equipment replacement proposal includes the parts that need to be replaced, the timing of the replacement, the performance improvement effect of the replacement, and the cost. The improvement plan generation unit 223 outputs the generated operation plan or equipment replacement proposal to the output device 206. This allows operators or managers to consider future operation policies in advance.

[0054] The adaptive control unit 224 adaptively updates the control of the valve and pump by reflecting the learning results of the feedback learning unit 222 and the improvement plan generated by the improvement plan generation unit 223 in the control parameters. The adaptive control unit 224 reads the updated prediction model and control model from the model storage unit 235 and calculates the control input using the model. The adaptive control unit 224 updates the control parameters stored in the setting value storage unit 236.

[0055] The adaptive control unit 224 can plan and execute setting value search experiments using safety-constrained Bayesian optimization and update the setting values ​​based on measured performance metrics. Safety-constrained Bayesian optimization is a method for searching for setting values ​​that maximize or minimize the objective function while satisfying constraints. The adaptive control unit 224 estimates the posterior distribution of the objective function using a probabilistic model such as a Gaussian process and selects the setting value that maximizes the acquisition function as the next search point. The adaptive control unit 224 executes experiments using the selected setting value and measures the performance metrics. The adaptive control unit 224 updates the probabilistic model using the measurement results and determines the next search point. By repeating this process, the optimal setting value can be discovered while ensuring safety.

[0056] The water quality optimization unit 225 adjusts the aeration rate or nutrient supply rate to compensate for temperature-induced water quality fluctuations. The water quality optimization unit 225 reads temperature information and water quality information from the temperature information storage unit 232 and the water quality information storage unit 233. The water quality optimization unit 225 analyzes the correlation between temperature and pH, EC, and DO to predict water quality fluctuations due to temperature changes. The water quality optimization unit 225 adjusts the aeration rate of the aeration system and the supply rate of the nutrient supply system to compensate for the predicted water quality fluctuations.

[0057] Figure 4 is a diagram illustrating the processing flow in the aquaponics integrated system 100.

[0058] The status acquisition unit 212 acquires computation load information from the liquid-cooled server group 10 (S301), and acquires temperature information and water quality information from the sensor group 80 for the hydroponic cultivation tank 40 and the aquaculture tank 50 (S302). The status acquisition unit 212 stores the acquired information in the computation load information storage unit 231, the temperature information storage unit 232, and the water quality information storage unit 233 (S303). The heat acquisition unit 211 receives waste heat from the coolant of the liquid-cooled server group 10 and calculates the amount of heat that can be received (S304).

[0059] The heat supply control unit 213 reads computation load information and temperature information from the computation load information storage unit 231 and the temperature information storage unit 232 (S305). The heat supply control unit 213 compares the current temperature of the hydroponic cultivation tank 40 and the aquaculture tank 50 with the target temperature range and calculates the temperature deviation (S306). Based on the temperature deviation and computation load information, the heat supply control unit 213 controls the opening of the valve group 70 of the heat exchange unit 30 and the flow rate of the circulation pump 60 to maintain the temperature of the circulating water within a predetermined range (S307).

[0060] The continuous monitoring unit 221 continuously monitors temperature information, water quality information, energy balance, and equipment status after control execution and stores them in the operation history storage unit 234 (S308). The feedback learning unit 222 updates the prediction model and control model online based on the data stored in the operation history storage unit 234 and stores the updated model in the model storage unit 235 (S309). The adaptive control unit 224 adaptively updates the control parameters using the updated model and set values ​​and reflects them in the control in the next control cycle (S310).

[0061] As described above, the aquaponics integrated system of this embodiment improves the cooling efficiency of the liquid-cooled server and reduces heating costs for agriculture and aquaculture by supplying waste heat from the server to the hydroponic cultivation tank and aquaculture tank. Furthermore, by dynamically controlling the flow rate of the heat exchanger valves and circulation pumps based on the server's computing load information and the temperature information of each tank, stable and efficient heat supply can be achieved. In addition, by utilizing the nutrients generated in the aquaculture tank through the aquaponics circulation system in the hydroponic cultivation tank, water quality can be maintained and wastewater can be reduced, thereby reducing the environmental burden. Moreover, by performing heat dissipation estimation using a machine learning model, cooperative optimization through model predictive control, and adaptive model updates through feedback learning, long-term optimal operation that responds to seasonal fluctuations and equipment deterioration can be achieved.

[0062] Although these embodiments have been described above, they are intended to facilitate understanding of the present invention and are not intended to limit its interpretation. The present invention can be modified and improved without departing from its spirit, and equivalents thereof are also included.

[0063] For example, the processing performed by each functional unit of the control device 20 described above may be executed by any of the functional units. Furthermore, different functional units may be added to perform some of the processing performed by each of the functional units described above. Also, the functional units of the control device 20 may be distributed across multiple computers. For example, the heat demand forecasting unit 215 and the model forecasting control unit 216 may be placed on a cloud server, while the other functional units are placed on an on-premise control device.

[0064] Furthermore, the information stored in each memory unit of the control device 20 may be stored in any of the memory units. That is, the information stored in the multiple memory units described above may be stored in a single memory unit, or some of the information stored in one memory unit may be stored in another memory unit. For example, the computation load information storage unit 231, the temperature information storage unit 232, and the water quality information storage unit 233 may be implemented as an integrated time-series database.

[0065] <Example 1> In the embodiment described above, the heat acquisition unit 211 receives waste heat from the coolant of the liquid-cooled server group 10, but the waste heat from the air-cooled servers may be recovered by a heat exchanger. In this case, the heat acquisition unit 211 monitors the exhaust temperature and airflow rate of the server room and transfers the waste heat to the circulating water using an air-water heat exchanger. The heat acquisition unit 211 can adjust the amount of heat received by controlling the rotation speed of the heat exchanger fan. Alternatively, a configuration may be used in combination with liquid-cooled servers, recovering waste heat from each. In this case, the heat acquisition unit 211 calculates the total amount of heat received by summing the waste heat from each heat source and provides it to the heat supply control unit 213.

[0066] <Modification 2> In the embodiment described above, the heat supply control unit 213 estimates the amount of heat dissipated from computation load information using a machine learning model, but it may also estimate the amount of heat dissipation using a physical model. The physical model is a model that calculates the amount of heat dissipated by thermodynamic calculations from the server's CPU usage, memory usage, power consumption, etc. The heat supply control unit 213 refers to the specification information of each server (TDP (Thermal Design Power), cooling efficiency, etc.) and calculates the amount of heat generated according to the current operating state. The heat supply control unit 213 estimates the amount of heat transferred to the coolant from the calculated amount of heat generated and calculates the amount of heat that can be received. A hybrid model combining a physical model and a machine learning model may also be used. In this case, the heat supply control unit 213 can use the estimated value from the physical model as an input feature of the machine learning model to improve the estimation accuracy.

[0067] <Variation 3> In the embodiment described above, the aquaponics circulation system is configured to circulate between the aquaculture tank 50 and the hydroponic cultivation tank 40, but multiple aquaculture tanks and multiple hydroponic cultivation tanks may be connected in a multi-stage configuration. For example, the water discharged from the first aquaculture tank may be purified in the first hydroponic cultivation tank, the water discharged from the first hydroponic cultivation tank may be supplied to the second aquaculture tank, the water discharged from the second aquaculture tank may be purified in the second hydroponic cultivation tank, and the water discharged from the second hydroponic cultivation tank may be returned to the first aquaculture tank. The heat supply control unit 213 maintains the temperature of each tank within an optimal range by individually controlling the amount of heat supplied to each tank. Furthermore, when different types of crops or fish are cultivated in each tank, the multi-stage heat exchange unit of the heat acquisition unit 211 can be used to supply hot water at a suitable temperature to each tank.

[0068] <Modification 4> In the embodiment described above, the explanation generation unit 217 outputs a summary of the control policy and a justification score using the generated AI. However, it may also generate and output a flowchart or graph that visualizes the control decision-making process. The explanation generation unit 217 acquires multiple control candidates evaluated by the model prediction control unit 216 during the process of solving the optimization problem and displays the evaluation value of each candidate in a graph. The explanation generation unit 217 highlights the control candidate that was ultimately selected and the reason for its selection. The explanation generation unit 217 may also display the contribution of major factors that influenced the control decision (such as ambient temperature, computation load, and water quality parameters) in a bar graph or pie chart. This allows the operator to intuitively understand the control decision-making process.

[0069] <Modification 5> In the embodiment described above, the thermoelectric conversion unit 228 converts unused waste heat into electricity, but an absorption chiller that uses waste heat for cooling or dehumidification may also be provided. An absorption chiller is a device that generates chilled water using waste heat as a driving source. In the summer, when the outside temperature is high and cooling of the hydroponic cultivation tank 40 or aquaculture tank 50 is necessary, the control device 20 operates the absorption chiller to generate chilled water and supplies the chilled water to each tank. The control device 20 switches between the hot water supply path and the chilled water supply path by controlling the valve group 70 of the heat exchange unit 30. In addition, dehumidification may be performed using the absorption chiller to maintain the humidity of the hydroponic cultivation tank 40 within an appropriate range. This makes it possible to effectively utilize waste heat throughout the year.

[0070] <Disclosure Items> Furthermore, this disclosure also includes the following configurations. [Item 1] A heat acquisition unit that receives waste heat from the coolant of a liquid-cooled server, A status acquisition unit that acquires computing load information of the server group and temperature information of hydroponic cultivation tanks and aquaculture tanks, Based on the acquired information, a heat supply control unit controls the flow rate of the heat exchanger valves and circulation pump to maintain the temperature of the circulating water within a predetermined range, An aquaponics circulation system that circulates and connects aquaculture tanks and hydroponic cultivation tanks, An integrated aquaponics system equipped with [features / equipment]. [Item 2] The aquaponics integrated system according to item 1, wherein the heat supply control unit optimizes the target residence time of each tank or the flow rate distribution between tanks to maintain the number of times the circulating water is replaced. [Item 3] The heat supply control unit estimates the target heat flux of the heat exchanger using a machine learning model that estimates the amount of heat dissipated from the computational load information of the server group, as described in item 1 of the aquaponics integrated system. [Item 4] An aquaponics integrated system as described in item 1, comprising a thermoelectric conversion unit that converts unused waste heat into electricity to power pumps, lighting, and sensors. [Item 5] The aquaponics integrated system according to item 1, wherein the heat acquisition unit includes a multi-stage heat exchange unit having multiple heat exchange stages with different temperature settings, and controlling heat distribution for each stage according to the crop species and fish species. [Item 6] The aquaponics integrated system described in item 1, comprising a group of water quality sensors for measuring pH, EC, and DO, and a water quality optimization unit that adjusts the aeration rate or nutrient supply rate to compensate for temperature-induced water quality fluctuations. [Item 7] The process of receiving waste heat from the coolant of a liquid-cooled server, A process for acquiring computing load information for the server group and temperature information for each tank, A step of controlling the flow rate of the heat exchanger valves and circulation pump based on the acquired information to maintain the temperature of the circulating water within a predetermined range, A computer-based aquaponics integration method. [Item 8] The process of receiving waste heat from the coolant of a liquid-cooled server, A process for acquiring computing load information for the server group and temperature information for each tank, A step of controlling the flow rate of the heat exchanger valves and circulation pump based on the acquired information to maintain the temperature of the circulating water within a predetermined range, A program that causes a computer to execute something. [Item 9] A heat acquisition unit that receives waste heat from the coolant of a liquid-cooled server, A status acquisition unit that acquires computing load information of the server group and temperature information of each tank, External data acquisition unit for obtaining weather forecasts and electricity prices, A heat demand forecasting unit that estimates the future heat demand of the circulation system based on the growth plan of crops and fish species, the external data, and the computation load information, A model prediction control unit controls the valve and pump by coordinating the optimization of heat exchange and water circulation in the time direction based on the above prediction, An explanation generation unit that outputs an explanation of the reason and basis for the control, An integrated aquaponics system equipped with [features / equipment]. [Item 10] The aquaponics integrated system described in item 9 outputs a summary of the control policy and a basis score for the data sources used, using a generating AI. [Item 11] The aquaponics integrated system described in item 9, comprising a digital twin unit that evaluates the results of virtual operation for assumed server load, weather, and cultivation conditions, and updates the constraints of the model prediction control unit based on the evaluation results. [Item 12] The aquaponics integrated system described in item 9, comprising a load coordination unit that works in conjunction with the server group's job scheduler to adjust the distribution or execution time of the computing load according to the heat demand and shape the waste heat supply. [Item 13] The aquaponics integrated system described in item 9, wherein the external data acquisition unit acquires demand response signals or electricity market prices, and the model prediction control unit simultaneously optimizes the output target of the thermoelectric conversion unit and the purchase and sale of electricity from the grid. [Item 14] The process of obtaining weather forecasts and electricity prices, A process for predicting the future heat demand of a circulating system based on the growth plans of crops and fish species, the aforementioned external data, and the computational load information of the server group, A process of controlling valves and pumps by coordinating and optimizing heat exchange and water circulation in the time direction based on the said prediction, A process to output an explanation of the reason and basis for the control, A computer-based aquaponics integration method. [Item 15] The process of obtaining weather forecasts and electricity prices, A process for predicting the future heat demand of a circulating system based on the growth plans of crops and fish species, the aforementioned external data, and the computational load information of the server group, A process of controlling valves and pumps by coordinating and optimizing heat exchange and water circulation in the time direction based on the said prediction, A process to output an explanation of the reason and basis for the control, A program that causes a computer to execute something. [Item 16] A heat acquisition unit that receives waste heat from the coolant of a liquid-cooled server, A status acquisition unit that acquires computing load information of the server group, temperature information and water quality information for each tank, A continuous monitoring unit that continuously monitors and stores the time-series data of the acquired information, energy balance, and equipment status, A feedback learning unit updates the prediction model and control model online based on the aforementioned accumulated data, An improvement plan generation unit that generates an operation plan or equipment upgrade proposal in anticipation of seasonal fluctuations and equipment deterioration, An adaptive control unit that reflects the learning results and improvement plan in the control parameters and adaptively updates the control of the valve and pump, An integrated aquaponics system equipped with [features / equipment]. [Item 17] The adaptive control unit plans and executes setpoint exploration experiments using safety-constrained Bayesian optimization and updates the setpoints based on measured performance indicators, as described in item 16 of the aquaponics integrated system. [Item 18] An aquaponics integrated system as described in item 16, comprising a preventive maintenance unit that estimates the remaining lifespan based on the vibration and pressure difference of the pump and the temperature difference of the heat exchanger and indicates the maintenance timing. [Item 19] The aquaponics integrated system described in item 16, comprising a carbon accounting unit that estimates the reduction in greenhouse gas emissions from waste heat reuse by comparing it with a baseline operation and uses the estimation result as a control objective function or constraint. [Item 20] An aquaponics integrated system as described in item 16, which includes a thermoelectric conversion unit and a microgrid management unit including a battery, which works in conjunction with a virtual power plant when connected to the grid and switches to independent operation in the event of a power outage. [Item 21] The process involves continuously monitoring and accumulating operational data, water quality data, energy balance, and equipment status. A process of updating the prediction model and control model online based on the aforementioned accumulated data, A process of adaptively updating control parameters based on update results to control the valve and pump, A process for generating an operation plan or equipment upgrade proposal that takes into account seasonal fluctuations and equipment deterioration, A computer-based aquaponics integration method. [Item 22] The process involves continuously monitoring and accumulating operational data, water quality data, energy balance, and equipment status. A process of updating the prediction model and control model online based on the aforementioned accumulated data, A process of adaptively updating control parameters based on update results to control the valve and pump, A process for generating an operation plan or equipment upgrade proposal that takes into account seasonal fluctuations and equipment deterioration, A program that causes a computer to execute something. [Explanation of Symbols]

[0071] 100 Aquaponics Integrated Systems 10 Liquid-cooled server clusters 20 Control device 30 Heat exchange unit 40 hydroponic growing tanks 50 Aquaculture tank 60 Circulation pump 70 valve group 80 sensor group

Claims

1. A group of liquid-cooled servers, A heat exchange unit that receives waste heat from the coolant of the aforementioned liquid-cooled server group, A hydroponic cultivation tank and an aquaculture tank to which circulating water heated by the heat received from the heat exchange unit is supplied, A circulation pump and valve group are provided in the piping that circulates and connects the hydroponic cultivation tank and the aquaculture tank, A group of sensors for measuring the temperature of the hydroponic cultivation tank and the aquaculture tank, Control device and an aquaponics integrated system comprising, The control device is A heat acquisition unit that monitors the temperature and flow rate of the cooling liquid in the aforementioned liquid-cooled server group and calculates the amount of heat that can be received, A state acquisition unit that acquires computational load information of the server group and temperature information of the hydroponic cultivation tank and the aquaculture tank, A heat supply control unit that controls the opening degree of the valve group and the flow rate of the circulation pump to maintain the temperature of the circulating water within a predetermined range, based on the acquired computation load information and temperature information, An aquaponics integrated system characterized by having the following features.

2. The aquaponics integrated system according to claim 1, wherein the heat supply control unit optimizes the target residence time of circulating water in the hydroponic cultivation tank and the aquaculture tank or the flow rate distribution between tanks to maintain the number of times the circulating water is replaced.

3. The aquaponics integrated system according to claim 1, wherein the heat supply control unit estimates the target heat flux of the heat exchange unit using a machine learning model that estimates the amount of heat dissipated from the computational load information of the server group.

4. The aquaponics integrated system according to claim 1, further comprising a thermoelectric conversion unit that converts unused waste heat not recovered in the heat exchange unit into electricity.

5. The aquaponics integrated system according to claim 1, wherein the heat exchange unit includes a multi-stage heat exchange unit having multiple heat exchange stages with different temperature settings and controlling heat distribution for each stage according to the crop species and fish species.

6. The sensor group includes water quality sensors that measure pH, EC and DO, The aquaponics integrated system according to claim 1, wherein the control device further comprises a water quality optimization unit that adjusts the aeration rate or nutrient supply rate in order to correct temperature-induced water quality fluctuations.

7. A group of liquid-cooled servers, A heat exchange unit that receives waste heat from the coolant of the aforementioned liquid-cooled server group, A hydroponic cultivation tank and an aquaculture tank to which circulating water heated by the heat received from the heat exchange unit is supplied, A circulation pump and valve group are provided in the piping that circulates and connects the hydroponic cultivation tank and the aquaculture tank, A group of sensors for measuring the temperature of the hydroponic cultivation tank and the aquaculture tank, Control device and an aquaponics integrated system comprising, The control device is A heat acquisition unit that monitors the temperature and flow rate of the cooling liquid in the aforementioned liquid-cooled server group and calculates the amount of heat that can be received, A state acquisition unit that acquires computational load information of the server group and temperature information of the hydroponic cultivation tank and the aquaculture tank, An external data acquisition unit that acquires external data including weather forecasts and electricity prices, A heat demand forecasting unit that predicts the future heat demand of the circulation system based on the growth plan of crops and fish species, the external data, and the computation load information, A model prediction control unit controls the heat exchange unit, the valve group, and the circulation pump in a coordinated optimization manner over time based on the prediction made by the heat demand forecasting unit, An explanation generation unit that outputs an explanation of the reason and basis for the control, An aquaponics integrated system characterized by having the following features.

8. The aquaponics integrated system according to claim 7, wherein the explanation generation unit outputs a summary of the control policy and the basis score of the data source used, using generating AI.

9. The control device further comprises a digital twin unit that evaluates the results of virtual operation for assumed server load, weather and cultivation conditions, The aquaponics integrated system according to claim 7, wherein the digital twin unit updates the constraints of the model prediction control unit based on the results of the evaluation.

10. The aquaponics integrated system according to claim 7, wherein the control device further comprises a load coordination unit that works in cooperation with the job scheduler of the server group to adjust the distribution of computation load or execution time according to the heat demand to shape the waste heat supply.

11. Further comprising a thermoelectric conversion unit that converts unused waste heat into electricity, The aquaponics integrated system according to claim 7, wherein the external data acquisition unit acquires a demand response signal or electricity market price, and the model prediction control unit simultaneously optimizes the output target of the thermoelectric conversion unit and the purchase and sale of electricity from the grid.

12. A group of liquid-cooled servers, A heat exchange unit that receives waste heat from the coolant of the aforementioned liquid-cooled server group, A hydroponic cultivation tank and an aquaculture tank to which circulating water heated by the heat received from the heat exchange unit is supplied, A circulation pump and valve group are provided in the piping that circulates and connects the hydroponic cultivation tank and the aquaculture tank, A group of sensors for measuring the temperature of the hydroponic cultivation tank and the aquaculture tank, Control device and an aquaponics integrated system comprising, The control device is A heat acquisition unit that monitors the temperature and flow rate of the cooling liquid in the aforementioned liquid-cooled server group and calculates the amount of heat that can be received, A status acquisition unit that acquires computational load information of the server group, temperature information and water quality information of the hydroponic cultivation tank and the aquaculture tank, A continuous monitoring unit continuously monitors and stores the information acquired by the status acquisition unit, the energy balance of the aquaponics integrated system, and the time-series status of the equipment including the heat exchange unit, the circulation pump, and the valve group. Based on the data accumulated by the continuous monitoring unit, a feedback learning unit updates online a prediction model for predicting temperature fluctuations of the circulating water and a control model for determining the opening degree of the valve group and the flow rate of the circulation pump. An improvement plan generation unit that generates an operation plan or equipment upgrade proposal for the aquaponics integrated system in anticipation of seasonal fluctuations and deterioration of the equipment, An adaptive control unit that reflects the learning results from the feedback learning unit and the improvement plan generated by the improvement plan generation unit in the control parameters and adaptively updates the control of the valve group and the circulation pump, An aquaponics integrated system characterized by having the following features.

Images