TWI655902B - Method and system for managing information of fish tank - Google Patents

Abstract

The invention provides a smart fish tank information management system, which comprises a network server, a sensor, a breeding device and a controller. The controller is used to obtain the breeding process through the sensor and transfer the breeding process to the network server. The network server calculates the equipment parameters corresponding to the farming equipment according to the farming history. The controller takes the equipment parameters and controls the corresponding farming equipment according to the equipment parameters.

Description

Smart fish tank information management system and method

The invention relates to a culture fish tank, and in particular to a smart fish tank information management system and method.

Traditional fish farmers rely on experience and are raised by the rules of thumb. For example, if you look at the color of the water with the naked eye to decide whether to change water or gas, etc., this method is easy to cause huge losses due to lack of experience, even if the color of the water is consistent. Ensure that the pH data in the water will be consistent. If there are new generations of young people to invest in the aquaculture industry, there will be huge bottlenecks in the cultivation, which will easily lead to the loss of technology.

Embodiments of the present invention provide a smart fish tank information management system including a network server, a sensor, a breeding device, and a controller. The controller is used to obtain the breeding process through the sensor and transfer the breeding process to the network server. The network server calculates the equipment parameters corresponding to the farming equipment according to the farming history. The controller takes the equipment parameters and controls the corresponding farming equipment according to the equipment parameters.

In some embodiments, the sensor includes a water quality detector, a thermometer, a light sensor, and an image capture module. Farming equipment includes air pumps, motors, filters, lamps, water quality processors, filtration systems and temperature control systems.

In some embodiments, the controller acquires multiple images at multiple points in time through the image capture module and detects the ornamental fish in each image. The controller calculates the number of aquarium fish in each image to obtain the number of fish, and calculates the number of these fish to obtain the estimated number of fish.

In some embodiments, the network server sets the breeding period and the sampling period, and the breeding period is greater than the sampling period. The state set S is established according to the value detected by the sensor, the sampling period and the breeding period, and the action set is established according to the device parameters. A, establishing a reward set R according to the estimated number of fish, establishing a first action value function q ₁ ( s, a ) and a second action value function q ₂ ( s, a ), wherein s S , a A. The network server determines one of the following equations (1), (2) according to the first probability to determine the action a _t corresponding to the current state s, and determines the device parameter according to the action a _t .

a _t =max _a q ₁ ( s,a )...(1)

a _t =max _a q ₂ ( s,a )...(2).

In some embodiments, the controller performs action a _t and calculates the current reward r and state s ' , where r R , s ' S. The web server performs one of the following equations (3), (4) with a second probability.

q ₂ ( s,a )← Where α , β , and γ are real numbers, and N _t ( s, a ₁ ) is the number of times a ₁ is selected under the state s.

In some embodiments, if the estimated number of fish is reduced, the controller sets the reward r to a first value, and if the estimated number of fish is constant, the controller sets the reward r to a second value, and if the estimated number of fish increases, the controller The set reward is a third value, wherein the third value is greater than the second value and the second value is greater than the first value.

In another aspect, an embodiment of the present invention provides a smart fish tank information management method for a smart fish tank information management system, the smart fish tank information management system comprising a plurality of sensors and a plurality of breeding equipment. The smart fish tank information management method includes: obtaining a breeding process through a sensor; calculating a plurality of equipment parameters corresponding to the breeding equipment according to the breeding history; and controlling the corresponding breeding equipment according to the equipment parameters.

According to the above system and method, the device parameters can be automatically determined, so that the user can culture the ornamental fish more effectively or more easily.

The above described features and advantages of the invention will be apparent from the following description.

100‧‧‧Smart Aquarium Information Management System

110‧‧‧fish tank

111‧‧‧Sensor

112‧‧‧ farming equipment

120‧‧‧ Controller

130‧‧‧Web server

201~203‧‧‧Steps

FIG. 1 is a schematic diagram showing a smart fish tank information management system according to an embodiment.

FIG. 2 is a flow chart showing a smart fish tank information management method according to an embodiment.

The terms "first", "second", "etc." used in this document are not intended to mean the order or the order, and are merely to distinguish between elements or operations described in the same technical terms.

FIG. 1 is a schematic diagram showing a smart fish tank information management system according to an embodiment. The smart aquarium information management system 100 includes a fish tank 110, a sensor 111, a farming device 112, a controller 120, and a network server 130. The sensor 111 may include a water quality detector, a thermometer, a light sensor, an image capturing module, and the like. The breeding equipment 112 can include a gas pump, a motor, a filter, a luminaire, a water quality processor, a filtration system, a temperature control system, and the like. However, the sensor 111 and the breeding device 112 are merely exemplary, and the present invention is not limited thereto, and the present invention does not limit the number of the sensor 111 and the breeding device 112. The controller 120 is, for example, any suitable circuit such as a central processing unit, a microprocessor, a microcontroller, a digital signal processor, a special application integrated circuit, or the like.

The values detected by the sensor 111 are collectively referred to as a farming history, and the controller 120 obtains the breeding history and transmits the farming history to the web server 130. In some embodiments, the farming history may also include the type of fish, the number of fish, the size of the aquarium 110, and the like. The network server 130 calculates device parameters corresponding to the breeding equipment 112 based on these farming schedules, which are used to control the farming equipment 112. For example, the control parameters can be used to control the speed of the motor, control the light intensity of the luminaire, and control The temperature, as well as controlling whether the filtration system is turned on, etc., are not limited thereto. The controller 120 can retrieve these device parameters from the network server 130 and control the corresponding breeding device 112 based on the acquired device parameters. In some embodiments, different culture courses and equipment parameters can be established for different ornamental fish and shrimp. In this way, when the user is not familiar with the culture of the fish, the corresponding device parameters can be obtained directly from the network server 130.

In some embodiments, the network server 130 can also provide a network service, and the user can use the network service through the software on the personal computer, the application on the mobile phone, or the webpage, thereby viewing the culture. History and equipment parameters, and even control equipment parameters. In some embodiments, based on the data collected by the web server 130, it is also possible to create a unique production history, including categories, characteristics, life status, etc., to create a unique brand, avoiding the variety of fish from other countries.

The following describes how to calculate device parameters. The information management method described below can be performed by the network server 130, but can also be performed by the controller 120 in some embodiments, and the present invention is not limited thereto. First, the culture cycle and the sampling cycle must be set first. The culture cycle is, for example, three days, one week, one month, or any other time. The sampling cycle is shorter than the culture cycle, and can be set to 3 hours, 6 hours, and the like. A set of states S can be established based on the values detected by the sensor 111, the breeding cycle, and the sampling period. This set of states S contains a number of different states. For example, different temperatures, different water quality, or different brightness are considered different states. In addition, every other sampling period is considered to be a different state, and every other culture cycle is repeated once, for example, every 6 weeks. Hours are in different states, but next week The first 6 hours and the first 6 hours of the previous week can belong to the same state. In some embodiments, since the number of possible states is too large, after collecting some data, a k-mean algorithm is performed to classify the values detected by the sensors 111 with time. All categories form the above-mentioned set of states S.

Next, an action set A is created based on the device parameters. This action set contains a number of actions, each of which has corresponding device parameters to control the corresponding breeding device 112. For example, each action can include temperature, light, and motor control parameters.

In addition, a reward set R must be established. In some embodiments, the survival of the fish can be used to determine the reward, but since it is not easy to track the status of each fish, the number of all fish can be counted. Specifically, the image capturing module can first acquire multiple images at multiple time points and detect the ornamental fish in each image. The detection step can adopt any object detection algorithm, for example, The neural network or the support vector machine performs the detection, and the present invention is not limited thereto. After detection, the number of ornamental fish in each image (called the number of fish) can be calculated. After collecting multiple images, multiple fish numbers can be obtained, and then the mode of these fish numbers is calculated as an estimate. Number of fish. After each sampling period, it can be judged that the estimated fish number is increased, unchanged, or decreased. If the estimated number of fish is reduced, set a reward r R is the first value. If the estimated number of fish is constant, the reward r is set to a second value. If the estimated number of fish is increased, the reward is set to a third value, wherein the third value is greater than the second value, and the second value is greater than The first value. In other words, the reward set R described above includes the first value, the second value, and the third value. For example, the first value can be -10, the second value can be 1, and the third value can be 10, because the purpose of breeding is to try to keep the fish alive (the number is the same), so it should be given more than 0 as time increases. The reward, and if there is a fish dying (the number is reduced), a reward of less than 0 should be given, and in a few cases the number of fish will increase (reproduction), so a reward greater than 0 can also be given.

Next, a first action value function q ₁ ( s, a ) and a second action value function q ₂ ( s, a ) are established, wherein s S , a A. In the initialization phase, it is possible to set all states s and the first action value function q ₁ ( s, a ) and the second action value function q ₂ ( s, a ) under action a are randomly determined (or both are constants) For example, 0), these functions are used to determine which action a should be used in the current state s. In particular, in this embodiment, one of the following equations (1), (2) is executed at a first probability (e.g., 50%) to determine the action a _t corresponding to the current state s.

a _t =max _a q ₁ ( s,a )...(1)

a _t =max _a q ₂ ( s,a )...(2).

After determining the action a _t , the value of the device parameter can be determined. Therefore, the key to this embodiment is how to determine the first action value function q ₁ ( s, a ) and the second action value function q ₂ ( s, a ). In this embodiment, the action value function is gradually updated through an experiment. Specifically, the controller 120 may first perform the above action a _t , and calculate the current reward r and the state s ' by the value detected by the sensor 111 , where s ' S. Next, one of the following equations (3), (4) is performed at a second probability (for example, 50%).

q ₁ ( s,a ))...(3)

α , β , and γ are real numbers, and N _t ( s, a ₁ ) is the number of times the action a ₁ is selected in the state s. The real number α indicates the updated ratio, and the reward r is used to update the action value function. Equation (3), although the updating operation of the first cost function q ₁ (s, a), but also refers to the second operation function result value q ₂ (s, a), and similarly, although in Equation (4 The second action value function q ₂ ( s, a ) is updated, but also refers to the result of the first action value function q ₁ ( s, a ), in order to avoid one of the action value functions falling into the area The best solution (local optimal). In addition, N _t ( s, a ₁ ) is used to adopt some of the infrequently used actions a ₁ . When a certain action is used in the state s, the above equations (3) and (4) are increased. The opportunity to take this action. Through the above equations (3) and (4), the action value function can be gradually updated. After the update, the above equations (1) and (2) can be re-executed in the next sampling cycle. The more experimental data, the effect of breeding. will be better.

2 is a flow chart showing a smart fish tank information management method according to an embodiment. Referring to FIG. 2, in step 201, the breeding process is obtained through the sensor. In step 202, a plurality of equipment parameters corresponding to the farming equipment are calculated based on the farming history. In step 203, the corresponding breeding equipment is controlled according to the equipment parameters. However, the steps in FIG. 2 have been described in detail above, and will not be described again here. It should be noted that the steps in FIG. 2 can be implemented as multiple code codes or circuits, and the present invention is not limited thereto. In addition, the method of Figure 2 can Used in combination with the above embodiments, it can also be used alone. In other words, other steps can be added between the steps of FIG. 2.

In another aspect, the present invention also proposes a computer program product, which can be written by any programming language and/or platform. When the computer program product is loaded into a computer system and executed, the above-mentioned method.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

Claims (7)

A smart fish tank information management system includes: a network server; a plurality of sensors; a plurality of breeding equipment; and a controller for obtaining a breeding process through the sensors, and transmitting the breeding process to the a network server, wherein the network server calculates a plurality of device parameters corresponding to the breeding equipment according to the breeding history, wherein the controller obtains the device parameters and controls the corresponding breeding according to the device parameters The device includes an image capturing module, and the controller obtains a plurality of images at multiple time points through the image capturing module, and detects a plurality of ornamental fish in each of the images. The controller calculates the number of the ornamental fish in each of the images to obtain a fish number, wherein the controller calculates a majority of the number of fish to obtain an estimated number of fish, wherein the network server setting a culture cycle and a sampling cycle, the culture cycle being greater than the sampling cycle, establishing a state set S according to the values detected by the sensors, the sampling cycle and the breeding cycle, An action set A is established according to the device parameters, and a reward set R is established according to the estimated fish number, and a first action value function q ₁ ( s , a ) and a second action value function q ₂ ( s , a are established. ), where s S, a A. The network server determines one of the following equations (1) and (2) according to a first probability to determine an action a _t corresponding to the current state s, and determines the according to the action a _t Equipment parameters: a _t =max _a q ₁ ( s , a )...(1) a _t =max _a q ₂ ( s , a ) (2). The smart fish tank information management system according to claim 1, wherein the sensors further comprise a water quality detector, a thermometer and a light sensor, wherein the breeding equipment comprises a gas pump, a motor, a filter, a lamp, Water quality processor, filtration system and temperature control system. The smart aquarium information management system according to claim 1, wherein the controller performs the action a _t and calculates a current reward r and a state s ' , wherein r R,s ' S, the network server performs one of the following equations (3), (4) at a second probability: Where α, β, γ are real numbers, and N _t ( s , a ₁ ) is the number of times the action a ₁ is selected in this state s. The smart aquarium information management system according to claim 3, wherein if the estimated number of fish is reduced, the controller sets the reward r to be a first value, and if the estimated fish number does not change, the controller Setting the reward r as a second value, if the estimated number of fish increases, the controller sets the reward r to a third value, wherein the third value is greater than the second value, and the second value is greater than the first A value. A smart fish tank information management method for a smart fish tank information management system, the smart fish tank information management system comprises a plurality of sensors and a plurality of breeding equipment, the sensors comprising an image capturing module, the smart fish tank information The management method comprises: obtaining a breeding process through the sensors; calculating a plurality of equipment parameters corresponding to the breeding equipment according to the breeding history; controlling the corresponding breeding equipment according to the equipment parameters; capturing the image through the image The module acquires multiple images at multiple time points and detects a plurality of ornamental fish in each of the images; calculating the number of the ornamental fish in each of the images to obtain a fish number; calculating the number of the fish a population to obtain an estimated number of fish; a culture cycle and a sampling cycle, wherein the culture cycle is greater than the sampling cycle; based on the values detected by the sensors, the sampling cycle and the breeding cycle a state set S, establishing an action set A according to the device parameters, establishing a reward set R according to the estimated fish number, and establishing a first action value letter a number q ₁ ( s , a ) and a second action value function q ₂ ( s , a ), where s S, a A; and determining one of the following equations (1), (2) according to a first probability to determine an action a _t corresponding to the current state s, and determining the device parameters according to the action a _t : a _t = max _a q ₁ ( s , a ) (1) a _t =max _a q ₂ ( s , a ) (2). The smart fish tank information management method according to claim 5, wherein the sensors further comprise a water quality detector, a thermometer and a light sensor, and the breeding equipment comprises a gas pump, a motor, a filter, a lamp, and a water quality. Processor, filtration system and temperature control system. For example, the smart fish tank information management method described in claim 5 of the patent application further includes: performing the action a _t and calculating the current reward r and the state s ' , wherein r R,s ' S`; performs one of the following equations (3), (4) with a second probability: Where α, β, γ are real numbers, and N _t ( s , a ₁ ) is the number of times the action a ₁ is selected in this state s.