Intelligent breeding factory big data decision optimization system and method

Abstract

The invention provides an intelligent breeding factory big data decision optimization system and method. The system comprises a database, a model library, a data acquisition module, a biomass prediction module, an intelligent decision module and an execution module. The specific optimization method comprises the following steps: firstly, obtaining a corresponding predicted biomass I and a predicted biomass II based on collected breeding parameters and biological images, and fusing to obtain a final predicted biomass; corresponding target breeding parameters are obtained according to the final predicted biomass, the target breeding parameters are compared with breeding parameters collected in real time, and when the difference value of one or more breeding parameters of the same type exceeds a preset threshold value, corresponding decision optimization information is generated based on the breeding parameter in the target breeding parameters; and finally, adjusting the breeding environment and feeding work according to the decision optimization information. According to the invention, the matched aquaculture data is obtained by predicting the biomass of the current fish in real time, and the corresponding decision optimization information is generated, so that the intelligent level of the current aquaculture is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of aquaculture, and in particular to a big data decision-making optimization system and method for an intelligent aquaculture factory. Background Art

[0002] With the increase of population and consumption level, the demand for aquatic products is increasing. Traditional aquaculture methods are restricted by natural conditions, such as seasonal changes, water pollution and other factors, and it is difficult to meet the market demand. In this context, factory recirculating aquaculture has emerged as an advanced aquaculture model. Its core concept is to build a closed or semi-closed aquaculture system through industrial means to achieve a high degree of control and management of the aquaculture process, thus overcoming the limitations of traditional aquaculture methods.

[0003] The operation of the factory-scale recirculating aquaculture system is based on the principle of water recycling, that is, the wastewater generated during the aquaculture process is treated and reused for aquaculture, which can not only significantly reduce water resource consumption, but also effectively reduce environmental pollution. In addition, the system also introduces modern information technology, such as the Internet of Things technology, which deploys various sensor devices to collect data on key environmental parameters such as water quality, temperature, dissolved oxygen, etc., and analyzes and processes them through the cloud computing platform to provide a scientific basis for aquaculture decisions.

[0004] Despite this, factory-scale recirculating aquaculture still faces many challenges in practical applications. Existing data management systems are often relatively static, and it is difficult to dynamically adjust aquaculture parameters (including pond environmental parameters and feeding parameters) according to the actual biomass of farmed fish (including fish body length and width). Therefore, how to build a big data decision-making control system that can flexibly adapt to different growth conditions has become a key issue in promoting the further development of factory-scale recirculating aquaculture technology. Summary of the invention

[0005] In view of the problem that existing aquaculture platforms are difficult to dynamically adjust breeding data according to the actual growth of farmed fish, the present invention provides an intelligent breeding factory big data decision optimization system and method, which obtains matching breeding data by real-time prediction of the current fish biomass and generates corresponding decision optimization information, thereby improving the current level of intelligence in aquaculture.

[0006] The present invention achieves the above technical objectives through the following technical means.

[0007] An intelligent breeding factory big data decision-making optimization system, including:

[0008] A database for storing historical aquaculture parameter-biomass data sets corresponding to different aquaculture fish; the aquaculture parameters include aquaculture pond environmental parameters and feeding working parameters, and the biomass includes the body length and body width of the fish;

[0009] A model library for storing a number of models, including a farming parameter-biomass prediction model constructed based on a historical farming parameter-biomass data set, and a machine vision biomass model for predicting biomass based on biological images; the farming parameter-biomass prediction model can be self-learning and optimized based on data updates of the historical farming parameter-biomass data set;

[0010] The data acquisition module includes a breeding parameter acquisition device and an image acquisition device. The breeding parameter acquisition device is used to collect breeding parameters of the breeding pond in real time, and the image acquisition device is used to take biological images.

[0011] The biomass prediction module is used to obtain the predicted biomass I corresponding to the real-time collected breeding parameters according to the breeding parameter-biomass prediction model; to obtain the predicted biomass II corresponding to the captured biological image according to the machine vision biomass model; and to perform weighted fusion of the predicted biomass I and the predicted biomass II to obtain the final predicted biomass;

[0012] The intelligent decision-making module is used to find the corresponding target aquaculture parameters in the historical aquaculture parameter-biomass data set according to the final predicted biomass, and compare it with the aquaculture parameters collected in real time. When the difference between one or more aquaculture parameters of the same type exceeds the set threshold, the corresponding decision optimization information is generated based on the aquaculture parameter in the target aquaculture parameter;

[0013] The execution module is used to adjust the breeding environment and feeding work based on the decision optimization information.

[0014] Furthermore, the environmental parameters of the culture pond include light, water temperature, dissolved oxygen, pH value and ammonia nitrogen content; the feeding working parameters include feeding time and feeding amount.

[0015] Furthermore, the method for constructing the aquaculture parameter-biomass prediction model is as follows: first, extracting a historical aquaculture parameter-biomass data set of the same farmed fish from the historical aquaculture parameter-biomass data set; second, preprocessing the data, including data cleaning, standardization and feature selection, and randomly dividing it into a training set and a validation set; then, using the data in the training set to train the neural network model, and optimizing the network weights through the back propagation algorithm to minimize the difference between the estimated value and the actual biomass; finally, verifying the model performance through the data in the validation set, optimizing the model parameters, and obtaining the aquaculture parameter-biomass prediction model.

[0016] Furthermore, the self-learning optimization process of the aquaculture parameter-biomass prediction model is as follows: based on the historical aquaculture parameter-biomass data set of the same farmed fish species updated in the database, the biomass estimation results of the aquaculture parameter-biomass prediction model for the updated aquaculture parameters are compared with the actual biomass to form a feedback mechanism; the model parameters of the aquaculture parameter-biomass prediction model are adjusted according to the feedback results to achieve self-learning optimization.

[0017] Furthermore, the execution module includes:

[0018] Environmental control module, including thermostat for controlling water temperature and lighting equipment for controlling light;

[0019] Water quality control module, including an aerator for controlling dissolved oxygen, acid-base adjustment and dosing equipment for controlling pH value, and ammonia nitrogen treatment equipment for controlling the amount of ammonia nitrogen;

[0020] The intelligent feeding module includes a feeding robot for controlling feeding time and feeding amount.

[0021] Furthermore, the breeding parameter collection device includes a thermometer for detecting water temperature, a dissolved oxygen meter for detecting dissolved oxygen, a pH meter for detecting pH value, an ammonia nitrogen detector for detecting ammonia nitrogen amount, a light sensor for detecting light intensity, and a signal collector; the signal collector is connected to the intelligent feeding module signal and is used to collect the feeding working parameters of the intelligent feeding module.

[0022] Furthermore, the specific method by which the biomass prediction module obtains the predicted biomass II from the captured biological images is: using a convolutional neural network to perform feature recognition and image segmentation on the biological images transmitted by the image acquisition device, separating the target fish from the background and other non-related objects; then using an edge detection algorithm to extract the edge contour shape of the target fish; and finally calculating the predicted biomass II of the target fish based on the extracted edge contour shape.

[0023] Furthermore, the biomass prediction module adopts a weighted average method to fuse the predicted biomass I and the predicted biomass II.

[0024] Furthermore, it also includes a biomass correction module, which is used to input the measured biomass detected manually into the biomass prediction module to correct the weight values ​​of the predicted biomass I and the predicted biomass II in the weighted fusion process.

[0025] The optimization method of the intelligent breeding factory big data decision optimization system according to any one of the above items is characterized by comprising the following steps:

[0026] S1: The breeding parameter acquisition device collects the breeding parameters of the breeding pond in real time, and the image acquisition device takes biological images in real time;

[0027] S2: The biomass prediction module obtains the corresponding predicted biomass I and predicted biomass II according to the farming parameters and biological images collected in real time, and obtains the final predicted biomass by weighted fusion of the predicted biomass I and the predicted biomass II;

[0028] S3: The intelligent decision-making module obtains the corresponding target breeding parameters according to the final predicted biomass, and compares them with the breeding parameters collected in real time. When the difference between one or more breeding parameters of the same type exceeds a preset threshold, the intelligent decision-making module generates corresponding decision optimization information based on the breeding parameter in the target breeding parameters;

[0029] S4: The execution module adjusts the breeding environment and feeding work based on the decision optimization information.

[0030] The beneficial effects of the present invention are as follows:

[0031] 1. The present invention obtains the corresponding predicted biomass I and predicted biomass II based on the breeding parameter-biomass prediction model and the machine vision biomass model, and obtains the final predicted biomass through weighted fusion. This diversified prediction method reduces the need for manual collection and reduces labor costs on the one hand, and greatly improves the accuracy and reliability of biomass prediction on the other hand.

[0032] 2. The present invention provides reliable data support for optimizing aquaculture decisions by improving the accuracy of biomass prediction, thereby being able to better adjust feeding strategies and aquaculture environmental conditions, reduce resource waste, improve aquaculture efficiency, and ensure that aquaculture organisms are always in the best growth state.

[0033] 3. The aquaculture data-biomass prediction model in the present invention can continuously optimize its performance as new data continues to accumulate, which means that as time goes by, the prediction accuracy of the system will continue to improve, thereby achieving more accurate biomass prediction and more effective aquaculture management.

[0034] 4. The present invention integrates a biomass correction module to allow the actual biomass data detected manually to be input into the biomass prediction module to correct the weight values ​​of the predicted biomass I and the predicted biomass II in the weighted fusion process. This function improves the prediction accuracy of the biomass prediction module and further optimizes the breeding management and resource utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a module diagram of the intelligent breeding factory big data decision optimization system described in Example 1 of the present invention.

[0036] Figure 2 This is a module diagram of the intelligent breeding factory big data decision optimization system described in Example 2 of the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited thereto.

[0038] Example 1

[0039] The intelligent breeding factory big data decision-making optimization system described in this embodiment includes a database, a model library, a data acquisition module, a biomass prediction module, an intelligent decision-making module and an execution module. Figure 1 This is a module diagram of the intelligent breeding factory big data decision-making optimization system described in this embodiment.

[0040] The database is used to store historical breeding parameter-biomass data sets corresponding to different farmed fish. The breeding parameters include breeding pond environmental parameters and feeding working parameters. The breeding pond environmental parameters include light, water temperature, dissolved oxygen, pH value and ammonia nitrogen content; the feeding working parameters include feeding time and feeding amount. The biomass includes the body length and body width of the fish.

[0041] The model library is used to store several models, including a farming parameter-biomass prediction model constructed based on a historical farming parameter-biomass data set, and a machine vision biomass model for predicting biomass based on biological images. The farming parameter-biomass prediction model can perform self-learning optimization based on data updates of the historical farming parameter-biomass data set.

[0042] The method for constructing the aquaculture parameter-biomass prediction model is as follows: first, extracting a historical aquaculture parameter-biomass data set of the same farmed fish species from the historical aquaculture parameter-biomass data set; second, preprocessing the data, including data cleaning, standardization and feature selection, and randomly dividing it into a training set and a validation set; then, using the data in the training set to train a neural network model, optimizing the network weights through a back propagation algorithm to minimize the difference between the estimated value and the actual biomass; finally, verifying the model performance through the data in the validation set, optimizing the model parameters, and obtaining the aquaculture parameter-biomass prediction model.

[0043] The self-learning optimization process of the aquaculture parameter-biomass prediction model is as follows: based on the historical aquaculture parameter-biomass data set of the same aquaculture fish species updated in the database, the biomass estimation results of the aquaculture parameter-biomass prediction model for the updated aquaculture parameters are compared with the actual biomass to form a feedback mechanism; the model parameters of the aquaculture parameter-biomass prediction model are adjusted according to the feedback results to achieve self-learning optimization.

[0044] The data acquisition module includes a breeding parameter acquisition device and an image acquisition device. The breeding parameter acquisition device is used to collect breeding parameters of the breeding pond in real time, including a thermometer for detecting water temperature, a dissolved oxygen meter for detecting dissolved oxygen, a pH meter for detecting pH value, an ammonia nitrogen detector for detecting ammonia nitrogen amount, a light sensor for detecting light intensity, and a signal collector. The signal collector is connected to the intelligent feeding module signal and is used to collect the feeding working parameters of the intelligent feeding module. The image acquisition device is used to take biological images.

[0045] The biomass prediction module is used to obtain the predicted biomass I corresponding to the real-time collected breeding parameters according to the breeding parameter-biomass prediction model. It is used to obtain the predicted biomass II corresponding to the captured biological image according to the machine vision biomass model. The predicted biomass I and the predicted biomass II are weighted and fused by the weighted average method to obtain the final predicted biomass.

[0046] The specific method for the biomass prediction module to obtain the predicted biomass II from the captured biological images is: using a convolutional neural network to perform feature recognition and image segmentation on the biological images transmitted by the image acquisition device, separating the target fish from the background and other non-related objects; then using an edge detection algorithm to extract the edge contour shape of the target fish; finally, calculating the predicted biomass II of the target fish based on the extracted edge contour shape.

[0047] The intelligent decision-making module is used to find the corresponding target breeding parameters in the historical breeding parameter-biomass data set according to the final predicted biomass, and compare it with the breeding parameters collected in real time. When the difference between one or more breeding parameters of the same type exceeds the set threshold, the corresponding decision optimization information is generated based on the breeding parameter in the target breeding parameter.

[0048] The execution module is used to adjust the breeding environment and feeding work according to the decision optimization information. The execution module includes:

[0049] Environmental control module, including thermostat for controlling water temperature and lighting equipment for controlling light;

[0050] Water quality control module, including an aerator for controlling dissolved oxygen, acid-base adjustment and dosing equipment for controlling pH value, and ammonia nitrogen treatment equipment for controlling the amount of ammonia nitrogen;

[0051] The intelligent feeding module includes a feeding robot for controlling feeding time and feeding amount.

[0052] The optimization method of the above-mentioned intelligent breeding factory big data decision optimization system includes the following steps:

[0053] S1: The breeding parameter acquisition device collects the breeding parameters of the breeding pond in real time, and the image acquisition device takes biological images in real time;

[0054] S2: The biomass prediction module obtains the corresponding predicted biomass I and predicted biomass II according to the farming parameters and biological images collected in real time, and obtains the final predicted biomass by weighted fusion of the predicted biomass I and the predicted biomass II;

[0055] S3: The intelligent decision-making module obtains the corresponding target breeding parameters according to the final predicted biomass, and compares them with the breeding parameters collected in real time. When the difference between one or more breeding parameters of the same type exceeds a preset threshold, the intelligent decision-making module generates corresponding decision optimization information based on the breeding parameter in the target breeding parameters;

[0056] S4: The execution module adjusts the breeding environment and feeding work based on the decision optimization information.

[0057] Example 2

[0058] The difference between the intelligent breeding factory big data decision-making optimization system described in this embodiment and embodiment 1 is that a biomass correction module is added, and the biomass correction module is used to input the measured biomass detected manually into the biomass prediction module to correct the weight values ​​of the predicted biomass I and the predicted biomass II in the weighted fusion process. Figure 2 This is a module diagram of the intelligent breeding factory big data decision optimization system described in this embodiment. The prediction accuracy of the biomass prediction module is improved, and the breeding management and resource utilization efficiency are further optimized.

[0059] The embodiments are preferred implementations of the present invention, but the present invention is not limited to the above-mentioned implementations. Any obvious improvements, substitutions or modifications that can be made by those skilled in the art without departing from the essential content of the present invention belong to the protection scope of the present invention.

Claims

1. An intelligent breeding factory big data decision-making optimization system, characterized in that: include: A database for storing historical aquaculture parameter-biomass data sets corresponding to different aquaculture fish; the aquaculture parameters include aquaculture pond environmental parameters and feeding working parameters, and the biomass includes the body length and body width of the fish; A model library for storing a number of models, including a farming parameter-biomass prediction model constructed based on a historical farming parameter-biomass data set, and a machine vision biomass model for predicting biomass based on biological images; the farming parameter-biomass prediction model can be self-learning and optimized based on data updates of the historical farming parameter-biomass data set; The data acquisition module includes a breeding parameter acquisition device and an image acquisition device. The breeding parameter acquisition device is used to collect breeding parameters of the breeding pond in real time, and the image acquisition device is used to take biological images. The biomass prediction module is used to obtain the predicted biomass I corresponding to the real-time collected breeding parameters according to the breeding parameter-biomass prediction model; to obtain the predicted biomass II corresponding to the captured biological image according to the machine vision biomass model; and to perform weighted fusion of the predicted biomass I and the predicted biomass II to obtain the final predicted biomass; The intelligent decision-making module is used to find the corresponding target aquaculture parameters in the historical aquaculture parameter-biomass data set according to the final predicted biomass, and compare it with the aquaculture parameters collected in real time. When the difference between one or more aquaculture parameters of the same type exceeds the set threshold, the corresponding decision optimization information is generated based on the aquaculture parameter in the target aquaculture parameter; The execution module is used to adjust the breeding environment and feeding work based on the decision optimization information.

2. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The breeding pond environmental parameters include light, water temperature, dissolved oxygen, pH value and ammonia nitrogen content; the feeding working parameters include feeding time and feeding amount.

3. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The method for constructing the aquaculture parameter-biomass prediction model is as follows: first, extracting a historical aquaculture parameter-biomass data set of the same farmed fish species from the historical aquaculture parameter-biomass data set; second, preprocessing the data, including data cleaning, standardization and feature selection, and randomly dividing it into a training set and a validation set; then, using the data in the training set to train a neural network model, optimizing the network weights through a back propagation algorithm to minimize the difference between the estimated value and the actual biomass; finally, verifying the model performance through the data in the validation set, optimizing the model parameters, and obtaining the aquaculture parameter-biomass prediction model.

4. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The self-learning optimization process of the aquaculture parameter-biomass prediction model is as follows: based on the historical aquaculture parameter-biomass data set of the same aquaculture fish species updated in the database, the biomass estimation results of the aquaculture parameter-biomass prediction model for the updated aquaculture parameters are compared with the actual biomass to form a feedback mechanism; the model parameters of the aquaculture parameter-biomass prediction model are adjusted according to the feedback results to achieve self-learning optimization.

5. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The execution module includes: Environmental control module, including thermostat for controlling water temperature and lighting equipment for controlling light; Water quality control module, including an aerator for controlling dissolved oxygen, acid-base adjustment and dosing equipment for controlling pH value, and ammonia nitrogen treatment equipment for controlling the amount of ammonia nitrogen; The intelligent feeding module includes a feeding robot for controlling feeding time and feeding amount.

6. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The breeding parameter collection device includes a thermometer for detecting water temperature, a dissolved oxygen meter for detecting dissolved oxygen, a pH meter for detecting pH value, an ammonia nitrogen detector for detecting ammonia nitrogen amount, a light sensor for detecting light intensity, and a signal collector; the signal collector is connected to the intelligent feeding module signal and is used to collect the feeding working parameters of the intelligent feeding module.

7. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The specific method for the biomass prediction module to obtain the predicted biomass II from the captured biological images is: using a convolutional neural network to perform feature recognition and image segmentation on the biological images transmitted by the image acquisition device, separating the target fish from the background and other non-related objects; then using an edge detection algorithm to extract the edge contour shape of the target fish; finally, calculating the predicted biomass II of the target fish based on the extracted edge contour shape.

8. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: The biomass prediction module adopts a weighted average method to fuse the predicted biomass I and the predicted biomass II.

9. The intelligent breeding factory big data decision optimization system according to claim 1 is characterized in that: It also includes a biomass correction module, which is used to input the artificially detected actual biomass into the biomass prediction module to correct the weight values ​​of the predicted biomass I and the predicted biomass II in the weighted fusion process.

10. The optimization method of the intelligent breeding factory big data decision optimization system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1: The breeding parameter acquisition device collects the breeding parameters of the breeding pond in real time, and the image acquisition device takes biological images in real time; S2: The biomass prediction module obtains the corresponding predicted biomass I and predicted biomass II according to the farming parameters and biological images collected in real time, and obtains the final predicted biomass by weighted fusion of the predicted biomass I and the predicted biomass II; S3: The intelligent decision-making module obtains the corresponding target breeding parameters according to the final predicted biomass, and compares them with the breeding parameters collected in real time. When the difference between one or more breeding parameters of the same type exceeds a preset threshold, the intelligent decision-making module generates corresponding decision optimization information based on the breeding parameter in the target breeding parameters; S4: The execution module adjusts the breeding environment and feeding work based on the decision optimization information.

Images