Healthy livestock breeding environment disease data analysis system and method based on machine learning

Abstract

The invention relates to the technical field of data analysis, and discloses a livestock healthy breeding environment disease data analysis system and method based on machine learning. According to the method, historical environment parameters in a healthy breeding area and disease data corresponding to the environment parameters are collected, a training set is constructed, meanwhile, the historical environment parameters in the training set and the disease data corresponding to the environment parameters are scored by collecting experience of different experts, and a quality evaluation system is established; after a quality evaluation system is established, historical environment parameters in a training set and disease data corresponding to the environment parameters are trained in a machine learning mode, optimal environment parameters are output, and a healthy breeding environment disease data analysis model is constructed. And then environment parameters in the healthy breeding area are collected in real time by installing the Internet of Things monitoring equipment and uploaded to the data analysis platform in real time, and meanwhile, an analysis result is output in a data matching mode, so that the factuality of environment disease data analysis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a system and method for analyzing livestock health breeding environment disease data based on machine learning. Background Art

[0002] So far, healthy livestock breeding is still in a relatively backward production state, with low technical content and information level; and healthy livestock breeding is easily affected by environmental diseases, so it is necessary to scientifically and rationally monitor and analyze environmental parameters and improve the information level of healthy livestock breeding.

[0003] The existing Chinese patent application CN110414146A, the method collects basic information of the water environment governance area, and constructs a conceptual model of the water environment governance area through the collected basic information; at the same time, a numerical model of the water environment governance area is established based on the constructed conceptual model of the water environment governance area, and the numerical model of the water environment governance area is corrected by establishing the numerical model of the water environment governance area; the corrected numerical model of the water environment governance area is run to generate alternative model training samples, and the alternative model is trained using deep learning; at the same time, multi-objective optimization calculation of the design parameters of the water environment governance project is performed based on the trained alternative model, and the optimal water environment governance design parameters are output. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the existing technology, the present invention provides a livestock healthy breeding environmental disease data analysis system and method based on machine learning, which has the advantages of real-time, accurate, efficient and safe, and solves the problem that livestock healthy breeding is easily affected by environmental diseases.

[0006] (II) Technical solution

[0007] In order to solve the above technical problem that healthy animal husbandry is easily affected by environmental diseases, the present invention provides the following technical solutions:

[0008] This embodiment discloses a method for analyzing livestock health breeding environment disease data based on machine learning, which specifically includes the following steps:

[0009] S1. Collect historical environmental parameters and disease data of corresponding environmental parameters in healthy breeding areas and construct training sets;

[0010] S2. Score the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by collecting the experience of different experts and establish a quality assessment system;

[0011] S3. Train the historical environmental parameters and disease data of the corresponding environmental parameters in the training set by machine learning, and output the optimal environmental parameters and build a healthy breeding environment disease data analysis model;

[0012] S4. Install IoT monitoring equipment to collect environmental parameters in the healthy breeding area in real time, and upload the real-time collected environmental parameters to the data analysis platform based on the message queue telemetry transmission method;

[0013] S5. The data analysis platform inputs the real-time collected environmental parameters into the constructed healthy aquaculture environment disease data analysis model and quality assessment system, and outputs the analysis results through data matching.

[0014] The present invention collects historical environmental parameters and disease data of corresponding environmental parameters in a healthy breeding area and constructs a training set, and at the same time collects the experience of different experts to score the historical environmental parameters and disease data of corresponding environmental parameters in the training set and establishes a quality evaluation system; after the quality evaluation system is established, the historical environmental parameters and disease data of corresponding environmental parameters in the training set are trained by machine learning, and the optimal environmental parameters are output and a healthy breeding environment disease data analysis model is constructed, and then the environmental parameters in the healthy breeding area are collected in real time by installing Internet of Things monitoring equipment and uploaded to the data analysis platform in real time, and the analysis results are output by data matching, thereby improving the factuality of environmental disease data analysis.

[0015] Preferably, the process of scoring the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by collecting the experience of different experts and establishing a quality assessment system comprises the following steps:

[0016] S21. Arrange the data sets in the training set in order and divide them into 10 categories in equal amounts. Different experts judge the historical environmental parameters and the disease data of the corresponding environmental parameters in each category based on their own experience, and convert the scoring results of different experts into fuzzy numbers. At the same time, the weighted average method is used to aggregate the fuzzy numbers.

[0017] S22. The fuzzy number of each expert scoring result is used as a set of expert quality assessment system data, and all expert quality assessment system data are summarized to preliminarily establish a quality assessment system.

[0018] Preferably, converting the scoring results of different experts into fuzzy numbers and aggregating the fuzzy numbers using a weighted average method comprises the following steps:

[0019] The weighted average method is as follows:

[0020]

[0021] Among them, Pz It is represented by the fuzzy score of the experts on the historical environmental parameters of the zth category in the training set and the disease data of the corresponding environmental parameters, P zr represents the fuzzy score of expert r on the historical environmental parameters of the zth category in the training set and the disease data of the corresponding environmental parameters, W r represents the weight of expert r, and y represents the total number of experts.

[0022] The present invention arranges and divides each group of data in the training set in sequence, collects scores of different experts on each group of data in the training set after division, and clusters them to obtain fuzzy scores for each group of data in the training set and establish a quality evaluation system, thereby improving the efficiency of environmental disease data analysis.

[0023] Preferably, the training of the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by machine learning and the construction of a healthy aquaculture environment disease data analysis model comprises the following steps:

[0024] S31. Determine the structure and related parameters of machine learning;

[0025] The structure of machine learning includes: input structure, intermediate structure, and output structure;

[0026] Set the input structure, intermediate structure and output structure to have n, m, and s nodes respectively;

[0027] Combine the historical environmental parameters and the disease data corresponding to the environmental parameters in the training set, and set each set of historical environmental parameters and the disease data corresponding to the environmental parameters to represent an input signal;

[0028] S32, determining a training model for a machine learning structure;

[0029] S33. The training model of the machine learning structure is improved through the particle swarm algorithm to obtain a healthy aquaculture environment disease data analysis model and optimal environmental parameters.

[0030] Preferably, the determining of the training model of the machine learning structure comprises the following steps:

[0031] S321, input signal forward propagation process;

[0032] The threshold of each structure is set. After the input structure receives a set of historical environmental parameters and disease data corresponding to the environmental parameters, a set of input signals is generated. When the threshold of each structure is not exceeded, the signals pass through the intermediate structure and are output from the output structure.

[0033] In the intermediate structure, the output formula of the jth node is as follows:

[0034]

[0035] Among them, a j represents the output of the jth node of the intermediate structure, f represents the activation function, and p i represents the input signal of the i-th node of the input structure, ω ij represents the weight between the i-th node of the input structure and the j-th node of the intermediate structure, b j represents the threshold of the jth node of the intermediate structure;

[0036] In the output structure, the output formula of the kth node is as follows:

[0037]

[0038] Among them, a k represents the output of the kth node of the output structure, ω jk represents the weight between the jth node of the intermediate structure and the kth node of the output structure, b k Represents the threshold of the kth node of the output structure;

[0039] S322, error back propagation process;

[0040] S323, determining whether the iteration is completed and outputting the training model of the machine learning structure;

[0041] After each iteration, the size of the output error is checked to determine whether the output error is smaller than the expected error. If it is smaller, the iteration is terminated and the training model of the machine learning structure corresponding to the current weight is output. Otherwise, the iteration is continued until the requirements are met.

[0042] Preferably, the error back propagation process comprises the following steps:

[0043] When the input signal exceeds the threshold of each structure during the forward propagation process, the machine learning structure is corrected through the error back propagation process;

[0044] Set the expected output of the machine learning structure to β k , the error function is as follows:

[0045]

[0046] Where E represents the error function;

[0047] According to the negative gradient descent method, when the output exceeds the expected output, the weight of the current machine learning structure is gradually reduced, and when the output is less than the expected output, the weight of the current machine learning structure is gradually increased.

[0048] Preferably, the method of improving the training model of the machine learning structure by using a particle swarm algorithm to obtain a healthy aquaculture environment disease data analysis model and optimal environmental parameters comprises the following steps:

[0049] S331, particle swarm optimization algorithm parameter initialization;

[0050] Set the group size and the maximum number of iterations d max , particle random position q, particle velocity u;

[0051] Each particle is set to represent a set of historical environmental parameters and disease data corresponding to the environmental parameters;

[0052] S332, calculating the fitness of each particle;

[0053]

[0054] Among them, H represents the fitness function of the particle, and t represents a constant between 0 and 1;

[0055] S333, updating the optimal position of a single particle based on the calculated fitness;

[0056] For each particle's fitness, the fitness of its current position is compared with the best position q it has passed. best If the fitness of the current position is greater than the best position q it has passed, best The fitness of the current position is taken as the current best position q best , if the fitness of the current position is less than or equal to the best position q it has passed best The fitness of the current best position q is not changed. best ;

[0057] S334, updating the optimal position of the group based on the calculated fitness;

[0058] For each particle's calculated fitness, the fitness of its current position is compared with the best position g that the particle in its population has passed. best If the fitness of the current position is greater than the best position g that the particle in the population has passed through, best The fitness of the current position is taken as the current best position g best , if the fitness of the current position is less than or equal to the best position g that the particle in the population has passed best The fitness of the current best position g is not changed. best ;

[0059] S334, repeating steps S332-S334 until the maximum number of iterations is reached, and outputting a healthy breeding environment disease data analysis model and optimal environmental parameters;

[0060] The optimal position where particles in the population pass is set as the optimal environmental parameter.

[0061] The present invention trains the historical environmental parameters in the training set and the disease data of the corresponding environmental parameters by machine learning, and uses the particle swarm algorithm to improve the training model of the machine learning structure, thereby obtaining a healthy breeding environment disease data analysis model and optimal environmental parameters, determining the optimal environmental parameters for healthy livestock breeding environment disease data, and ensuring the accuracy of environmental disease data analysis.

[0062] Preferably, the data analysis platform inputs the real-time collected environmental parameters into the constructed healthy aquaculture environment disease data analysis model and quality assessment system, and outputs the analysis results by data matching, including the following steps:

[0063] S51, analyzing and outputting the environmental parameters collected in real time through a healthy breeding environment disease data analysis model to obtain analyzed environmental parameters;

[0064] S52, calculating the distance between the analyzed environmental parameters and the optimal environmental parameters;

[0065] S53, outputting the analysis result by data matching based on the distance between the calculated and analyzed environmental parameters and the optimal environmental parameters;

[0066] The optimal environmental parameter score is set to 100, and the analyzed environmental parameters are matched to corresponding categories based on the calculated distance between the analyzed environmental parameters and the optimal environmental parameters, and the score of the corresponding category is used as the score of the analyzed environmental parameters.

[0067] Preferably, the distance between the calculated and analyzed environmental parameters and the optimal environmental parameters comprises the following steps:

[0068]

[0069] Among them, d represents the distance between the analyzed environmental parameters and the optimal environmental parameters, x1 represents the horizontal coordinate of the analyzed environmental parameters, y1 represents the vertical coordinate of the analyzed environmental parameters, and x best The horizontal axis represents the optimal environmental parameters, y best The vertical axis represents the optimal environmental parameters.

[0070] The present invention analyzes and outputs the environmental parameters collected in real time through a healthy breeding environment disease data analysis model, and simultaneously calculates the distance between the analyzed environmental parameters and the optimal environmental parameters and performs data matching to determine the scores of the environmental parameters collected in real time.

[0071] This embodiment discloses a livestock health breeding environment disease data analysis system based on machine learning, including: a data acquisition module, a scoring and evaluation module, a model building module, a data analysis platform and a data matching module;

[0072] The data acquisition module is used to collect environmental parameters in the healthy breeding area in real time;

[0073] The scoring and evaluation module is used to score the collected environmental parameters in the healthy breeding area;

[0074] The data analysis platform includes a model building module and a data matching module;

[0075] The model building module is used to build a healthy breeding environment disease data analysis model and output optimal environmental parameters through machine learning;

[0076] The data matching module is used to perform score matching according to the analysis results.

[0077] (III) Beneficial effects

[0078] Compared with the prior art, the present invention provides a system and method for analyzing livestock health breeding environment disease data based on machine learning, which has the following beneficial effects:

[0079] 1. The invention collects historical environmental parameters and disease data of corresponding environmental parameters in healthy breeding areas and constructs a training set. At the same time, the historical environmental parameters and disease data of corresponding environmental parameters in the training set are scored by collecting the experience of different experts and a quality assessment system is established. After the quality assessment system is established, the historical environmental parameters and disease data of corresponding environmental parameters in the training set are trained by machine learning, and the optimal environmental parameters are output and a healthy breeding environment disease data analysis model is constructed. Then, the environmental parameters in the healthy breeding area are collected in real time by installing Internet of Things monitoring equipment and uploaded to the data analysis platform in real time. At the same time, the analysis results are output by data matching, thereby improving the factuality of environmental disease data analysis.

[0080] 2. The invention improves the efficiency of environmental disease data analysis by sequentially arranging and dividing each group of data in the training set, collecting scores from different experts on each group of data in the training set after division, and clustering them to obtain fuzzy scores for each group of data in the training set and establish a quality assessment system.

[0081] 3. This invention trains the historical environmental parameters in the training set and the disease data of the corresponding environmental parameters through machine learning, and uses the particle swarm algorithm to improve the training model of the machine learning structure, thereby obtaining a healthy breeding environment disease data analysis model and optimal environmental parameters, and determining the optimal environmental parameters for the healthy breeding environment disease data of livestock, thereby ensuring the accuracy of the environmental disease data analysis.

[0082] 4. The invention analyzes and outputs the environmental parameters collected in real time through a healthy breeding environment disease data analysis model, and at the same time calculates the distance between the analyzed environmental parameters and the optimal environmental parameters and performs data matching to determine the score of the environmental parameters collected in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 This is a structural schematic diagram of the livestock health breeding environment disease data analysis process of the present invention. DETAILED DESCRIPTION

[0084] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0085] Example 1

[0086] See also Figure 1 This embodiment discloses a method for analyzing livestock health breeding environment disease data based on machine learning, which specifically includes the following steps:

[0087] S1. Collect historical environmental parameters and disease data of corresponding environmental parameters in healthy breeding areas and construct training sets;

[0088] S2. Score the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by collecting the experience of different experts and establish a quality assessment system;

[0089] Scoring the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by collecting the experience of different experts includes the following steps:

[0090] S21. Arrange the data sets in the training set in order and divide them into 10 categories in equal amounts. Different experts judge the historical environmental parameters and the disease data of the corresponding environmental parameters in each category based on their own experience, and convert the scoring results of different experts into fuzzy numbers. At the same time, the weighted average method is used to aggregate the fuzzy numbers.

[0091] The weighted average method is as follows:

[0092]

[0093] Among them, P z It is represented by the fuzzy score of the experts on the historical environmental parameters of the zth category in the training set and the disease data of the corresponding environmental parameters, P zrrepresents the fuzzy score of expert r on the historical environmental parameters of the zth category in the training set and the disease data of the corresponding environmental parameters, W r represents the weight of expert r, and y represents the total number of experts;

[0094] S22, taking the fuzzy number of each expert's scoring result as a set of expert quality assessment system data, and summarizing all expert quality assessment system data to preliminarily establish a quality assessment system;

[0095] S3. Train the historical environmental parameters and disease data of the corresponding environmental parameters in the training set by machine learning, and output the optimal environmental parameters and build a healthy breeding environment disease data analysis model;

[0096] The historical environmental parameters and the disease data of the corresponding environmental parameters in the training set are trained by machine learning, and a healthy aquaculture environment disease data analysis model is constructed, including the following steps:

[0097] S31. Determine the structure and related parameters of machine learning;

[0098] The structure of machine learning includes: input structure, intermediate structure, and output structure;

[0099] Furthermore, the input structure, the intermediate structure, and the output structure are set to have n, m, and s nodes respectively;

[0100] Combine the historical environmental parameters and the disease data corresponding to the environmental parameters in the training set, and set each set of historical environmental parameters and the disease data corresponding to the environmental parameters to represent an input signal;

[0101] S32, determining a training model for a machine learning structure;

[0102] S321, input signal forward propagation process;

[0103] The threshold of each structure is set. After the input structure receives a set of historical environmental parameters and disease data corresponding to the environmental parameters, a set of input signals is generated. When the threshold of each structure is not exceeded, the signals pass through the intermediate structure and are output from the output structure.

[0104] In the intermediate structure, the output formula of the jth node is as follows:

[0105]

[0106] Among them, a j represents the output of the jth node of the intermediate structure, f represents the activation function, and p i represents the input signal of the i-th node of the input structure, ω ij represents the weight between the i-th node of the input structure and the j-th node of the intermediate structure, bj represents the threshold of the jth node of the intermediate structure;

[0107] In the output structure, the output formula of the kth node is as follows:

[0108]

[0109] Among them, a k represents the output of the kth node of the output structure, ω jk represents the weight between the jth node of the intermediate structure and the kth node of the output structure, b k Represents the threshold of the kth node of the output structure;

[0110] S322, error back propagation process;

[0111] When the input signal exceeds the threshold of each structure during the forward propagation process, the machine learning structure is corrected through the error back propagation process;

[0112] Set the expected output of the machine learning structure to β k , the error function is as follows:

[0113]

[0114] Where E represents the error function;

[0115] According to the negative gradient descent method, when the output exceeds the expected output, the weight of the current machine learning structure is gradually reduced, and when the output is less than the expected output, the weight of the current machine learning structure is gradually increased;

[0116] S323, determining whether the iteration is completed and outputting the training model of the machine learning structure;

[0117] After each iteration, the size of the output error is checked to determine whether the output error is smaller than the expected error. If it is smaller, the iteration is terminated and the training model of the machine learning structure corresponding to the current weight is output. Otherwise, the iteration is continued until the requirements are met.

[0118] S33, improving the training model of the machine learning structure through the particle swarm algorithm, and obtaining a healthy breeding environment disease data analysis model and optimal environmental parameters;

[0119] S331, particle swarm optimization algorithm parameter initialization;

[0120] Set the group size and the maximum number of iterations d max , particle random position q, particle velocity u;

[0121] Each particle is set to represent a set of historical environmental parameters and disease data corresponding to the environmental parameters;

[0122] S332, calculating the fitness of each particle;

[0123]

[0124] Among them, H represents the fitness function of the particle, and t represents a constant between 0 and 1;

[0125] S333, updating the optimal position of a single particle based on the calculated fitness;

[0126] For each particle's fitness, the fitness of its current position is compared with the best position q it has passed. best If the fitness of the current position is greater than the best position q it has passed, best The fitness of the current position is taken as the current best position q best , if the fitness of the current position is less than or equal to the best position q it has passed best The fitness of the current best position q is not changed. best ;

[0127] S334, updating the optimal position of the group based on the calculated fitness;

[0128] For each particle's calculated fitness, the fitness of its current position is compared with the best position g that the particle in its population has passed. best If the fitness of the current position is greater than the best position g that the particle in the population has passed through, best The fitness of the current position is taken as the current best position g best , if the fitness of the current position is less than or equal to the best position g that the particle in the population has passed best The fitness of the current best position g is not changed. best ;

[0129] S334, repeating steps S332-S334 until the maximum number of iterations is reached, and outputting a healthy breeding environment disease data analysis model and optimal environmental parameters;

[0130] Set the best position where particles in the population pass as the optimal environmental parameters;

[0131] S4. Install IoT monitoring equipment to collect environmental parameters in the healthy breeding area in real time, and upload the real-time collected environmental parameters to the data analysis platform based on the message queue telemetry transmission method;

[0132] S5. The data analysis platform inputs the real-time collected environmental parameters into the constructed healthy aquaculture environment disease data analysis model and quality assessment system, and outputs the analysis results through data matching;

[0133] The data analysis platform inputs the real-time collected environmental parameters into the constructed healthy aquaculture environment disease data analysis model and quality assessment system, and outputs the analysis results through data matching, including the following steps:

[0134] S51, analyzing and outputting the environmental parameters collected in real time through a healthy breeding environment disease data analysis model to obtain analyzed environmental parameters;

[0135] S52, calculating the distance between the analyzed environmental parameters and the optimal environmental parameters;

[0136]

[0137] Among them, d represents the distance between the analyzed environmental parameters and the optimal environmental parameters, x1 represents the horizontal coordinate of the analyzed environmental parameters, y1 represents the vertical coordinate of the analyzed environmental parameters, and x best The horizontal axis represents the optimal environmental parameters, y best The ordinate represents the optimal environmental parameters;

[0138] S53, outputting the analysis result by data matching based on the distance between the calculated and analyzed environmental parameters and the optimal environmental parameters;

[0139] The best environmental parameter score is set to 100, and the analyzed environmental parameters are matched to corresponding categories based on the calculated distance between the analyzed environmental parameters and the best environmental parameters, and the scores of the corresponding categories are used as the scores of the analyzed environmental parameters;

[0140] Example 2

[0141] This embodiment discloses a livestock health breeding environment disease data analysis system based on machine learning, including: a data acquisition module, a scoring and evaluation module, a model building module, a data analysis platform and a data matching module;

[0142] The data acquisition module is used to collect environmental parameters in the healthy breeding area in real time;

[0143] The scoring and evaluation module is used to score the collected environmental parameters in the healthy breeding area;

[0144] The data analysis platform includes a model building module and a data matching module;

[0145] The model building module is used to build a healthy breeding environment disease data analysis model and output optimal environmental parameters through machine learning;

[0146] The data matching module is used to perform score matching according to the analysis results.

[0147] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for analyzing livestock health breeding environment disease data based on machine learning, characterized in that: The following steps are involved: S1. Collect historical environmental parameters and disease data of corresponding environmental parameters in healthy breeding areas and construct training sets; S2. Score the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by collecting the experience of different experts and establish a quality assessment system; S3. Train the historical environmental parameters and disease data of the corresponding environmental parameters in the training set by machine learning, and output the optimal environmental parameters and build a healthy breeding environment disease data analysis model; S4. Install IoT monitoring equipment to collect environmental parameters in the healthy breeding area in real time, and upload the real-time collected environmental parameters to the data analysis platform based on the message queue telemetry transmission method; S5. The data analysis platform inputs the real-time collected environmental parameters into the constructed healthy aquaculture environment disease data analysis model and quality assessment system, and outputs the analysis results through data matching.

2. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 1, characterized in that: The process of collecting the experience of different experts to score the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set and establishing a quality assessment system includes the following steps: S21. Arrange the data sets in the training set in order and divide them into 10 categories in equal amounts. Different experts judge the historical environmental parameters and the disease data of the corresponding environmental parameters in each category based on their own experience, and convert the scoring results of different experts into fuzzy numbers. At the same time, the weighted average method is used to aggregate the fuzzy numbers. S22. The fuzzy number of each expert scoring result is used as a set of expert quality assessment system data, and all expert quality assessment system data are summarized to preliminarily establish a quality assessment system.

3. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 2 is characterized in that: The step of converting the scoring results of different experts into fuzzy numbers and aggregating the fuzzy numbers using the weighted average method includes the following steps: The weighted average method is as follows: Among them, P z It is represented by the fuzzy score of the experts on the historical environmental parameters of the zth category in the training set and the disease data of the corresponding environmental parameters, P zr represents the fuzzy score of expert r on the historical environmental parameters of the zth category in the training set and the disease data of the corresponding environmental parameters, W r represents the weight of expert r, and y represents the total number of experts.

4. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 1, characterized in that: The method of training the historical environmental parameters and the disease data of the corresponding environmental parameters in the training set by machine learning and constructing a healthy breeding environment disease data analysis model includes the following steps: S31. Determine the structure and related parameters of machine learning; The structure of machine learning includes: input structure, intermediate structure, and output structure; Set the input structure, intermediate structure and output structure to have n, m, and s nodes respectively; Combine the historical environmental parameters and the disease data corresponding to the environmental parameters in the training set, and set each set of historical environmental parameters and the disease data corresponding to the environmental parameters to represent an input signal; S32, determining a training model for a machine learning structure; S33. The training model of the machine learning structure is improved through the particle swarm algorithm to obtain a healthy aquaculture environment disease data analysis model and optimal environmental parameters.

5. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 4 is characterized in that: The training model for determining the machine learning structure comprises the following steps: S321, input signal forward propagation process; The threshold of each structure is set. After the input structure receives a set of historical environmental parameters and disease data corresponding to the environmental parameters, a set of input signals is generated. When the threshold of each structure is not exceeded, the signals pass through the intermediate structure and are output from the output structure. In the intermediate structure, the output formula of the jth node is as follows: Among them, a j represents the output of the jth node of the intermediate structure, f represents the activation function, and p i represents the input signal of the i-th node of the input structure, ω ij represents the weight between the i-th node of the input structure and the j-th node of the intermediate structure, b j represents the threshold of the jth node of the intermediate structure; In the output structure, the output formula of the kth node is as follows: Among them, a k represents the output of the kth node of the output structure, ω jk represents the weight between the jth node of the intermediate structure and the kth node of the output structure, b k Represents the threshold of the kth node of the output structure; S322, error back propagation process; S323, determining whether the iteration is completed and outputting the training model of the machine learning structure; After each iteration, the size of the output error is checked to determine whether the output error is smaller than the expected error. If it is smaller, the iteration is terminated and the training model of the machine learning structure corresponding to the current weight is output. Otherwise, the iteration is continued until the requirements are met.

6. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 5, characterized in that: The error back propagation process includes the following steps: When the input signal exceeds the threshold of each structure during the forward propagation process, the machine learning structure is corrected through the error back propagation process; Set the expected output of the machine learning structure to β k , the error function is as follows: Where E represents the error function; According to the negative gradient descent method, when the output exceeds the expected output, the weight of the current machine learning structure is gradually reduced, and when the output is less than the expected output, the weight of the current machine learning structure is gradually increased.

7. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 4 is characterized in that: The method of improving the training model of the machine learning structure by using the particle swarm algorithm to obtain the healthy breeding environment disease data analysis model and the optimal environmental parameters includes the following steps: S331, particle swarm optimization algorithm parameter initialization; Set the group size and the maximum number of iterations d max , particle random position q, particle velocity u; Each particle is set to represent a set of historical environmental parameters and disease data corresponding to the environmental parameters; S332, calculating the fitness of each particle; Among them, H represents the fitness function of the particle, and t represents a constant between 0 and 1; S333, updating the optimal position of a single particle based on the calculated fitness; For each particle's fitness, the fitness of its current position is compared with the best position q it has passed. best If the fitness of the current position is greater than the best position q it has passed, best The fitness of the current position is taken as the current best position q best , if the fitness of the current position is less than or equal to the best position q it has passed best The fitness of the current best position q is not changed. best ; S334, updating the optimal position of the group based on the calculated fitness; For each particle's calculated fitness, the fitness of its current position is compared with the best position g that the particle in its population has passed. best If the fitness of the current position is greater than the best position g that the particle in the population has passed through, best The fitness of the current position is taken as the current best position g best , if the fitness of the current position is less than or equal to the best position g that the particle in the population has passed best The fitness of the current best position g is not changed. best ; S334, repeating steps S332-S334 until the maximum number of iterations is reached, and outputting a healthy breeding environment disease data analysis model and optimal environmental parameters; The optimal position where particles in the population pass is set as the optimal environmental parameter.

8. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 1, characterized in that: The data analysis platform inputs the real-time collected environmental parameters into the constructed healthy aquaculture environment disease data analysis model and quality assessment system, and outputs the analysis results by data matching, including the following steps: S51, analyzing and outputting the environmental parameters collected in real time through a healthy breeding environment disease data analysis model to obtain analyzed environmental parameters; S52, calculating the distance between the analyzed environmental parameters and the optimal environmental parameters; S53, outputting the analysis result by data matching based on the distance between the calculated and analyzed environmental parameters and the optimal environmental parameters; The optimal environmental parameter score is set to 100, and the analyzed environmental parameters are matched to corresponding categories based on the calculated distance between the analyzed environmental parameters and the optimal environmental parameters, and the score of the corresponding category is used as the score of the analyzed environmental parameters.

9. The method for analyzing livestock health breeding environment disease data based on machine learning according to claim 8, characterized in that: The distance between the calculated and analyzed environmental parameters and the optimal environmental parameters comprises the following steps: Among them, d represents the distance between the analyzed environmental parameters and the optimal environmental parameters, x1 represents the horizontal coordinate of the analyzed environmental parameters, y1 represents the vertical coordinate of the analyzed environmental parameters, and x best The horizontal axis represents the optimal environmental parameters, y best The vertical axis represents the optimal environmental parameters.

10. A system for implementing the method for analyzing livestock health breeding environment disease data based on machine learning according to any one of claims 1 to 9, characterized in that: include: Data collection module, scoring and evaluation module, model building module, data analysis platform and data matching module; The data acquisition module is used to collect environmental parameters in the healthy breeding area in real time; The scoring and evaluation module is used to score the collected environmental parameters in the healthy breeding area; The data analysis platform includes a model building module and a data matching module; The model building module is used to build a healthy breeding environment disease data analysis model and output optimal environmental parameters through machine learning; The data matching module is used to perform score matching according to the analysis results.

Images