A livestock and poultry house environmental temperature detection system based on a cloud platform

Abstract

The present invention discloses a livestock and poultry house environmental temperature detection system based on a cloud platform, which is composed of a livestock and poultry house environmental parameter collection and control platform and a livestock and poultry house environmental temperature big data processing subsystem, and realizes the functions of collecting, processing and predicting the livestock and poultry house breeding environmental parameters; the present invention effectively solves the problems that the existing livestock and poultry breeding environmental temperature detection system cannot accurately detect and predict the livestock and poultry breeding environmental temperature due to the non-linear, large lag and complex dynamic changes of the breeding environmental temperature, thus greatly affecting the effective management of the livestock and poultry breeding environmental temperature, and that the temperature parameters are not managed, thus greatly affecting the breeding efficiency of the livestock and poultry house.

Description

Technical Field

[0001] The present invention relates to the technical field of automated equipment for detecting the ambient temperature of livestock and poultry houses, and in particular to a livestock and poultry house ambient temperature detection system based on a cloud platform. Background Art

[0002] Temperature and humidity are important environmental factors that affect the growth of livestock and poultry. When the temperature and humidity in the environment are too high, the bacteria in the environment will grow wildly, resulting in a decrease in the resistance of the livestock and poultry bodies. When the temperature and humidity in the environment are too low, the water vapor in the air will not be able to effectively absorb dust and bacteria, and livestock and poultry are prone to respiratory diseases and other diseases, which may cause death in severe cases.

[0003] Lighting plays an important role in the production performance of poultry and livestock. Under appropriate light intensity, light not only promotes the growth of poultry and livestock, but also has a positive effect on the reproduction and fattening of poultry and livestock. The detection terminal obtains sunlight irradiation data through the light intensity sensor, and its high resolution can detect a large range of light intensity changes in the breeding environment. Most of the practitioners in my country's poultry and livestock farming are family households or small and medium-sized enterprises. Compared with developed countries, there are some contradictions in breeding production, such as low degree of informatization, low technical input cost, and high labor input cost. With the advent of the agricultural informatization era, the combination of modern information technology and traditional breeding industry can effectively improve the intelligence level of poultry and livestock breeding. A cloud platform-based livestock and poultry house environmental temperature detection system is invented to realize the detection of livestock and poultry house environmental parameters and predict the temperature, so as to improve the economic benefits and efficiency of livestock and poultry breeding. Summary of the invention

[0004] The present invention provides a livestock and poultry house environment temperature detection system based on a cloud platform. The present invention effectively solves the problems that the existing livestock and poultry breeding environment temperature detection system cannot accurately detect and predict the livestock and poultry breeding environment temperature according to the characteristics of nonlinearity, large hysteresis and complex dynamic changes of breeding environment temperature changes, thereby greatly affecting the effective management of livestock and poultry breeding environment temperature, and the temperature parameters are not managed, thereby greatly affecting the breeding efficiency of livestock and poultry houses.

[0005] The present invention is achieved through the following technical solutions:

[0006] A livestock house environmental temperature detection system based on a cloud platform consists of a livestock house environmental parameter acquisition and control platform and a livestock house environmental temperature big data processing subsystem. The livestock house environmental parameter detection system platform consists of detection nodes, control nodes, gateway nodes, on-site monitoring terminals, mobile APPs, and a cloud platform. The detection nodes and control nodes are responsible for detecting and controlling the environmental parameters of the livestock house. The environmental parameters of the livestock house are uploaded to the cloud platform through the gateway nodes. Breeding managers can view the livestock breeding environmental data on the cloud platform in real time from the mobile APP, realizing the functions of remote monitoring and intelligent regulation of the environmental parameters of the livestock house. The structure diagram of the livestock house environmental parameter acquisition and control platform is shown in Figure 1 as follows.

[0007] The further technical improvement scheme of the present invention is:

[0008] The livestock house environmental temperature big data processing subsystem includes a temperature detection unit and a temperature level classifier. The outputs of multiple temperature sensors are respectively the inputs of the corresponding multiple tapped delay lines TDL (Tapped Delay Line) of the temperature detection unit. The nursery temperature trapezoidal fuzzy numbers, growth period temperature trapezoidal fuzzy numbers, and fattening period temperature trapezoidal fuzzy numbers output by the temperature detection unit at different growth stages of livestock are used as the inputs of the corresponding 3 tapped delay lines TDL (Tapped Delay Line) of the temperature classifier. The trapezoidal fuzzy number output by the temperature classifier represents the temperature suitability level of the livestock house. The livestock house environmental temperature big data processing subsystem is shown in Figure 2 .

[0009] The further technical improvement scheme of the present invention is:

[0010] The temperature detection unit consists of multiple tapped delay lines (TDLs), multiple NARX neural network temperature models, a DRNN neural network model, a FLNN neural network model, an ANFIS neural network model, three integral circuits, and a GMDH neural network model. Two integral operators S are connected in series to form an integral circuit. The outputs of the connection ends of the two integral operators of each integral circuit serve as one corresponding input of the GMDH neural network model, and the output of each integral circuit serves as one corresponding input of the GMDH neural network model. The outputs of multiple temperature sensors are respectively used as the inputs of multiple corresponding tapped delay lines (TDLs). The temperature sensor values over a period of time output by each tapped delay line (TDL) are respectively used as the inputs of the corresponding NARX neural network temperature models. The outputs of multiple NARX neural network temperature models are respectively used as the inputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model. The outputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model are respectively used as the inputs of the corresponding integral circuits and one corresponding input of the GMDH neural network model. The output of the GMDH neural network model is a trapezoidal fuzzy number representing the magnitudes of multiple temperature sensor values in the livestock and poultry house environment over a period of time. The trapezoidal fuzzy number is [a, b, c, d], and [a, b, c, d] constitutes the trapezoidal fuzzy numerical values of the output of the temperature detection unit as multiple temperature sensor values over a period of time. a, b, c, and d respectively represent the minimum value, minimum extreme value, maximum extreme value, and maximum value of the temperature in the livestock and poultry house environment. The temperature detection unit converts multiple temperature sensor values over a period of time into trapezoidal fuzzy temperature values.

[0011] A further technical improvement scheme of the present invention is:

[0012] The temperature grade classifier is composed of three tapped delay lines (TDL), three dynamic recursive wavelet neural networks, three autoassociative neural networks and RBF neural network classifiers. The temperature trapezoidal fuzzy numbers of the nursery period, the growth period and the fattening period output by the temperature detection unit at different growth stages of livestock and poultry are used as the first, second and third tapped delay lines (TDL) of the temperature classifier. Line), the trapezoidal fuzzy numbers of the livestock and poultry house temperature over a period of time output by the three beat delay lines TDL are used as the inputs of the corresponding first, second and third autoassociative neural networks respectively, the outputs of the three autoassociative neural networks are used as the inputs of the corresponding first, second and third dynamic recursive wavelet neural networks respectively, the output of the first dynamic recursive wavelet neural network is used as the input of the second dynamic recursive wavelet neural network and the corresponding input of the RBF neural network classifier respectively, the output of the second dynamic recursive wavelet neural network is used as the input of the third dynamic recursive wavelet neural network and the corresponding input of the RBF neural network classifier respectively, the third dynamic recursive wavelet neural network is used as the corresponding input of the RBF neural network classifier, the numbers 1-5 representing different types of livestock and poultry are used as one corresponding input of the RBF neural network classifier, where the number 1 represents pigs, the number 2 represents chickens, the number 3 represents beef cattle, the number 4 represents sheep, and the number 5 represents pigeons, and the trapezoidal fuzzy number output by the RBF neural network classifier represents the temperature suitability level;.

[0013] The further technical improvement scheme of the present invention is:

[0014] According to the engineering practice of livestock and poultry house temperature on livestock and poultry growth suitability, the RBF neural network classifier divides the influence of livestock and poultry house environment temperature on livestock and poultry growth process into five suitability levels. The five suitability levels are generally suitable, relatively suitable, very suitable, unsuitable and very unsuitable, corresponding to five different trapezoidal fuzzy numbers respectively. A corresponding relationship table between the five trapezoidal fuzzy numbers and the five suitability levels is constructed, and the similarity between the trapezoidal fuzzy number output by the RBF neural network classifier and the five trapezoidal numbers representing the five suitability levels is calculated. The suitability level corresponding to the trapezoidal fuzzy number with the largest similarity is determined as the suitability level of the livestock and poultry house environment temperature.

[0015] Compared with the prior art, the present invention has the following obvious advantages:

[0016] 1. The FLNN function - connected neural network model of the present invention consists of an input layer and an output layer without a hidden layer. Therefore, compared with traditional neural networks, the FLNN function - connected neural network model has less network computation and faster training speed. It can avoid updating the weights of the hidden layer and only needs to adjust the weights of the output layer, thus having a faster convergence speed and less online computation. At the same time, by expanding the input variables of the temperature parameters in the livestock and poultry breeding environment, the network resolution ability of the FLNN function - connected neural network model can be improved, and the accuracy of detecting the temperature of the livestock and poultry house environment in the present invention can be enhanced.

[0017] 2. The GMDH neural network model of the present invention has the following two basic ideas: dealing with the input - output relationship of the system of the livestock and poultry house environment temperature by the method of analyzing the black box, and describing the function of the network by the interconnection relationship between the elements in the network. The construction process of the GMDH neural network model is mainly a process of continuously generating active neurons, screening the neurons by an external criterion, strongly combining the selected neurons to generate the next - layer neurons until the model with the best complexity is selected. ① It can obtain the model result expressed by a clear function formula. The self - organizing GMDH neural network model combines the ideas of neural networks and statistical modeling and can give the result expressed by a function formula, even a multi - variable high - order regression equation that is difficult to achieve by other modeling methods. ② The modeling process is self - organized and controlled without any initial assumptions. The GMDH neural network model allows hundreds of input variables, and then generates a large number of candidate models layer by layer with a large number of variables. The algorithm finds the input items that have a substantial impact on the explained variable according to data - driven, self - organizes to generate the optimal network structure, and minimizes the influence of the subjective factors of the modeler. ③ Optimal complexity and high - precision prediction. The optimal complex characteristics of the GMDH neural network model ensure that it can make decisions in an approximate, uncertain, and even contradictory knowledge environment. Also, because it avoids over - fitting and under - fitting of the model structure at the same time, the GMDH neural network model is closer to the real situation of the livestock and poultry house environment temperature change system, so the livestock and poultry house environment temperature has higher prediction reliability.

[0018] 3. The present invention uses the NARX neural network livestock and poultry house environment temperature prediction model. Since the NARX neural network establishes a dynamic recursive network of the model by introducing a delay module and output feedback, it introduces the delay feedback of the input and output vectors into the network training to form a new input vector, and has good non - linear mapping ability. The input of the NARX neural network includes not only the original input data of the livestock and poultry house environment temperature but also the temperature output data after training. The generalization ability of the network is improved, making it have better prediction accuracy and adaptive ability in the prediction of the livestock and poultry house environment temperature compared with traditional static neural networks.

[0019] IV. The temperature of the livestock and poultry house environment in the present invention has characteristics such as non-linearity, large lag, and complex dynamic changes. The temperature sensors for measuring the livestock and poultry house environment are easily interfered, so there is often a large amount of noise in the measurement of the livestock and poultry house environment temperature. On the other hand, the measured temperature of the livestock and poultry house environment is more than the number of its independent variables, that is, there is redundant information in these measured temperatures. Through the compression and decompression process of the livestock and poultry house environment temperature information, the auto-associative neural network can use the redundant information to suppress its measurement noise. In the process of processing the big data of the livestock and poultry house environment, applying the auto-associative neural network to preprocess the measured temperature can greatly improve the accuracy of the livestock and poultry house environment temperature.

[0020] V. Regarding the scientificity and reliability of the classification of the temperature suitability level of the livestock and poultry house environment in the present invention, the RBF neural network classifier of this patent classifies the temperature suitability level of the livestock and poultry house environment. The temperature suitability level of the livestock and poultry house environment is based on the size of the trapezoidal fuzzy number of the temperature suitability for livestock growth in the livestock and poultry house environment. According to the engineering practice experience of the temperature control of the livestock and poultry house environment, through the RBF neural network classifier, the dynamic influence of the temperature in the nursery period, growth period, and fattening period of the livestock and poultry house on livestock growth is quantitatively converted into a suitability level. Through the trapezoidal fuzzy number, the livestock and poultry house environment temperature is divided into five situations, and the five suitability levels of the livestock and poultry house environment temperature are respectively corresponding to five different trapezoidal fuzzy numbers: generally suitable, relatively suitable, very suitable, unsuitable, and very unsuitable. Calculate the similarity between the trapezoidal fuzzy number output by the RBF neural network classifier and the five trapezoidal fuzzy numbers representing the five suitability levels, and determine the suitability level corresponding to the trapezoidal fuzzy number of the similarity as the temperature suitability level of the livestock and poultry house environment, realizing the dynamic performance and scientific classification of the classification of the temperature suitability level of the livestock and poultry house environment.

[0021] VI. In the present invention, since the first and second change amounts of the predicted value of the livestock and poultry house temperature parameter are introduced through 3 integral circuits, applying the GMDH neural network model in the time series prediction of non-linear parameters has better prediction accuracy and adaptive ability to convert the detected parameter into a trapezoidal fuzzy number according to the predicted value of the detected parameter and the influence of the change amount, improving the generalization ability of the GMDH neural network model. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is the acquisition and control platform for the livestock and poultry house environment parameters of the present invention;

[0023] Figure 2 is the big data processing subsystem for the livestock and poultry house environment temperature of the present invention;

[0024] Figure 3 is the temperature level classifier of the present invention;

[0025] Figure 4 is the detection node of the present invention;

[0026] Figure 5 is the control node of the present invention;

[0027] Figure 6 is the gateway node of the present invention;

[0028] Figure 7 is the on-site monitoring terminal software of the present invention. Specific embodiments

[0029] Combined with the attached Figures 1-7 , the technical solution of the present invention is further described:

[0030] I. Design of the overall system functions

[0031] The present invention consists of a livestock and poultry house environmental parameter collection and control platform and a livestock and poultry house environmental temperature big data processing subsystem. The livestock and poultry house environmental parameter collection and control platform includes a detection node, a control node, a gateway node, an on-site monitoring terminal, a cloud platform and a mobile APP for the environmental parameters of the livestock and poultry house. Communication between the detection node, the control node and the gateway node is realized by constructing a CAN communication network; the detection node sends the detected environmental parameters of the livestock and poultry house to the on-site monitoring terminal through the RS232 interface of the gateway node, and the on-site monitoring terminal processes the sensor data and predicts the temperature; the gateway node realizes the bidirectional transmission of the environmental parameters of the livestock and poultry house between the NB-IoT module and the cloud platform and between the cloud platform and the mobile APP through the 5G network, and realizes the bidirectional transmission of the environmental parameters of the livestock and poultry house between the gateway node and the on-site monitoring terminal through the RS232 interface. The mobile APP provides real-time livestock and poultry house environmental data, warning management, and query of historical data for managers, meeting the convenient visualization of livestock and poultry house environmental data information. All the data collected by the sensors from the detection node have been uploaded to the database of the cloud platform, and managers can remotely view the current environmental parameters of the livestock and poultry house through the mobile APP. The cloud platform realizes functions such as user management, livestock and poultry house environmental data management, real-time monitoring and alarm. The cloud platform is mainly responsible for processing, storing, analyzing and displaying the received livestock and poultry house environmental information. The interaction between the cloud platform and users is mainly through the web page end and the mobile device end. The completeness of the web page end functions and the convenience of the mobile end operation maximize the efficient interaction between the cloud platform and users. User management provides management operations such as registering an account, logging in to an account, and changing account information for managers; data management provides operations such as querying historical data and classifying data management for users; the real-time monitoring function processes the parameters into visual data and can present the data changes in an easy-to-analyze manner through a histogram or a line chart. The structure of the livestock and poultry house environmental parameter collection and control platform is shown in Figure 1 as shown.

[0032] II. Design of the detection node

[0033] The detection node consists of a sensor, a conditioning circuit, an STM32 single-chip microcomputer, and a CAN bus interface. It is mainly used to collect environmental parameters of livestock and poultry houses by collecting temperature sensors, humidity sensors, light intensity sensors, and wind speed sensors in the livestock and poultry environment. The environmental data information will realize real-time interaction of information between the detection node and the gateway node through the CAN bus interface of the detection node and the CAN bus interface of the gateway node.

[0034] III. Control Node Design

[0035] The control node consists of a CAN bus interface, an STM32 single-chip microcomputer, a temperature control device, a humidity control device, a light intensity control device, and a wind speed control device. It regulates the stability of various factor data in the livestock and poultry house breeding environment by adjusting the operation of equipment for heating, humidifying, and ventilation. After the microprocessor of the control node receives the set values of environmental parameters sent by the cloud platform management personnel, it will control the operation of the temperature control device, humidity control device, wind speed control device, and light intensity control device through relays to complete the regulation of the livestock and poultry house environment. The two-way data communication between the control node and the gateway node is through the CAN bus interface. To ensure that the livestock and poultry house environment is always in a more suitable environment, the operating state and operating efficiency of the equipment should be in a dynamic adjustment.

[0036] IV. Gateway Node Design

[0037] The gateway node consists of a CAN bus interface, an NB-IoT module, an STM32 single-chip microcomputer, and an RS232 interface. It realizes two-way data transmission between the detection node and the control node and the on-site monitoring terminal through the CAN bus interface and the RS232 interface, and realizes two-way transmission between the cloud platform, the mobile APP, the detection node, the control node, and the on-site monitoring terminal through the CAN bus interface, the NB-IoT module, and the RS232 interface.

[0038] V. On-site Monitoring Terminal Software Design

[0039] The on-site monitoring terminal is an industrial control computer. The on-site monitoring terminal mainly realizes the collection of livestock and poultry house parameters and the prediction of the livestock and poultry house environment temperature, and realizes information interaction with the gateway node. The main functions of the on-site monitoring terminal are communication parameter setting, data analysis and data management, and the big data processing subsystem of the livestock and poultry house environment temperature. The big data processing subsystem of the livestock and poultry house environment temperature includes a temperature detection unit and a temperature level classifier. The structure of the big data processing subsystem of the livestock and poultry house environment temperature is shown in Figure 2 . This management software selects Microsoft Visual++6.0 as the development tool and calls the Mscomm communication control of the system to design the communication program. The functions of the on-site monitoring terminal software are shown in Figure 7The big data processing subsystem for the environmental temperature of livestock and poultry houses includes a temperature detection unit and a temperature level classifier. The outputs of multiple temperature sensors are respectively the inputs of multiple corresponding tapped delay lines (TDLs) of the temperature detection unit. The nursery temperature trapezoidal fuzzy numbers, growth period temperature trapezoidal fuzzy numbers, and fattening period temperature trapezoidal fuzzy numbers output by the temperature detection unit during different growth stages of livestock and poultry are used as the inputs of 3 corresponding tapped delay lines (TDLs) of the temperature classifier. The trapezoidal fuzzy number output by the temperature classifier represents the temperature suitability level of the livestock and poultry house. The characteristics of the temperature detection unit and the temperature level classifier are as follows:

[0040] 1. Design of the temperature detection unit

[0041] The temperature detection unit consists of multiple tapped delay lines (TDLs), multiple NARX neural network temperature models, a DRNN neural network model, a FLNN neural network model, an ANFIS neural network model, 3 integral circuits, and a GMDH neural network model. Two integral operators S are connected in series to form an integral circuit. The outputs of the connection ends of the 2 integral operators of each integral circuit are used as 1 corresponding input of the GMDH neural network model, and the output of each integral circuit is used as 1 corresponding input of the GMDH neural network model. The outputs of multiple temperature sensors are respectively used as the inputs of multiple corresponding tapped delay lines (TDLs). The temperature sensor values over a period of time output by each tapped delay line (TDL) are respectively used as the inputs of the corresponding NARX neural network temperature models. The outputs of multiple NARX neural network temperature models are used as the inputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model. The outputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model are respectively used as the inputs of the corresponding integral circuits and 1 corresponding input of the GMDH neural network model. The output of the GMDH neural network model is a trapezoidal fuzzy number representing the magnitudes of multiple temperature sensor values in the livestock and poultry house environment over a period of time. The trapezoidal fuzzy number is [a, b, c, d], and [a, b, c, d] constitutes the trapezoidal fuzzy numerical values of the outputs of multiple temperature sensor values by the temperature detection unit over a period of time. a, b, c, and d respectively represent the minimum value, minimum extreme value, maximum extreme value, and maximum value of the environmental temperature of the livestock and poultry house. The temperature detection unit converts the temperature sensor values over multiple time periods into temperature trapezoidal fuzzy numerical values;

[0042] A. Design of the NARX neural network temperature model

[0043] The outputs of multiple NARX neural network temperature models are used as the inputs of the DRNN neural network model, the FLNN function - linked neural network model, and the ANFIS neural network model. The outputs of the DRNN neural network model, the FLNN function - linked neural network model, and the ANFIS neural network model are respectively used as the inputs of each corresponding integral loop and one corresponding input of the GMDH neural network model. The NARX neural network temperature model (Nonlinear Auto - Regression with External input neural network) is a dynamic feed - forward neural network. The NARX neural network temperature model is a nonlinear autoregressive network with external input. It has a dynamic characteristic of multi - step time delay and closes several layers of the network through feedback connections. The recurrent neural network of the NARX neural network temperature model is one of the most widely used dynamic neural networks in nonlinear dynamic systems, and its performance is generally better than that of the full - recurrent neural network. A typical NARX neural network temperature model mainly consists of an input layer, a hidden layer, an output layer, and input and output delays. Generally, the delay orders of the input and output and the number of hidden - layer neurons need to be determined in advance before application. The current output of the NARX neural network temperature model not only depends on the past output y(t - n), but also depends on the current input temperature vector X(t) and the delay order of the input temperature vector, etc. Among them, the input temperature signal is transmitted to the hidden layer through the time - delay layer. The hidden layer processes the input temperature signal and then transmits it to the output layer. The output layer linearly weights the output signal of the hidden layer to obtain the final neural network output signal. The time - delay layer delays the signal fed back by the network and the signal output by the input layer, and then transports it to the hidden layer. The NARX neural network temperature model has characteristics such as nonlinear mapping ability, good robustness and self - adaptability, and is suitable for predicting the temperature of the livestock and poultry breeding environment. x(t) represents the external input of the neural network temperature model, that is, the value of the temperature sensor in the livestock and poultry breeding environment; m represents the delay order of the external input; y(t) is the output of the neural network, that is, the predicted value of the temperature in the livestock and poultry breeding environment in the next time period; n is the output delay order; s is the number of neurons in the hidden layer; the output of the j - th hidden unit can be obtained as follows:

[0044]

[0045] In the above formula, w ji is the connection weight between the i - th input and the j - th hidden neuron, and b j is the bias value of the j - th hidden neuron. The value of the network output y(t + 1) is:

[0046] y(t + 1) = f[y(t), y(t - 1), …, y(t - n), x(t), x(t - 1), …, x(t - m + 1); W] (2)

[0047] B. DRNN Neural Network Design

[0048] The outputs of multiple NARX neural network temperature models serve as the inputs to the DRNN neural network model, FLNN function - linked neural network model, and ANFIS neural network model. The outputs of the DRNN neural network model, FLNN function - linked neural network model, and ANFIS neural network model are respectively used as the inputs to each corresponding integral circuit and one corresponding input to the GMDH neural network model. The output of the DRNN neural network model is used as the input to the corresponding integral circuit and one corresponding input to the GMDH neural network model. The DRNN neural network model is a dynamic regression neural network with feedback and the ability to adapt to time - varying characteristics. This network can more directly and vividly represent the dynamic change performance of the detection parameters of the livestock and poultry breeding environment. The DRNN neural network model can more accurately measure the temperature of the livestock and poultry breeding environment. The network structure of the DRNN neural network model is a three - layer network structure of n - 2n+1 - 1, and its hidden layer is a dynamic regression layer. Let \(I = [I_1(t),I_2(t),\cdots,I n (t)]\) be the input vector of the DRNN neural network model, where \(I i (t)\) is the input of the \(i\) - th neuron at time \(t\) in the input layer of the multi - DRNN neural network model, and the output of the \(j\) - th neuron in the regression layer is \(X j (t)\), \(S j (t)\) is the sum of the inputs to the \(j\) - th regression neuron, \(f(\cdot)\) is a function of \(S\), then \(O(t)\) is the output of the DRNN neural network model. Then the output of the output layer of the DRNN neural network model is:

[0049]

[0050] C. FLNN Function - Linked Neural Network Model Design

[0051] The outputs of multiple NARX neural network temperature models are respectively used as the inputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model. The outputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model are respectively used as the inputs of each corresponding integral loop and one corresponding input of the GMDH neural network model. The output of the parameter measurement sensor is used as the input of the corresponding tapped delay line TDL. The parameter measurement sensor values for a period of time output by each tapped delay line TDL are respectively used as the inputs of the corresponding FLNN function-connected neural network model. The outputs of multiple FLNN function-connected neural network models are respectively used as the inputs of multiple DRNN neural network models. The FLNN function-connected neural network is a function-type neural network model. The role of the function connection in this model is to multiply each component of the input pattern of the detection parameters of the livestock and poultry breeding environment by the entire pattern vector, and the result is to generate a tensor product of the original pattern vector. The FLNN function-connected neural network pre-nonlinearly expands the input pattern of the detection parameters of the livestock and poultry breeding environment, introduces "higher-order" terms in the FLNN function-connected neural network, and maps the input pattern of the detection parameters of the livestock and poultry breeding environment to a larger pattern space through the non-linear expansion of the input pattern of the detection parameters of the livestock and poultry breeding environment, enhancing the pattern expression of the input signal of the detection parameters of the livestock and poultry breeding environment and greatly simplifying the network structure of the FLNN function-connected neural network model. Although the information of the detection parameters of the livestock and poultry breeding environment input to the FLNN function-connected neural network model does not increase, the enhancement of the FLNN function-connected neural network model pattern brings about the simplification of the network structure of the FLNN function-connected neural network model and the improvement of the learning speed. "Supervised" learning can be achieved with a single-layer network, which has great advantages compared to multi-layer feedforward neural networks. The FLNN function-connected neural network model uses a single-layer network to achieve supervised learning, and this solution process can be completed by the following adaptive supervised learning algorithm. The learning algorithm of the FLNN function-connected neural network model can be expressed by the following formula:

[0052]

[0053] Weight adjustment:

[0054]

[0055] Where: F i (k), e i (k) and w n(k) is the expected output, estimated output, error of the i-th input pattern, and the n-th connection weight of the functional neural network at the k-th step; α is the learning factor, which affects stability and convergence speed. The FLNN functional connection neural network model uses the method of function expansion to expand the input of the detection parameters of the original livestock and poultry breeding environment, so that the input of the detection parameters of the original livestock and poultry breeding environment is transformed into another space, and the enhanced pattern is used as the input of the network input layer of the FLNN functional connection neural network model. Through this method, nonlinear problems can be better handled; the FLNN functional connection neural network model consists of an input layer and an output layer, without a hidden layer. Therefore, compared with traditional neural networks, the FLNN functional connection neural network model has less network calculation and faster training speed. It can avoid updating the hidden layer weights and only needs to adjust the output layer weights, so it has a faster convergence speed and less online calculation. At the same time, expanding the input variables of the detection parameters of the livestock and poultry breeding environment can improve the network resolution ability of the FLNN functional connection neural network model.

[0056] D. Design of ANFIS Neural Network Model

[0057] The outputs of multiple NARX neural network temperature models are respectively used as the inputs of the DRNN neural network model, FLNN neural network model, and ANFIS neural network model. The outputs of the DRNN neural network model, FLNN neural network model, and ANFIS neural network model are respectively used as the inputs of the corresponding each integral loop and one corresponding input of the GMDH neural network model. The ANFIS neural network model is an adaptive fuzzy inference system ANFIS based on neural networks, also known as Adaptive Neuro-Fuzzy Inference System, which organically combines neural networks and adaptive fuzzy inference systems, can not only give full play to the advantages of both, but also make up for their respective deficiencies. The fuzzy membership function and fuzzy rules in the ANFIS neural network are obtained through the learning of a large amount of known historical data of the livestock and poultry house environment. The biggest feature of the ANFIS neural network model is the data-based modeling method, rather than arbitrarily given based on experience or intuition. The input of the ANFIS neural network model is the output value of the NARX neural network, and the output of the ANFIS neural network model is the re-predicted value of the livestock and poultry house environment temperature. The main operation steps are as follows:

[0058] Layer 1: Fuzzify the livestock and poultry house values of the output of the input NARX neural network. The output corresponding to each node can be expressed as:

[0059]

[0060] Equation n is the number of membership functions for each input, and the membership function uses a Gaussian membership function.

[0061] Layer 2: Implement rule operations and output the applicability of the rules. The rule operations of the ANFIS neural network model use multiplication.

[0062]

[0063] Layer 3: Normalize the applicability of each rule:

[0064]

[0065] Layer 4: The transfer function of each node is a linear function, representing a local linear model. The output of each adaptive node i is:

[0066]

[0067] Layer 5: The single node in this layer is a fixed node, and the output of the ANFIS neural network model is calculated as:

[0068]

[0069] In the ANFIS neural network model, the conditional parameters that determine the shape of the membership function and the conclusion parameters of the inference rules can be trained through the learning process. The parameters are adjusted using an algorithm that combines linear least squares estimation and gradient descent. In each iteration of the ANFIS neural network, the input signal is first propagated forward along the network until the fourth layer, and the least squares estimation algorithm is used to adjust the conclusion parameters. The signal continues to be propagated forward along the network until the output layer. The error signal obtained by the ANFIS neural network model is propagated backward along the network, and the conditional parameters are updated using the gradient method. By adjusting the given conditional parameters in the ANFIS neural network model in this way, the global optimal point of the conclusion parameters can be obtained, which can not only reduce the dimensionality of the search space in the gradient method but also improve the convergence speed of the parameters of the ANFIS neural network model. The output of the ANFIS neural network model is the fusion value of the output values of multiple NARX neural networks.

[0070] E. Design of the GMDH neural network model

[0071] The outputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model are respectively used as the inputs for each corresponding integral circuit and one corresponding input of the GMDH neural network model. The output of the GMDH neural network model is a trapezoidal fuzzy number representing the magnitudes of multiple temperature sensor values in the livestock and poultry house environment for a period of time. The trapezoidal fuzzy number is [a, b, c, d], and [a, b, c, d] constitutes the trapezoidal fuzzy numerical value of the outputs of multiple temperature sensors by the temperature detection unit for a period of time. a, b, c, and d respectively represent the minimum value, the minimum extreme value, the maximum extreme value, and the maximum value of the temperature in the livestock and poultry house environment. The temperature detection unit converts the temperature sensor values at multiple times into trapezoidal fuzzy temperature values. The GMDH neural network model (GMDH) is an algorithm for self-organizing data mining. If the GMDH neural network model has m input variables x1, x2, …, x m and the output is Y. The purpose of GMDH is to establish a functional relationship f with undetermined coefficients and a known form between the input and the output. The function f can be approximated by applying a polynomial expanded by the Volterra series:

[0072]

[0073] The GMDH neural network model is mainly used to process small sample data and construct a prediction model for the livestock and poultry house environment parameters by automatically finding the correlations between variables in the samples. First, the first-generation intermediate candidate models are generated according to the initial model of the reference function, and then several items are selected from the first-generation intermediate candidate models and combined with calculation rules to generate the second-generation intermediate candidate models. This process is repeated until the optimal prediction model for the livestock and poultry house environment parameters is obtained. Therefore, the GMDH neural network model can adaptively establish a high-order polynomial model that has an explanatory ability for the dependent variable according to the independent variables. Let R j be the maximum number of neurons in the j-th layer, x kl be the k-th dimension of the l-th input sample, y jkl be the predicted value of the k-th neuron in the j-th layer of the network for the l-th input sample, be the root mean square of the thresholds of the k-th neuron in the j-th layer of the network, and Y be the predicted value of the network. The GMDH neural network model constructs the network structure by using an adaptive multi-layer iterative method, selects the optimal model of the network according to the minimum deviation criterion, and constructs a non-linear mapping between the input and the output based on the Kolmogorov-Gabor polynomial. Data preprocessing divides the data set into a training set and a test set; pairs the input quantities, and identifies the local polynomial models to generate a set of competing models, calculates the selection criterion value as the input for the next layer until the optimal complexity model is selected. The learning and evolution process of the GMDH neural network model is as follows: ① Set the maximum number of neurons R jGiven the initial number of variables \(d_0\) of the network, select the minimum deviation criterion of the network. ② Construct an initial network containing only the neurons in the first layer according to the dimension of the input data. ③ Calculate the root mean square of the thresholds of each neuron in turn. For the \(j\)-th layer of the network, sort from largest to smallest. Select the first \(R\) j ones as the selected neurons to be retained, and the rest as the unselected neurons. For the selected neurons, find the minimum and compare it with the minimum of the previous layer If is less than then execute step ④, otherwise execute step ⑤. ④ Generate the neurons in the next layer according to the currently selected neurons. ⑤ The network construction is completed.

[0074] 2. Design of Temperature Grade Classifier

[0075] The temperature grade classifier consists of 3 tapped delay lines (TDLs), 3 dynamic recursive wavelet neural networks, 3 self-associative neural networks, and an RBF neural network classifier. The trapezoidal fuzzy numbers of the conservation period temperature, growth period temperature, and fattening period temperature output by the temperature detection unit at different growth stages of livestock and poultry are used as the inputs of the corresponding first, second, and third tapped delay lines (TDLs) of the temperature classifier respectively. The trapezoidal fuzzy numbers of the livestock house temperature for a period of time output by the 3 tapped delay lines (TDLs) are used as the inputs of the corresponding first, second, and third self-associative neural networks respectively. The outputs of the 3 self-associative neural networks are used as the inputs of the corresponding first, second, and third dynamic recursive wavelet neural networks respectively. The output of the first dynamic recursive wavelet neural network is used as the input of the second dynamic recursive wavelet neural network and the corresponding input of the RBF neural network classifier respectively. The output of the second dynamic recursive wavelet neural network is used as the input of the third dynamic recursive wavelet neural network and the corresponding input of the RBF neural network classifier respectively. The third dynamic recursive wavelet neural network is used as the corresponding input of the RBF neural network classifier. The numbers 1 - 5 representing different livestock and poultry species are used as one of the corresponding inputs of the RBF neural network classifier, where the number 1 represents live pigs, the number 2 represents chickens, the number 3 represents beef cattle, the number 4 represents sheep, and the number 5 represents pigeons. The trapezoidal fuzzy number output by the RBF neural network classifier represents the temperature suitability grade. The design process of the temperature grade classifier is as follows:

[0076] A. Design of Self-Associative Neural Network

[0077] The temperature trapezoidal fuzzy numbers of the conservation period, growth period, and fattening period output by the temperature detection unit at different growth stages of livestock and poultry are respectively used as the inputs of the 1st, 2nd, and 3rd tapped delay lines (TDL) of the temperature classifier corresponding to the tapped delay line TDL. The trapezoidal fuzzy numbers of the livestock and poultry house temperature for a period of time output by the 3 tapped delay lines TDL are respectively used as the inputs of the corresponding 1st, 2nd, and 3rd auto-associative neural networks. The trapezoidal fuzzy numbers of the temperature output by the 3 auto-associative neural networks are respectively used as the inputs of the corresponding 1st, 2nd, and 3rd dynamic recursive wavelet neural networks. The auto-associative neural network (AANN), a feedforward neural network with a special structure, the structure of the auto-associative neural network includes an input layer, a certain number of hidden layers, and an output layer. First, through the input layer, mapping layer, and bottleneck layer, the compression of the input data information of the trapezoidal fuzzy numbers of the livestock and poultry house environmental temperature is realized. The most representative low-dimensional subspace reflecting the system structure of the trapezoidal fuzzy numbers of the livestock and poultry house environmental temperature is extracted from the high-dimensional parameter space of the network input. At the same time, the noise and measurement errors in the input data of the trapezoidal fuzzy numbers of the livestock and poultry house environmental temperature are effectively filtered. Then, through the bottleneck layer, demapping layer, and output layer, the decompression of the trapezoidal fuzzy numbers of the livestock and poultry house environmental temperature data is realized, and the previously compressed information is restored to each parameter value, so as to realize the reconstruction of the input data of the trapezoidal fuzzy numbers of the livestock and poultry house environmental temperature. In order to achieve the purpose of information compression, the number of nodes in the bottleneck layer of the auto-associative neural network is significantly less than that of the input layer. In order to prevent the formation of a simple single mapping between the input and output layers, except that the activation function of the output layer uses a linear function, other layers all use non-linear activation functions. Essentially, the first layer of the hidden layer of the auto-associative neural network is called the mapping layer, and the node transfer function of the mapping layer may be an S-shaped function or other similar non-linear functions; the second layer of the hidden layer is called the bottleneck layer, and the dimension of the bottleneck layer is the smallest in the network. Its transfer function may be linear or non-linear. The bottleneck layer avoids the one-to-one mapping relationship where the output and input are easily equal. It enables the network to encode and compress the trapezoidal fuzzy number signal of the livestock and poultry house environmental temperature to obtain a relevant model of the input temperature sensor data, and perform decoding and decompression after the bottleneck layer to generate an estimated value of the trapezoidal fuzzy number input signal of the livestock and poultry house environmental temperature; the third layer or the last layer of the hidden layer is called the demapping layer, and the node transfer function of the demapping layer is usually a non-linear S-shaped function. The auto-associative neural network is trained using the error backpropagation algorithm.

[0078] B. Design of the dynamic recursive wavelet neural network prediction model

[0079] The outputs of the 3 self-associative neural networks are respectively used as the inputs of the corresponding 1st, 2nd, and 3rd dynamic recurrent wavelet neural networks. The output of the 1st dynamic recurrent wavelet neural network is respectively used as the input of the 2nd dynamic recurrent wavelet neural network and the corresponding input of the RBF neural network classifier. The output of the 2nd dynamic recurrent wavelet neural network is respectively used as the input of the 3rd dynamic recurrent wavelet neural network and the corresponding input of the RBF neural network classifier. The output of the 3rd dynamic recurrent wavelet neural network is used as the corresponding input of the RBF neural network classifier. The theoretical basis of the wavelet neural network WNN (Wavelet Neural Networks) is a feedforward network proposed by combining the wavelet function as the activation function of neurons with the artificial neural network. In the wavelet neural network, the scaling, translation factors of the wavelet, and connection weights are adaptively adjusted during the optimization process of the error energy function. Let the input signal of the wavelet neural network be represented as the one-dimensional vector x of the input i (i = 1, 2, …, n), and the output signal be represented as y k (k = 1, 2, …, m). The calculation formula for the predicted value of the output layer of the wavelet neural network prediction model is:

[0080]

[0081] In the formula, ω ij is the connection weight between the i-th node of the input layer and the j-th node of the hidden layer, is the wavelet basis function, b j is the translation factor of the wavelet basis function, a j is the scaling factor of the wavelet basis function, ω jkis the connection weight between the j-th node of the hidden layer and the k-th node of the output layer. The difference between the dynamic recursive wavelet neural network prediction model of this patent and the ordinary static wavelet neural network is that the dynamic recursive wavelet neural network prediction model has two associated layer nodes that play the role of storing the "internal state" of the network. A self-feedback loop with a fixed gain is added to the two associated layer nodes to enhance the memory performance of time series feature information, thereby enhancing the tracking accuracy of the evolution trajectory of the livestock and poultry house breeding output to ensure better prediction accuracy. The first associated layer node is used to store the state of the hidden layer nodes at the previous moment phase point and then pass it to the hidden layer nodes at the next moment. The second associated layer node is used to store the state of the output layer nodes at the previous moment phase point and then pass it to the hidden layer nodes at the next moment. The feedback information of the neurons in the hidden layer and the output layer will affect the dynamic processing ability of the dynamic recursive wavelet neural network prediction model. The two associated layers both belong to the state feedback inside the dynamic recursive wavelet neural network prediction model, forming the unique dynamic memory performance of the recursiveness of the dynamic recursive wavelet neural network prediction model, and improving the accuracy and dynamic performance of the dynamic recursive wavelet neural network prediction model for predicting the environmental temperature of the livestock and poultry house. A set of connection weights is added between the first associated layer node and the output layer node of the dynamic recursive wavelet neural network prediction model to enhance the dynamic approximation ability of the dynamic recursive wavelet neural network prediction model for predicting the environmental temperature of the livestock and poultry house and improve the prediction accuracy of the environmental temperature of the livestock and poultry house. The weight and threshold correction algorithm of the dynamic recursive wavelet neural network livestock and poultry house environmental temperature prediction model in this patent uses the gradient correction method to update the network weights and wavelet basis function parameters, so that the output of the dynamic recursive wavelet neural network livestock and poultry house environmental temperature prediction model continuously approaches the expected output.

[0082] C. Design of RBF neural network classifier

[0083] The outputs of the 3 self-associative neural networks are respectively used as the inputs of the corresponding 1st, 2nd, and 3rd dynamic recursive wavelet neural networks. The output of the 1st dynamic recursive wavelet neural network is respectively used as the input of the 2nd dynamic recursive wavelet neural network and the corresponding input of the RBF neural network classifier. The output of the 2nd dynamic recursive wavelet neural network is respectively used as the input of the 3rd dynamic recursive wavelet neural network and the corresponding input of the RBF neural network classifier. The 3rd dynamic recursive wavelet neural network is used as the corresponding input of the RBF neural network classifier. The numbers 1-5 representing different livestock and poultry species are used as one corresponding input of the RBF neural network classifier, where the number 1 represents live pigs, the number 2 represents chickens, the number 3 represents beef cattle, the number 4 represents sheep, and the number 5 represents pigeons. The trapezoidal fuzzy number output by the RBF neural network classifier represents the temperature suitability level; the radial basis vectors of the RBF neural network classifier are H = [h1, h2, …, h p T , h p ​Taking it as the basis function, the commonly used radial basis function in the radial basis neural network is the Gaussian function, and its expression is:

[0084]

[0085] In the formula, X is the time series output of the outputs of two tapped delay lines TDL, C is the center point coordinate vector of the Gaussian basis function of the hidden layer neurons, and δ j is the width of the Gaussian basis function of the j-th neuron in the hidden layer; the output connection weight vector of the network is w ij , and the output expression of the RBF neural network classifier is:

[0086]

[0087] The trapezoidal fuzzy number output by the RBF neural network classifier represents the suitability level value of the environmental temperature in the livestock and poultry house; according to the engineering practice of the suitability of the livestock and poultry house temperature for the growth of livestock and poultry, the RBF neural network classifier divides the influence degree of the environmental temperature in the livestock and poultry house on the growth process of livestock and poultry into 5 suitability levels. The 5 suitability levels are generally suitable, relatively suitable, very suitable, unsuitable, and very unsuitable, respectively corresponding to 5 different trapezoidal fuzzy numbers. A corresponding relationship table between 5 trapezoidal fuzzy numbers and 5 suitability levels is constructed, and the similarity between the trapezoidal fuzzy number output by the RBF neural network classifier and the 5 trapezoidal numbers representing 5 suitability levels is calculated. Among them, the suitability level corresponding to the trapezoidal fuzzy number with the largest similarity is determined as the suitability level of the environmental temperature in the livestock and poultry house. The corresponding relationship between the livestock and poultry house temperature and the trapezoidal fuzzy number for the suitability of livestock and poultry growth is as follows, see Table 1.

[0088] Table 1 Corresponding relationship table between the suitability level of the environmental temperature in the livestock and poultry house and the trapezoidal fuzzy number

[0089] Serial number Suitability level Trapezoidal fuzzy number 1 Generally suitable (0.0，0.05，0.15，0.3) 2 Relatively suitable (0.1，0.15，0.3，0.4) 3 Highly suitable (0.3，0.35，0.45，0.7) 4 Unsuitable (0.6，0.75，0.8，0.9) 5 Highly unsuitable (0.8，0.85，0.9，1.0)

[0090] V. Design example of a livestock and poultry house environmental temperature detection system based on a cloud platform

[0091] According to the actual situation of the livestock and poultry house big data detection system, the system arranges the detection nodes, gateway nodes, and the plane layout installation diagram of the on-site monitoring terminal of the livestock and poultry house parameter acquisition platform. Among them, the sensors of the detection nodes are evenly arranged in all directions of the livestock and poultry house according to the detection needs, and the acquisition of livestock and poultry house parameters is realized through this system.

[0092] The technical means disclosed in the solution of the present invention are not limited to the technical means disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features. It should be noted that for those of ordinary skill in the art in this technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A livestock and poultry house environmental temperature detection system based on a cloud platform, characterized in that: The detection system consists of a livestock and poultry house environment parameter acquisition and control platform and a big data processing subsystem for the livestock and poultry house environment temperature, realizing the functions of collecting, processing, and predicting the breeding environment parameters of the livestock and poultry house; The big data processing subsystem for the livestock and poultry house environment temperature includes a temperature detection unit and a temperature grade classifier. The outputs of multiple temperature sensors are respectively the inputs of multiple corresponding tapped delay lines (TDLs) of the temperature detection unit. The nursery temperature trapezoidal fuzzy numbers, growth period temperature trapezoidal fuzzy numbers, and fattening period temperature trapezoidal fuzzy numbers output by the temperature detection unit at different growth stages of livestock and poultry are used as the inputs of the corresponding tapped delay lines (TDLs) of the temperature classifier. The trapezoidal fuzzy numbers output by the temperature classifier represent the temperature suitability grade of the livestock and poultry house; The temperature detection unit consists of a tapped delay line (TDL), a NARX neural network temperature model, a DRNN neural network model, a FLNN neural network model, an ANFIS neural network model, an integration circuit, and a GMDH neural network model. Two integration operators S are connected in series to form an integration circuit. The outputs of the connection ends of the two integration operators of each integration circuit are used as the corresponding inputs of the GMDH neural network model, and the output of each integration circuit is used as the corresponding input of the GMDH neural network model; the outputs of the temperature sensors are respectively used as the inputs of multiple corresponding tapped delay lines (TDLs). The temperature sensor values over a period of time output by each tapped delay line (TDL) are respectively used as the inputs of the corresponding NARX neural network temperature model. The outputs of the NARX neural network temperature model are respectively used as the inputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model. The outputs of the DRNN neural network model, the FLNN neural network model, and the ANFIS neural network model are respectively used as the inputs of the corresponding each integration circuit and the corresponding inputs of the GMDH neural network model. The output of the GMDH neural network model is a trapezoidal fuzzy number representing the magnitudes of multiple temperature sensor values of the livestock and poultry house environment over a period of time and is used as the output of the temperature detection unit. The temperature detection unit converts the multiple temperature sensor values over multiple time periods into temperature trapezoidal fuzzy values; The temperature grade classifier includes a tapped delay line (TDL), a dynamic recurrent wavelet neural network, a self-associative neural network, and an RBF neural network classifier. The nursery temperature trapezoidal fuzzy numbers, growth period temperature trapezoidal fuzzy numbers, and fattening period temperature trapezoidal fuzzy numbers output by the temperature detection unit at different growth stages of livestock and poultry are respectively used as the inputs of the corresponding tapped delay lines (TDLs) of the temperature classifier. The trapezoidal fuzzy numbers of the livestock and poultry house temperature over a period of time output by the tapped delay line (TDL) are respectively used as the inputs of the corresponding self-associative neural networks. The outputs of the self-associative neural networks are respectively used as the inputs of the corresponding dynamic recurrent wavelet neural networks; The outputs of the first dynamic recurrent wavelet neural network are respectively used as the inputs of the second dynamic recurrent wavelet neural network and the corresponding inputs of the RBF neural network classifier. The outputs of the second dynamic recurrent wavelet neural network are respectively used as the inputs of the third dynamic recurrent wavelet neural network and the corresponding inputs of the RBF neural network classifier. The output of the third dynamic recurrent wavelet neural network is used as the corresponding input of the RBF neural network classifier, and the numbers representing different livestock and poultry species are used as the corresponding inputs of the RBF neural network classifier. The trapezoidal fuzzy numbers output by the RBF neural network classifier represent the temperature suitability levels.

2. The livestock and poultry house environment temperature detection system based on a cloud platform according to claim 1, characterized in that: The RBF neural network classifier divides the influence degree of the livestock and poultry house environmental temperature on the growth process of livestock and poultry into 5 suitability levels. The 5 suitability levels are generally suitable, relatively suitable, very suitable, unsuitable, and very unsuitable, which respectively correspond to 5 different trapezoidal fuzzy numbers. A corresponding relationship table between the 5 trapezoidal fuzzy numbers and the 5 suitability levels is constructed, and the similarity between the trapezoidal fuzzy number output by the RBF neural network classifier and the 5 trapezoidal numbers representing the 5 suitability levels is calculated. The suitability level corresponding to the trapezoidal fuzzy number with the largest similarity is determined as the environmental temperature suitability level of the livestock and poultry house.

3. A livestock house environmental temperature detection system based on a cloud platform according to claim 1 or 2, characterized in that: The livestock and poultry house environmental parameter acquisition and control platform consists of detection nodes, control nodes, gateway nodes, on-site monitoring terminals, cloud platforms, and mobile APPs for livestock and poultry house environmental parameters. Communication between the detection nodes, control nodes, and gateway nodes is achieved by constructing a CAN communication network. The detection nodes send the detected livestock and poultry house environmental parameters to the on-site monitoring terminal through the communication interface of the gateway node. The on-site monitoring terminal processes the sensor data and predicts the temperature. Bidirectional transmission of livestock and poultry house environmental parameters is achieved between the gateway node and the cloud platform through the communication module, and between the cloud platform and the mobile APP through the wireless network. Bidirectional transmission of livestock and poultry house environmental parameters is achieved between the gateway node and the on-site monitoring terminal through the communication interface. The mobile APP provides managers with queries of real-time and historical livestock and poultry house environmental data. Managers can remotely view the current livestock and poultry house environmental parameters through the mobile APP. The cloud platform is mainly responsible for processing, storing, analyzing, and displaying the received livestock and poultry house environmental parameters.

Images