A rotary kiln temperature intelligent control method and Internet of Things system

Abstract

The present invention discloses a rotary kiln temperature control method. The outputs of multiple temperature sensors are used as inputs of a temperature detection module to collect multiple temperature data of the rotary kiln and pre-process the temperature data. A material weight sensor is arranged in a rotary kiln feed transmission device to detect the weight of the material and heat the rotary kiln through a burner. A rotary kiln temperature control system is constructed to control the burner according to the detected temperature data and a temperature target value. The rotary kiln temperature control system also dynamically controls the combustion progress of the burner in real time according to the detected feed weight and the real-time temperature of the rotary kiln at the current moment, combined with the set temperature target value and material weight target value, adjusts the temperature in the rotary kiln in real time to make it meet the temperature target value set by the rotary kiln temperature control system, and enables the system to meet the material weight target value through the feed transmission device, thereby improving the accuracy, dynamics and robustness of the rotary kiln temperature control.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control automation equipment for rotary kiln temperature, and in particular to an intelligent control method for rotary kiln temperature and an Internet of Things system. Background Art

[0002] The existing rotary kiln temperature control method mainly uses simple means such as PID controller and PI controller to achieve temperature control during the heating process of the rotary kiln. Due to the nonlinearity, large hysteresis and uncertain changes of the rotary kiln temperature, the rotary kiln temperature control stability and accuracy are low, and it is difficult to control the rotary kiln temperature at the target value and control it smoothly. In addition, adding materials to the rotary kiln will inevitably affect the temperature conditions in the rotary kiln. Adding materials will inevitably reduce the temperature and combustion temperature in the rotary kiln, thereby affecting the working efficiency and quality of the rotary kiln. How to dynamically and real-time adjust the temperature of the rotary kiln is a technical problem that technicians in this field urgently need to solve. With the development of intelligent control technology, intelligent control of the rotary kiln temperature has become an inevitable trend. Summary of the Invention

[0003] Purpose of the invention: In response to the problems pointed out in the background technology, the present invention discloses a rotary kiln temperature intelligent control method and Internet of Things system, which can more accurately and stably control the rotary kiln temperature to ensure the heating effect and efficiency of the rotary kiln, and provide an Internet of Things system for intelligent control of the rotary kiln temperature.

[0004] Technical solution: The present invention discloses a rotary kiln temperature control method, comprising the following steps:

[0005] Step 1: The outputs of multiple temperature sensors are used as inputs to the temperature detection module, which collects multiple temperature data of the rotary kiln and pre-processes the temperature data;

[0006] Step 2: A feed conveyor is provided at the front end of the rotary kiln. A material weight sensor is arranged in the feed conveyor to detect the weight of the material, and the rotary kiln is heated by a burner;

[0007] Step 3: Construct a rotary kiln temperature control system, which controls the burner according to the temperature data detected by the temperature detection module and the temperature target value set by the rotary kiln temperature control system; the rotary kiln temperature control system also dynamically controls the burner combustion progress in real time according to the feed weight detected by the material weight sensor and the real-time temperature detected by the temperature detection module at the current moment, combined with the temperature target value and material weight target value set by the rotary kiln temperature control system, and adjusts the temperature in the rotary kiln in real time to meet the temperature target value set by the rotary kiln temperature control system, and enables the system to meet the material weight target value through the feed transmission device.

[0008] Furthermore, the rotary kiln temperature control system includes a WOA NARX neural network model 3-5, a WOA ESN neural network model-PID controller 1-3, a WOA fuzzy recursive wavelet neural network model 3-5, a differential unit 1-3, and a fuzzy self-tuning controller;

[0009] The temperature target value set by the rotary kiln temperature control system is obtained, and the cumulative sum of the temperature target value and the output of the fuzzy recursive wavelet neural network model 5 of WOA and the error and the error change rate of the temperature detection module are used as the input of the ESN neural network model-PID controller 3 of WOA; the error and the error change rate of the temperature target value and the output of the NARX neural network model 5 of WOA are used as the input of the fuzzy self-tuning controller; the temperature target value is used as the input of the NARX neural network model 3 and the differential unit 1 of WOA, the output of the differential unit 1 is used as the input of the NARX neural network model 4 of WOA, the output of the fuzzy recursive wavelet neural network model 5 of WOA is used as the input of the NARX neural network model 3 and the differential unit 3 of WOA, the output of the differential unit 3 is used as the input of the NARX neural network model 4 of WOA, and the difference between the output of the NARX neural network model 3 of WOA and the output of the temperature detection module and the change rate of the difference as the input of the ESN neural network model-PID controller 1 of the WOA, the difference between the output of the NARX neural network model 4 of the WOA and the output of the differential unit 2 and the rate of change of the difference as the input of the ESN neural network model-PID controller 2 of the WOA, the outputs of the ESN neural network model-PID controller 1 of the WOA and the ESN neural network model-PID controller 2 of the WOA as the input of the fuzzy recursive wavelet neural network model 3 of the WOA, and the difference between the cumulative sum of the output of the fuzzy recursive wavelet neural network model 3 of the WOA and the output of the ESN neural network model-PID controller 3 of the WOA and the fuzzy recursive wavelet neural network model 4 of the WOA as the input of the burner and the fuzzy recursive wavelet neural network model 4 of the WOA; the output of the temperature detection module as the input of the differential unit 2, the NARX neural network model 5 of the WOA and the fuzzy recursive wavelet neural network model 4 of the WOA;

[0010] The error and error change between the material weight target value set by the rotary kiln temperature control system and the material weight sensor output are used as the input of the fuzzy recursive wavelet neural network model 5 of the WOA, and the material weight sensor output is used as the input of the fuzzy recursive wavelet neural network model 4 of the WOA. Figure 2 .

[0011] Furthermore, the temperature detection module includes WOA's NARX neural network model 1-2, WOA's fuzzy recursive wavelet neural network model 1-2 and WOA's ESN neural network model; the temperature detection module structure is shown in Figure 1shown.

[0012] The outputs of multiple temperature sensors are respectively used as inputs of WOA's NARX neural network model 1, WOA's ESN neural network model and WOA's fuzzy recursive wavelet neural network model 1; the outputs of WOA's NARX neural network model 1 and WOA's ESN neural network model are used as inputs of WOA's fuzzy recursive wavelet neural network model 1; the output of WOA's fuzzy recursive wavelet neural network model 1 is used as inputs of WOA's NARX neural network model 2 and WOA's fuzzy recursive wavelet neural network model 2; the output of WOA's NARX neural network model 2 is used as input of WOA's fuzzy recursive wavelet neural network model 2; and the output of WOA's fuzzy recursive wavelet neural network model 2 is used as the predicted values ​​of the output temperatures of multiple temperature sensors.

[0013] Furthermore, the WOA NARX neural network model 1 and the WOA NARX neural network model 2 have the same structure, are both WOA NARX neural network models, and both use the whale optimization algorithm to optimize the NARX neural network model parameters;

[0014] The structures of WOA's fuzzy recursive wavelet neural network models 1 and 2 are the same, both of which are WOA's fuzzy recursive wavelet neural network models. The whale optimization algorithm is used to optimize the fuzzy recursive wavelet neural network model parameters.

[0015] WOA's ESN neural network model uses the whale optimization algorithm to optimize the ESN neural network model parameters.

[0016] Furthermore, the WOA NARX neural network models 3-5 have the same structure, are all WOA NARX neural network models, and all use the whale optimization algorithm to optimize the NARX neural network model parameters;

[0017] The WOA ESN neural network model-PID controllers 1-3 have the same structure, and are all WOA ESN neural network model-PID controllers, and the WOA ESN neural network model outputs are used as parameters of the PID controller;

[0018] The WOA fuzzy recursive wavelet neural network models 3-5 have the same structure, are all WOA fuzzy recursive wavelet neural network models, and all use the whale optimization algorithm to optimize the fuzzy recursive wavelet neural network model parameters.

[0019] Furthermore, the fuzzy self-tuning controller is composed of two parts in parallel: fuzzy control and integral action. The specific fuzzy control rules are as follows:

[0020] U f =k0×[a×e+(1-a)×e']

[0021] Among them, U f is the output of the fuzzy controller, k0 is the output coefficient, a is the adaptive correction factor, and the size of a reflects the degree of influence of the rotary kiln temperature error e and the rotary kiln temperature error change rate e' on the output of the fuzzy self-tuning controller.

[0022] The present invention also discloses an Internet of Things system for intelligent control of rotary kiln temperature, including a rotary kiln parameter detection node, a rotary kiln temperature control node, a rotary kiln parameter communication node, a rotary kiln cloud platform server and a rotary kiln monitoring mobile phone APP. Two-way communication is achieved between the rotary kiln parameter detection node, the rotary kiln temperature control node, the rotary kiln parameter communication node, the rotary kiln cloud platform server and the rotary kiln monitoring mobile phone APP. The rotary kiln parameter detection node collects rotary kiln parameters, the rotary kiln temperature control node controls the temperature of the rotary kiln, the rotary kiln parameter communication node collects the rotary kiln parameters of the rotary kiln parameter detection node and the rotary kiln temperature control node and uploads them to the rotary kiln cloud platform server, and the rotary kiln cloud platform server stores and manages the parameters of the rotary kiln. The rotary kiln parameter detection node and the rotary kiln temperature control node execute the above-mentioned rotary kiln temperature control method. The structure of the Internet of Things system for intelligent control of rotary kiln temperature is shown in FIG. Figure 3 shown.

[0023] Compared with the prior art, the present invention has the following obvious advantages:

[0024] 1. The reserve pool of the ESN neural network model used in the present invention contains a large number of randomly sparsely connected neurons, which has the function of short-term memory of the rotary kiln temperature parameters. The reserve pool of the ESN neural network model can be regarded as a nonlinear time kernel, mapping the temperature dynamic time input to a high-dimensional static space, fully mining the timing information of the dynamic changes of the rotary kiln temperature, and improving the dynamic performance, accuracy and robustness of processing the rotary kiln temperature.

[0025] 2. The NARX neural network model 3 of WOA, the fuzzy recursive wavelet neural network model 5 of WOA, the fuzzy self-adjusting controller, the material weight target value and the material weight sensor used in the present invention constitute a compensation control loop for the dynamic change of the rotary kiln temperature. The output of the fuzzy recursive wavelet neural network model 5 of WOA is used to perform real-time dynamic compensation control on the dynamic change of the rotary kiln temperature, thereby effectively reducing the influence and error of the sudden addition of materials and the temperature change of the rotary kiln on the temperature tracking performance of the rotary kiln, and reducing the influence of the uncertain disturbance of the material weight and the temperature change of the rotary kiln on the dynamic tracking control process of the rotary kiln temperature, thereby improving the accuracy, dynamics and robustness of the rotary kiln temperature control.

[0026] 3. The NARX neural network model of WOA in the present invention uses the output data of the previous rotary kiln temperature detection and control process in the next calculation by adding a delay link, so that the NARX neural network model used to detect and control the rotary kiln temperature process not only has dynamic characteristics but also the information of the rotary kiln temperature detection and control process is more complete, thereby improving the dynamics, robustness and accuracy of predicting and controlling the rotary kiln temperature process, and conforming to the dynamic mechanism of the change of rotary kiln temperature parameters.

[0027] 4. The rotary kiln temperature detection and control process is a complex system with nonlinearity and large hysteresis. To address the problem that the traditional PID control method has poor control effect in it, the present invention designs a WOA ESN neural network model-PID controller, which utilizes the self-learning dynamic adaptive characteristics of the WOA ESN neural network model to improve the dynamics, accuracy and robustness of the rotary kiln temperature control process.

[0028] 5. The temperature target value, temperature detection module and WOA's ESN neural network model-PID controller 3 constitute the real-time control loop of the rotary kiln temperature. WOA's NARX neural network model 3, WOA's ESN neural network model-PID controller 3 and WOA's fuzzy recursive wavelet neural network model 3 constitute the steady-state control loop of the rotary kiln temperature control, realizing the interaction, mutual influence and mutual complementation of the real-time control and steady-state control of the rotary kiln temperature to improve the accuracy, dynamics and robustness of the rotary kiln temperature control.

[0029] 6. The differential units 1-3 of the present invention, the NARX neural network model 4 of WOA, and the ESN neural network model-PID controller 2 of WOA constitute the temperature target value change rate of the rotary kiln, the temperature detection module output change rate, and the fuzzy recursive wavelet neural network model 5 of WOA output compensation control value change rate. The three jointly control to achieve dynamic real-time tracking of the temperature target value change rate of the rotary kiln, thereby improving the dynamic performance, accuracy and robustness of the rotary kiln temperature control. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The temperature detection module of the present invention;

[0031] Figure 2 The rotary kiln temperature control system of the present invention;

[0032] Figure 3 The Internet of Things system for intelligent temperature control of a rotary kiln of the present invention;

[0033] Figure 4 It is the rotary kiln parameter detection node of the present invention;

[0034] Figure 5 It is the temperature control node of the rotary kiln of the present invention;

[0035] Figure 6 It is the rotary kiln parameter communication node of the present invention;

[0036] Figure 7 This is the software function of the rotary kiln parameter processing, analysis and monitoring terminal of the present invention. DETAILED DESCRIPTION

[0037] Combined with attachment Figure 1-7 , the technical solution of the present invention is further described:

[0038] The present invention discloses a rotary kiln temperature intelligent control method and an Internet of Things system, which specifically includes the following steps:

[0039] 1. Design of temperature detection module

[0040] The temperature detection module includes WOA's NARX neural network model 1-2, WOA's fuzzy recursive wavelet neural network model 1-2 and WOA's ESN neural network model. The temperature detection module structure diagram is shown in Figure 1 .

[0041] 1. Design of WOA’s ESN Neural Network Model

[0042] The reserve pool of the ESN neural network model contains a large number of randomly sparsely connected neurons, which have the function of short-term memory of the rotary kiln temperature parameters. The reserve pool of the ESN neural network model can be regarded as a nonlinear time kernel, mapping the temperature dynamic time input to a high-dimensional static space, fully mining the time series information of the dynamic changes in the rotary kiln temperature, and improving the dynamic performance and accuracy of processing the rotary kiln temperature. The number of neurons in the reserve pool is called the reserve pool scale, which is one of the important parameters of the ESN neural network model. The ESN neural network model consists of three parts: input layer, reserve pool and output layer, which contain K, R and L neurons respectively. Win is the connection weight matrix between the input layer and the reserve pool. The connection matrix of the neurons in the reserve pool is randomly generated during the network initialization phase. Wout is the readout connection weight matrix of the forward connection reserve pool and the output layer. Let the input temperature sequence value at time t be X(t), the output temperature process value of the reserve pool neuron be S(t), and the temperature prediction value output sequence be y(t). Then the update model of the ESN neural network model during training is:

[0043] S(t)=f(W in X(t)+WS(t+1)) (1)

[0044] y(t)=f out (W out S(t)+α) (2)

[0045] Among them, f(·) is the nonlinear reservoir activation function, f out (·) is the activation function of the output layer, α is the bias vector at the output, and W is the connection weight.

[0046] The WOA ESN neural network model uses the whale optimization algorithm to optimize the ESN neural network model parameters. The WOA algorithm completes the optimization of the target value by simulating the three predation behaviors of humpback whales: surrounding prey, attacking prey with a spiral bubble net, and randomly searching for prey. The steps for constructing the WOA ESN neural network model are as follows: (1) Initialize the number of whale populations and the location of whale groups, set the maximum number of iterations, and map the weights and hidden layer thresholds of the ESN neural network model to the position vectors of individual whale groups.

[0047] (2) Calculate the fitness value corresponding to the individual whale position and obtain the optimal individual whale position.

[0048] (3) By comparing the whale position update probability p and the coefficient vector A to see whether they meet the cutoff condition, the whale position is updated. When the maximum number of iterations is met, the position vector of the optimal whale individual is output, the optimal weights and thresholds are obtained, and the ESN neural network model is input for training and recognition.

[0049] (4) Construct the ESN neural network model of WOA.

[0050] 2. Design of Fuzzy Recursive Wavelet Neural Network Model 1-2 of WOA

[0051] The fuzzy recursive wavelet neural network model has five layers, with three hidden layers: the membership function layer, the rule layer, and the recursive wavelet function layer. The fuzzy recursive wavelet neural network model combines fuzzy logic, wavelet processing, and recursive structures to improve its ability to process rotary kiln temperature and its accuracy, while also addressing the shortcomings of static mapping. The first layer is the input layer, where both neuron nodes are input nodes, equivalent to input variables. The second layer is the membership function layer, where the output of each neuron corresponds to three neurons in the membership function layer. The nonlinear transformation in the membership function layer uses a Gaussian function, which incorporates fuzzy logic inference to improve the inductive performance of the fuzzy recursive wavelet neural network model. The third layer is the rule layer, where each neuron is the antecedent of a fuzzy logic rule. The neurons in this layer perform product operations on the input signals of this layer. The fourth layer is the recursive wavelet layer, which includes the consequent of wavelet function operations, recursive operations, and fuzzy logic rules. To increase sensitivity to the wavelet function and consequent input of the rotary kiln temperature signal and achieve real-time self-adjustment and control of the rotary kiln temperature, recursive processing is incorporated to meet the requirements of dynamic kiln temperature processing. The fifth layer is the output layer, which defuzzifies all input rotary kiln temperature signals. The output layer value is the product of the fourth layer's output and the connection weights of this layer. A gradient descent algorithm is used to adjust the connection weights, membership function, and wavelet function parameters in the fuzzy recursive wavelet neural network model.

[0052] The fuzzy recursive wavelet neural network model 1-2 of WOA refers to the design method of the ESN neural network model of WOA in this patent.

[0053] 3. Design of WOA’s NARX neural network model 1-2

[0054] The NARX neural network model introduces the input delay and output delay feedback values ​​of the NARX neural network model over the past period of time as the input of the NARX neural network model, simulating the autoregressive relationship between the input and output of the rotary kiln temperature detection and control process. Its training method is divided into two stages: forward propagation and backpropagation.

[0055] (1) Forward propagation stage: The historical input and output data of the rotary kiln temperature detection and control process are passed to the NARX neural network model, and the current NARX neural network model prediction output is generated.

[0056] (2) Back propagation stage. The weights and biases of the NARX neural network model are updated according to the error of predicting and controlling the temperature process of the rotary kiln to improve the accuracy of the NARX neural network model. The NARX neural network model consists of an input layer, a hidden layer, and an output layer. The NARX neural network model uses the external input u(t) at the current time t, the m-order delay of the external input, and the n-order delay of the output predicted by the neural network model as the input of the NARX neural network model, where the relationship between input and output is:

[0057] y(t)=f(y(t-1),y(t-2),…,y(t-ny),u(t),u(t-1),…,u(t-mu)) (3)

[0058] Where: u is input; y is output; m and n are the number of input and output variables.

[0059] The NARX neural network model of WOA refers to the design method of the ESN neural network model of WOA in this patent.

[0060] 2. Design of rotary kiln temperature control system

[0061] The rotary kiln temperature control system includes WOA's NARX neural network model 3-5, WOA's ESN neural network model-PID controller 1-3, WOA's fuzzy recursive wavelet neural network model 3-5, differential unit 1-3, temperature detection module and fuzzy self-adjusting controller. The structure of the rotary kiln temperature control system is shown in Figure 2 shown.

[0062] 1. Design of ESN Neural Network Model of WOA - PID Controller 1-3

[0063] WOA's ESN neural network model - PID controller is the Kp, Ki and Kd parameters of the PID controller output by WOA's ESN neural network model. The input of WOA's ESN neural network model - PID controller is the error and error change rate between the given value and the feedback detection value of the controlled quantity, which realizes online adjustment of PID controller parameters and further optimizes the dynamics, accuracy and robustness of the controlled parameters. The output value of WOA's ESN neural network model - PID controller is:

[0064]

[0065] Where: u(k) is the output value of the PID controller, e(k) is the error of the controlled object, and k is the number of sampling times.

[0066] 2. Design of fuzzy self-tuning controller

[0067] The fuzzy self-tuning controller consists of two parts in parallel: fuzzy control and integral action. The fuzzy control rule is:

[0068] U f =k0×[a×e+(1-a)×e'] (5)

[0069] U f is the output of the fuzzy controller, k0 is the output coefficient, a is the adaptive correction factor, and the size of a reflects the degree of influence of the rotary kiln temperature error e and the rotary kiln temperature error change rate e″ on the output of the fuzzy self-adjusting controller.

[0070] In the initial stage, the temperature error of the rotary kiln is relatively large. At this time, a larger a value should be selected to eliminate the existence of the rotary kiln temperature error as soon as possible, and the weight of the rotary kiln temperature error in the fuzzy self-adjusting controller should be increased; in the mid-term stage, the rotary kiln temperature error decreases and the rotary kiln temperature rise rate accelerates. In order to reduce the overshoot of the rotary kiln temperature, the control effect of the rotary kiln temperature error change should be highlighted, and a smaller a value should be selected. When the rotary kiln temperature is close to the expected temperature value, since the rotary kiln temperature error and its change are both small at this time, the rotary kiln temperature error and the error change rate can take the same weight.

[0071] The design methods of other models refer to the design method of the temperature detection module of this patent.

[0072] 3. Design of IoT system for intelligent temperature control of rotary kiln

[0073] The Internet of Things system for intelligent control of rotary kiln temperature includes a rotary kiln parameter detection node, a rotary kiln temperature control node, a rotary kiln parameter communication node, a rotary kiln cloud platform server, and a rotary kiln monitoring mobile phone APP. Two-way communication is achieved between the rotary kiln parameter detection node, the rotary kiln temperature control node, the rotary kiln parameter communication node, the rotary kiln cloud platform server, and the rotary kiln monitoring mobile phone APP. The rotary kiln parameter detection node collects rotary kiln parameters, the rotary kiln temperature control node controls the temperature of the rotary kiln, the rotary kiln parameter communication node collects the rotary kiln parameters of the rotary kiln parameter detection node and the rotary kiln temperature control node and uploads them to the rotary kiln cloud platform server. The rotary kiln cloud platform server stores and manages the parameters of the rotary kiln. The rotary kiln parameter processing and analysis monitoring terminal processes and analyzes the collected rotary kiln big data and monitors the rotary kiln parameters. Structure of the Internet of Things system for intelligent control of rotary kiln temperature Figure 3 .

[0074] 1. Design of rotary kiln parameter detection nodes

[0075] The present invention adopts a rotary kiln parameter detection node as a rotary kiln parameter sensing terminal to be detected. The rotary kiln parameter detection node and the rotary kiln parameter communication node realize two-way communication through the LoRa communication module. The rotary kiln parameter detection node includes temperature, crack, deformation, gas sensors and corresponding signal conditioning circuits, cameras, STM32 microprocessors, LoRa communication modules, and gas sensors include oxygen, carbon dioxide, sulfur dioxide and nitrogen oxides. The software of the rotary kiln parameter detection node mainly realizes the communication of the LoRa communication module, the collection and preprocessing of the rotary kiln parameters. The software is designed with C language programming and has a high degree of compatibility, which greatly improves the work efficiency of software design and development and enhances the reliability, readability and portability of the program code. See the rotary kiln parameter detection node. Figure 4 shown.

[0076] 2. Design of rotary kiln temperature control nodes

[0077] The rotary kiln temperature control node includes an STM32 microprocessor, a camera, a LoRa communication module, a rotary kiln parameter processing and analysis monitoring terminal and a control device, which are used for real-time monitoring of the rotary kiln temperature. The camera obtains real-time images of the rotary kiln for video monitoring. The control equipment includes a feed transmission device, a discharge transmission device, a burner and a material weight sensor to realize intelligent control to ensure that the temperature parameters of the rotary kiln are within the appropriate heating range. The rotary kiln temperature control node exchanges information with the rotary kiln parameter communication node through the LoRa communication module. The rotary kiln parameter processing and analysis monitoring terminal is an industrial control computer, which is connected to the STM32 microprocessor through a USB interface. The rotary kiln parameter processing and analysis monitoring terminal mainly realizes the rotary kiln communication parameter setting, rotary kiln data analysis, rotary kiln data management and rotary kiln temperature control system. The rotary kiln temperature control node structure diagram is shown in Figure 5 , the software functions of rotary kiln parameter processing and analysis monitoring terminal are shown in Figure 6 .

[0078] 3. Design of rotary kiln parameter communication node

[0079] The rotary kiln parameter communication node includes an STM32 microprocessor, a LoRa communication module, a camera, and a wireless communication module. The LoRa communication module is used to build a self-organizing communication network to realize data interaction between the rotary kiln parameter detection node, the rotary kiln temperature control node, and the rotary kiln parameter communication node. The rotary kiln parameter communication node realizes two-way data interaction with the rotary kiln cloud platform server through the wireless communication module. The detection parameters of the rotary kiln parameter detection node are uploaded to the rotary kiln cloud platform server through the rotary kiln parameter communication node to realize the storage and processing of the rotary kiln parameters. Figure 7 .

[0080] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made in accordance with the spirit of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. A rotary kiln temperature control method, characterized in that: The steps include: Step 1: The outputs of multiple temperature sensors are used as inputs to the temperature detection module, which collects multiple temperature data of the rotary kiln and pre-processes the temperature data; Step 2: A feed conveyor is provided at the front end of the rotary kiln. A material weight sensor is arranged in the feed conveyor to detect the weight of the material, and the rotary kiln is heated by a burner; Step 3: Constructing a rotary kiln temperature control system, the rotary kiln temperature control system controls the burner according to the temperature data detected by the temperature detection module and the temperature target value set by the rotary kiln temperature control system; the rotary kiln temperature control system also dynamically controls the burner combustion progress in real time according to the feed weight detected by the material weight sensor and the real-time temperature detected by the temperature detection module at the current moment, combined with the temperature target value and the material weight target value set by the rotary kiln temperature control system, and adjusts the temperature in the rotary kiln in real time to meet the temperature target value set by the rotary kiln temperature control system, and enables the system to meet the material weight target value through the feed transmission device; The rotary kiln temperature control system includes a NARX neural network model 3 of WOA, a NARX neural network model 4 of WOA, a NARX neural network model 5 of WOA, an ESN neural network model-PID controller 1 of WOA, an ESN neural network model-PID controller 2 of WOA, an ESN neural network model-PID controller 3 of WOA, a fuzzy recursive wavelet neural network model 3 of WOA, a fuzzy recursive wavelet neural network model 4 of WOA, a fuzzy recursive wavelet neural network model 5 of WOA, a differential unit 1, a differential unit 2, a differential unit 3, and a fuzzy self-tuning controller; The temperature target value set by the rotary kiln temperature control system is obtained, and the cumulative sum of the temperature target value and the output of the fuzzy recursive wavelet neural network model 5 of WOA and the error and the error change rate of the temperature detection module are used as the input of the ESN neural network model-PID controller 3 of WOA; the error and the error change rate of the temperature target value and the output of the NARX neural network model 5 of WOA are used as the input of the fuzzy self-tuning controller; the temperature target value is used as the input of the NARX neural network model 3 and the differential unit 1 of WOA, the output of the differential unit 1 is used as the input of the NARX neural network model 4 of WOA, the output of the fuzzy recursive wavelet neural network model 5 of WOA is used as the input of the NARX neural network model 3 and the differential unit 3 of WOA, the output of the differential unit 3 is used as the input of the NARX neural network model 4 of WOA, and the difference between the output of the NARX neural network model 3 of WOA and the output of the temperature detection module and the change rate of the difference as the input of the ESN neural network model-PID controller 1 of the WOA, the difference between the output of the NARX neural network model 4 of the WOA and the output of the differential unit 2 and the rate of change of the difference as the input of the ESN neural network model-PID controller 2 of the WOA, the outputs of the ESN neural network model-PID controller 1 of the WOA and the ESN neural network model-PID controller 2 of the WOA as the input of the fuzzy recursive wavelet neural network model 3 of the WOA, and the difference between the cumulative sum of the output of the fuzzy recursive wavelet neural network model 3 of the WOA and the output of the ESN neural network model-PID controller 3 of the WOA and the fuzzy recursive wavelet neural network model 4 of the WOA as the input of the burner and the fuzzy recursive wavelet neural network model 4 of the WOA; the output of the temperature detection module as the input of the differential unit 2, the NARX neural network model 5 of the WOA and the fuzzy recursive wavelet neural network model 4 of the WOA; The error and error change between the material weight target value set by the rotary kiln temperature control system and the material weight sensor output are used as the input of the fuzzy recursive wavelet neural network model 5 of WOA, and the material weight sensor output is used as the input of the fuzzy recursive wavelet neural network model 4 of WOA.

2. A rotary kiln temperature control method according to claim 1, characterized in that: The temperature detection module includes WOA's NARX neural network model 1, WOA's NARX neural network model 2, WOA's fuzzy recursive wavelet neural network model 1, WOA's fuzzy recursive wavelet neural network model 2 and WOA's ESN neural network model; The outputs of multiple temperature sensors are respectively used as inputs of WOA's NARX neural network model 1, WOA's ESN neural network model and WOA's fuzzy recursive wavelet neural network model 1; the outputs of WOA's NARX neural network model 1 and WOA's ESN neural network model are used as inputs of WOA's fuzzy recursive wavelet neural network model 1; the output of WOA's fuzzy recursive wavelet neural network model 1 is used as inputs of WOA's NARX neural network model 2 and WOA's fuzzy recursive wavelet neural network model 2; the output of WOA's NARX neural network model 2 is used as input of WOA's fuzzy recursive wavelet neural network model 2; and the output of WOA's fuzzy recursive wavelet neural network model 2 is used as the predicted values ​​of the output temperatures of multiple temperature sensors.

3. A rotary kiln temperature control method according to claim 2, characterized in that, The WOA NARX neural network model 1 and the WOA NARX neural network model 2 have the same structure, are both WOA NARX neural network models, and both use the whale optimization algorithm to optimize the NARX neural network model parameters; The structures of WOA's fuzzy recursive wavelet neural network model 1 and WOA's fuzzy recursive wavelet neural network model 2 are the same, both of which are WOA's fuzzy recursive wavelet neural network models. The whale optimization algorithm is used to optimize the fuzzy recursive wavelet neural network model parameters. WOA's ESN neural network model uses the whale optimization algorithm to optimize the ESN neural network model parameters.

4. A rotary kiln temperature control method according to claim 1, characterized in that: The WOA NARX neural network model 3, the WOA NARX neural network model 4, and the WOA NARX neural network model 5 have the same structure, are all WOA NARX neural network models, and all use the whale optimization algorithm to optimize the NARX neural network model parameters; The WOA ESN neural network model-PID controller 1, the WOA ESN neural network model-PID controller 2, and the WOA ESN neural network model-PID controller 3 have the same structure, and are all WOA ESN neural network model-PID controllers, and the WOA ESN neural network model output is used as the parameter of the PID controller; The WOA fuzzy recursive wavelet neural network model 3, the WOA fuzzy recursive wavelet neural network model 4, and the WOA fuzzy recursive wavelet neural network model 5 have the same structure, are all WOA fuzzy recursive wavelet neural network models, and all use the whale optimization algorithm to optimize the fuzzy recursive wavelet neural network model parameters.

5. A rotary kiln temperature control method according to claim 1, characterized in that: The fuzzy self-tuning controller consists of two parts in parallel: fuzzy control and integral action. The specific fuzzy control rules are as follows: ； in, is the fuzzy controller output, is the output coefficient, a is the adaptive correction factor, and the size of a reflects the rotary kiln temperature error e and the rotary kiln temperature error change rate The degree of influence on the output of the fuzzy self-tuning controller.

6. An Internet of Things system for intelligent temperature control of a rotary kiln, characterized in that: The invention comprises a rotary kiln parameter detection node, a rotary kiln temperature control node, a rotary kiln parameter communication node, a rotary kiln cloud platform server and a rotary kiln monitoring mobile phone APP. Two-way communication is realized among the rotary kiln parameter detection node, the rotary kiln temperature control node, the rotary kiln parameter communication node, the rotary kiln cloud platform server and the rotary kiln monitoring mobile phone APP. The rotary kiln parameter detection node collects rotary kiln parameters, the rotary kiln temperature control node controls the temperature of the rotary kiln, the rotary kiln parameter communication node collects the rotary kiln parameters of the rotary kiln parameter detection node and the rotary kiln temperature control node and uploads them to the rotary kiln cloud platform server, and the rotary kiln cloud platform server stores and manages the parameters of the rotary kiln. The rotary kiln parameter detection node and the rotary kiln temperature control node execute the rotary kiln temperature control method according to any one of claims 1 to 5.

Images