Heat exchanger automatic cooling circulation control method and system based on Internet of Things

Abstract

The invention relates to the technical field of temperature control, and discloses a heat exchanger automatic cooling circulation control method and system based on the Internet of Things. Firstly, a heat exchanger related parameter data set is obtained, and a heat exchanger automatic cooling circulation control model is established; secondly, establishing a BP neural network, and searching an initial weight of the BP neural network by using an improved red-mouth blue-magpie optimization algorithm to obtain a globally optimal solution; using the globally optimal solution as an initial weight of a BP neural network, training PID control parameters by using a PID control algorithm to obtain optimized PID control parameters, and completing cooling circulation cooling control in combination with an automatic cooling circulation control model of the heat exchanger; and finally, training to obtain a GRU neural network early warning model, outputting a temperature early warning value of the heat exchanger, and realizing temperature control early warning of the heat exchanger. By designing the heat exchanger automatic cooling circulation control system, the purpose of heat exchanger automatic cooling circulation control is achieved, and the method is accurate and objective.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of temperature control, in particular to a heat exchanger automatic cooling circulation control method and system based on Internet of Things. BACKGROUND

[0002] The traditional automatic cooling circulation control method usually cools through a cooling tower, which not only has large energy consumption but also cannot meet the process requirements in terms of cooling effect. Meanwhile, the traditional automatic cooling circulation control method does not use machine learning technology, and the temperature control switch is not intelligently controlled, resulting in low control precision and inability to meet the requirements of high-precision temperature control, which may affect the normal and stable operation of the equipment and cause safety hazards. SUMMARY

[0003] In view of the problems in the related art, the present application provides a heat exchanger automatic cooling circulation control method and system based on Internet of Things to overcome the above technical problems existing in the prior art.

[0004] To solve the above technical problems, the present application is realized by the following technical scheme:

[0005] The present application is a heat exchanger automatic cooling circulation control method based on Internet of Things, comprising the following steps:

[0006] S1, obtain heat exchanger related parameters to form a heat exchanger related parameter data set, and establish a heat exchanger automatic cooling circulation control model according to the heat exchanger related parameter data set;

[0007] S2, establish a BP neural network, introduce chaos mapping and dynamic weight to improve the Magpie optimization algorithm, obtain an improved Magpie optimization algorithm, use the improved Magpie optimization algorithm to find the initial weight of the BP neural network, and obtain a global optimal solution;

[0008] S3, use the global optimal solution as the initial weight of the BP neural network, combine it with the PID control algorithm, train the PID control parameters, obtain optimized PID control parameters, use the optimized PID control parameters and the heat exchanger automatic cooling circulation control model to control the cooling circulation of the heat exchanger, and complete the cooling circulation cooling control;

[0009] S4, train a GRU neural network to obtain a GRU neural network early warning model, output a heat exchanger temperature early warning value according to the heat exchanger related parameter data set and in combination with the GRU neural network early warning model, and realize heat exchanger temperature control early warning.

[0010] The application obtains heat exchanger related parameter data, and establishes a heat exchanger automatic cooling cycle control model, which clearly describes the parameter data relationship of the heat exchanger through the establishment of a heat balance formula, facilitates the leaving of a lag time when the heat exchanger is controlled to cool and circulate, and reduces temperature control errors; secondly, an improved magpie optimization algorithm is used to find the initial weight value of the BP neural network to obtain a global optimal solution, and the training time is effectively reduced by directly using the global optimal solution, thereby reducing the cooling time and cost; the algorithm simulates the predatory behavior of magpies to continuously iterate the position to determine the global optimal solution, introduces chaos mapping to increase the randomness of the population and reduce the risk of falling into a local optimal solution, and uses dynamic weights to accelerate the optimization speed and greatly reduce the optimization time compared with traditional algorithms; then, the PID control algorithm and the BP neural network are used to train the PID control parameters to obtain optimized PID control parameters, and an incremental PID controller outputs a control signal to control the heat exchanger to cool and circulate in combination with the heat exchanger automatic cooling cycle control model; the use of the neural network can effectively reduce the temperature control error and greatly improve the control ability, and compared with the traditional temperature control method, the temperature control method reduces the overshoot of the cooling speed, improves the energy efficiency ratio of the heat exchanger under the premise of accurate temperature control; finally, a GRU neural network early warning model is trained to output a heat exchanger temperature early warning value to realize heat exchanger temperature control early warning; the neural network is suitable for learning the correlation between temperature data and time, has high prediction accuracy, can ensure that the abnormal temperature of the heat exchanger can be found in advance, realizes early warning, and guarantees stable operation and equipment safety.

[0011] Preferably, the S1 comprises the following steps:

[0012] S11, obtain heat exchanger related parameters, including heat exchanger outlet temperature, cooling water mass, cooling water flow rate, etc., collect data of the heat exchanger related parameters, and give a time sequence to obtain heat exchanger related parameter data to form a heat exchanger related parameter data set; according to the heat exchanger related parameter data set, set the cooling water flow rate, the specific heat capacity of the cooling water, the heat exchanger inlet temperature, the heat exchanger outlet temperature, the cooling water mass, establish a heat exchanger heat balance formula, and take the heat exchanger outlet temperature as a time function to record the heat exchanger outlet temperature at time t to obtain the heat exchanger heat balance formula;

[0013] S12, Laplace transform is performed on the heat exchanger heat balance formula to obtain a heat exchanger outlet temperature expression and a heat exchanger inlet temperature expression; the impedance of the cooling water transferred in the heat exchanger at time t, the cooling water flow rate at time t and a constant are set to obtain an impedance calculation formula;

[0014] The heat exchanger temperature first-order lag function is obtained by converting the heat exchanger outlet temperature expression, the heat exchanger inlet temperature expression and the impedance calculation formula, and the heat exchanger temperature first-order lag function is the heat exchanger automatic temperature reduction cycle control model.

[0015] The application obtains the heat exchanger related parameter data, and establishes the heat exchanger automatic temperature reduction cycle control model.

[0016] Preferably, the S2 comprises the following steps:

[0017] S21, setting the BP neural network to be composed of an input layer, a hidden layer and an output layer, using an incremental PID controller to adjust the temperature control system of the heat exchanger, the incremental PID controller containing a controller input and a controller output, recording the control deviation of the controller input and the controller output as a control deviation, recording the control deviation at time d, the control deviation at time d-1 and the control deviation at time d-2, setting the proportional coefficient, the integral time coefficient and the differential time coefficient of the incremental PID controller, and obtaining the control parameter calculation formula of the incremental PID controller at time d;

[0018] The control parameter calculation formula of the incremental PID controller at time d is taken as the input of the input layer of the BP neural network, the proportional coefficient, the integral time coefficient and the differential time coefficient of the incremental PID controller are taken as the output of the output layer of the BP neural network, and the BP neural network is established;

[0019] S22, obtaining the control deviation of the incremental PID controller, obtaining a control deviation set, dividing the control deviation set into a deviation training set and a deviation test set, inputting the deviation training set into the BP neural network, taking the output error of the BP neural network as a fitness function, introducing chaos mapping and dynamic weight to improve the Magpie optimization algorithm to obtain an improved Magpie optimization algorithm, using the improved Magpie optimization algorithm to optimize the initial weight of the BP neural network, and obtaining a global optimal solution, and the specific steps are as follows:

[0020] The S22 comprises the following steps:

[0021] S221, regarding the BP neural network as a search space, there is a Magpie population in the search space, the Magpie individuals in the Magpie population represent the weights of the BP neural network, the positions of the Magpie individuals in the Magpie population are initialized using chaos mapping, and the Magpie population initialization is completed; after the Magpie population initialization, the Magpie population performs a foraging stage, and the current iteration number is set.

[0022] Set food position, maximum iteration number, divide the blue magpie population into several blue magpie groups, set the number of blue magpie groups, and update the position of the blue magpie individual according to the blue magpie group at this time;

[0023] The fitness function value corresponding to the position of the blue magpie individual in the current iteration is calculated, the position of the blue magpie individual with the best fitness function value is taken as the current best weight of the BP neural network, and the current optimal solution is obtained;

[0024] S222, introduce dynamic weight, give the blue magpie group weight, record it as weight set, the blue magpie individual moves along with the blue magpie group, and the position of the blue magpie individual is updated using dynamic weight;

[0025] The best fitness function value is screened to obtain the first stage blue magpie population; repeat iteration until the current iteration number reaches the maximum iteration number, stop iteration, and obtain the final blue magpie population; the position of the blue magpie individual corresponding to the best fitness function value in the final blue magpie population is recorded as the global optimal solution, and the global optimal solution is the initial weight of the BP neural network.

[0026] The improved blue magpie optimization algorithm is used to find the initial weight of the BP neural network, the position is iteratively moved by simulating the predation behavior of the blue magpie, the global optimal solution is determined, the training time is effectively reduced by directly using the global optimal solution, the cooling time and cost are reduced; and the chaos mapping is introduced, the randomness of the population is increased, the risk of falling into local optimal solution is reduced, the dynamic weight is used, compared with the traditional algorithm, the optimization speed is accelerated, and the optimization time is greatly reduced.

[0027] Preferably, the S3 comprises the following steps:

[0028] S31, the global optimal solution is taken as the initial weight of the BP neural network, the BP neural network is trained until convergence, and the trained BP neural network is obtained; the PID control algorithm is used, and the trained BP neural network is combined to train the PID control parameters, and the optimized PID control parameters are obtained, and the specific steps are as follows:

[0029] S311, the deviation test set is input into the trained BP neural network, the first iteration output result is collected at the first iteration, an error threshold is set, the size of the first iteration output result and the error threshold is recorded, the first iteration output result is the proportional coefficient, integral time coefficient and differential time coefficient of the incremental PID controller, and the control parameters of the first iteration incremental PID controller are calculated according to the control parameter calculation formula of the incremental PID controller at time d;

[0030] S312, in the second iteration, the control parameters of the second iteration incremental PID controller are calculated by adjusting the weights of the output layer and the hidden layer in the trained BP neural network until the output result is less than the error threshold, the iteration is stopped, the BP neural network model is obtained, and the final output result is obtained as the optimized PID control parameters;

[0031] S32, the optimized PID control parameters correspond to the optimized proportional coefficient, the optimized integral time coefficient and the optimized differential time coefficient in turn, the optimized PID control parameters are taken as the input of the incremental PID controller, the control signal is output by the incremental PID controller, the heat exchanger is controlled in the cooling cycle by combining the heat exchanger automatic cooling cycle control model, and the specific steps are as follows:

[0032] S321, the rated temperature upper limit of the heat exchanger is set, when the current temperature in the heat exchanger is greater than or equal to the rated temperature upper limit, the cooling signal of the temperature sensor is received by the incremental PID controller, the control signal of the incremental PID controller is delayed according to the heat exchanger automatic cooling cycle control model, the cooling tower controls the cooling water injection to cool the heat exchanger, and the adjusted heat exchanger temperature is obtained by the incremental PID controller.

[0033] S322, the cooling water quality threshold and the cooling water flow rate threshold of the heat exchanger are set, when the current cooling water quality in the heat exchanger is greater than or equal to the cooling water quality threshold, the cooling water flow rate is reduced, and the heat exchanger drainage is started; when the adjusted heat exchanger temperature is greater than or equal to the rated temperature upper limit, the cooling water flow rate is increased, when the current temperature in the heat exchanger is less than the rated temperature upper limit, the cooling treatment is stopped, otherwise the heat exchanger temperature abnormality alarm is started, and the heat exchanger cooling cycle control is completed.

[0034] The application trains the PID control parameters by combining the BP neural network, obtains the optimized PID control parameters, effectively reduces the temperature control error by using the neural network, greatly improves the control ability, outputs the control signal by the incremental PID controller to control the heat exchanger in the cooling cycle, and compared with the traditional temperature control method, the overshoot of the cooling speed is reduced, the energy efficiency ratio of the heat exchanger is improved under the premise of accurate temperature control.

[0035] Preferably, the S4 comprises the following steps:

[0036] S41, obtain the heat exchanger related parameters of previous years, wherein the heat exchanger related parameters of previous years include heat exchanger temperature of previous years and heat exchanger outlet temperature of previous years, and time series are given, the time step is gamma, and a heat exchanger related parameter data set of previous years is obtained;The time series corresponding to the heat exchanger related parameter data set of previous years;A GRU neural network early warning model is obtained by training using the heat exchanger related parameter data set of previous years, and the specific steps are as follows:

[0037] S411, the heat exchanger related parameter data set of previous years is normalized to obtain a normalized parameter data set, a time window is set, the normalized parameter data set is divided using the time window to obtain a normalized parameter data segment set, and the normalized parameter data segment set is divided into a sample training set and a sample test set;The GRU neural network is set to obtain a predicted value using forward propagation, the weight is optimized using the Adam algorithm, the sample training set is input into the GRU neural network in turn until convergence, and a trained GRU neural network is obtained;

[0038] S412, the sample test set is input into the trained GRU neural network in turn, and the precision threshold is set;When the output prediction result is less than the precision threshold, the iteration is stopped, and a GRU neural network early warning model is obtained;

[0039] S42, the heat exchanger related parameter data of heat exchanger temperature and heat exchanger outlet temperature in the heat exchanger related parameter data set is obtained to obtain a heat exchanger early warning data set, the heat exchanger early warning data set is normalized and input into the GRU neural network early warning model, a heat exchanger temperature early warning value is output, and the early warning threshold is set;When the heat exchanger temperature early warning value is greater than the early warning threshold, the heat exchanger temperature is greater than or equal to the upper limit of the rated temperature after the time step gamma, and the heat exchanger is cooled again to realize heat exchanger temperature control early warning.

[0040] The application realizes heat exchanger temperature control early warning by training a GRU neural network early warning model to output a heat exchanger temperature early warning value;The neural network is suitable for learning the correlation between temperature data and time, has high prediction accuracy, can detect abnormal heat exchanger temperature in advance, realize early warning, and ensure stable operation and equipment safety.

[0041] The application also discloses a heat exchanger automatic cooling circulation control method based on the Internet of Things, which specifically comprises: a cooling model establishing module, a neural network parameter optimization module, a cooling circulation cooling control module and a heat exchanger temperature control early warning module.

[0042] The cooling model establishing module is used to establish a heat exchanger automatic cooling circulation control model using heat exchanger related parameter data;

[0043] The neural network parameter optimization module is used for finding the initial weight value of the BP neural network using the improved red-billed blue magpie optimization algorithm.

[0044] The cooling cycle temperature control module is used for controlling the heat exchanger in a cooling cycle using the optimized PID control parameter and the cooling model.

[0045] The heat exchanger temperature control early warning module is used for establishing a neural network early warning model and outputting a heat exchanger temperature early warning value.

[0046] The present application has the following advantages:

[0047] 1. The present application establishes an automatic cooling cycle control model of the heat exchanger, which clearly describes the parameter data relationship of the heat exchanger by establishing a heat balance equation, and facilitates the cooling cycle control of the heat exchanger to leave a lag time and reduce temperature control error.

[0048] 2. The present application uses the improved red-billed blue magpie optimization algorithm to find the initial weight value of the BP neural network, determines the global optimal solution, directly uses the global optimal solution, effectively reduces the training time, reduces the cooling time and cost, introduces chaos mapping, increases the randomness of the population, reduces the risk of falling into a local optimal solution, uses dynamic weight, and compared with the traditional algorithm, accelerates the optimization speed and greatly reduces the optimization time.

[0049] 3. The present application trains the PID control parameter by combining the BP neural network to obtain the optimized PID control parameter, uses the neural network to effectively reduce the temperature control error, greatly improves the control ability, reduces the overshoot of the cooling speed compared with the traditional temperature control method, improves the energy efficiency ratio of the heat exchanger under the premise of accurate temperature control.

[0050] 4. The present application trains the GRU neural network early warning model to realize the temperature control early warning of the heat exchanger, the neural network is suitable for learning the correlation between temperature data and time, has high prediction accuracy, ensures that the abnormal temperature of the heat exchanger can be found in advance, realizes early warning, and guarantees stable operation and equipment safety.

[0051] Of course, any product implementing the present application does not necessarily need to achieve all the advantages described above. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and those skilled in the art can obtain other drawings according to these drawings without any creative effort.

[0053] Figure 1The heat exchanger automatic cooling circulation control system based on the Internet of Things provided by the application is a flow diagram of heat exchanger automatic cooling circulation control. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0055] In the description of the application, it should be understood that the terms "opening", "upper", "lower", "top", "middle", "inner" and the like indicate the orientation or positional relationship and are only used to facilitate the description of the application and simplify the description, and do not indicate or imply that the indicated component or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the application.

[0056] Embodiment 1

[0057] Please refer to Figure 1 The application is a heat exchanger automatic cooling circulation control method based on the Internet of Things, comprising the following steps:

[0058] S1, obtaining heat exchanger related parameters to form a heat exchanger related parameter data set, and establishing a heat exchanger automatic cooling circulation control model according to the heat exchanger related parameter data set;

[0059] The S1 comprises the following steps:

[0060] S11, obtaining heat exchanger related parameters, the heat exchanger related parameters including heat exchanger outlet temperature, cooling water mass, cooling water flow rate and the like, collecting data of the heat exchanger related parameters and giving time sequence to obtain heat exchanger related parameter data to form a heat exchanger related parameter data set A={A1, A2, A3,..., A m}, wherein A m represents the mth heat exchanger related parameter, and each heat exchanger related parameter contains n heat exchanger related parameter data; according to the heat exchanger related parameter data set, setting the cooling water flow rate as α1, the specific heat capacity of cooling water as α2, the heat exchanger inlet temperature as α3, the heat exchanger outlet temperature as α4, and the cooling water mass as β1, establishing a heat exchanger heat balance formula, and regarding the heat exchanger outlet temperature as a time function, using α4(t) to represent the heat exchanger outlet temperature at time t, then the heat exchanger heat balance formula is as follows:

[0061]

[0062] wherein, represents the integral of the heat exchanger outlet temperature at time t;

[0063] S12, Laplace transform is carried out on the heat exchanger heat balance formula to obtain a heat exchanger outlet temperature expression and a heat exchanger inlet temperature expression; the impedance of the cooling water transferred in the heat exchanger at time t is β2(t), the flow rate of the cooling water at time t is α1(t), b1, b2 and b3 represent constants, and the impedance calculation formula is as follows:

[0064]

[0065] The heat exchanger temperature first-order lag function, i.e., the heat exchanger automatic temperature reduction cycle control model, is obtained by conversion of the heat exchanger outlet temperature expression, the heat exchanger inlet temperature expression and the impedance calculation formula, and is as follows:

[0066]

[0067] wherein F(c) represents the heat exchanger temperature first-order lag function, c represents a Laplace transform complex variable, a represents a static gain, represents a lag time;

[0068] S2, a BP neural network is established, a chaotic mapping and a dynamic weight are introduced to improve the red-billed blue magpie optimization algorithm to obtain an improved red-billed blue magpie optimization algorithm, and the improved red-billed blue magpie optimization algorithm is used to find the initial weight of the BP neural network to obtain a global optimal solution;

[0069] The S2 includes the following steps:

[0070] S21, the BP neural network is set to consist of an input layer, a hidden layer and an output layer, an incremental PID controller is used to adjust the temperature control system of the heat exchanger, the incremental PID controller includes a controller input and a controller output, the deviation of the controller input and the controller output is recorded as a control deviation, the control deviation at time d is recorded as B(d), the control deviation at time d-1 is recorded as B(d-1), the control deviation at time d-2 is recorded as B(d-2), the proportional coefficient, the integral time coefficient and the differential time coefficient of the incremental PID controller are A', A'' and A''', respectively, and the control parameter calculation formula of the incremental PID controller at time d is as follows:

[0071] C(d) = C(d-1) + A'·(B(d)-B(d-1))+A''·B(d)+A'''·(B(d)-2B(d-1)+B(d-2));

[0072] Wherein, C(d) represents the control parameter of the incremental PID controller at d time, C(d-1) represents the control parameter of the incremental PID controller at d-1 time;

[0073] Taking B(d), B(d)-B(d-1) and B(d)-2B(d-1)+B(d-2) in the control parameter calculation formula of the incremental PID controller at d time as the input layer input of the BP neural network, taking the proportional coefficient, integral time coefficient and differential time coefficient of the incremental PID controller as the output layer output of the BP neural network, and establishing the BP neural network;

[0074] S22, obtain the control deviation of the incremental PID controller, obtain a control deviation set, divide the control deviation set into a deviation training set and a deviation test set, input the deviation training set into the BP neural network, and take the output error of the BP neural network as a fitness function; the improved Magpie optimization algorithm is obtained by introducing chaos mapping and dynamic weight to improve the Magpie optimization algorithm, and the initial weight of the BP neural network is optimized by using the improved Magpie optimization algorithm to obtain a global optimal solution, and the specific steps are as follows:

[0075] S221, the BP neural network is regarded as a search space, and there is a Magpie population in the search space, the number of the Magpie population is e, the Magpie individuals in the Magpie population represent the weights of the BP neural network, the upper limit and the lower limit of the search space are set, the dimension of the search space is j, the position of the Magpie individual in the Magpie population is initialized by using chaos mapping, and the Magpie population initialization is completed; after the Magpie population initialization, the Magpie population enters the foraging stage, the current iteration number is set as k, the position of the i-th Magpie individual in the k-th iteration is represented as D i (k), the position of the g-th Magpie individual in the k-th iteration is represented as D g (k), h1 represents a random number between the interval [0, 1], then the position of the i-th Magpie individual in the k+1-th iteration is represented as D i (k+1) is calculated as follows:

[0076]

[0077] The food position is set as D'(k), and the maximum iteration number is K, then the foraging coefficient is The Magpie population is divided into several Magpie groups, the number of the Magpie groups is set as e', at this time, the position of the Magpie individual D i (k+1) is updated according to the Magpie group,

[0078] The fitness function value corresponding to the current iteration of the individual position of the red-billed blue magpie is calculated, the red-billed blue magpie individual position where the current best fitness function value is located is taken as the current best weight value of the BP neural network, and a current optimal solution is obtained;

[0079] In S222, a dynamic weight is introduced, and the position of the hth red-billed blue magpie group in the kth iteration is set as E h (k), a weight is given to the red-billed blue magpie group, denoted as a weight set B1={δ1, δ2, δ3,..., δ h} where δ h represents the weight of the hth red-billed blue magpie group, h2 represents a random number between the interval [0, 1], and the dynamic coefficient φ h

[0080] The fitness function value corresponding to the position of the ith red-billed blue magpie individual in the k+1th iteration and the fitness function value corresponding to the position of the ith red-billed blue magpie individual in the kth iteration are recorded, and are denoted as the k+1th iteration fitness function value and the kth iteration fitness function value, respectively. When the k+1th iteration fitness function value is greater than the kth iteration fitness function value, the k+1th iteration fitness function value is used to replace the kth iteration fitness function value, otherwise it is not replaced, to obtain a first-stage red-billed blue magpie population. The iteration is repeated until the current iteration number reaches the maximum iteration number, and the iteration is stopped, to obtain a final red-billed blue magpie population. The red-billed blue magpie individual position corresponding to the best fitness function value in the final red-billed blue magpie population is denoted as a global optimal solution, and the global optimal solution is the initial weight value of the BP neural network;

[0081] In S3, the global optimal solution is taken as the initial weight value of the BP neural network, and the PID control parameters are trained in combination with the PID control algorithm to obtain optimized PID control parameters. The heat exchanger is controlled in a temperature cycle by using the optimized PID control parameters and the heat exchanger automatic temperature cycle control model, and the temperature cycle control is completed.

[0082] The S3 includes the following steps:

[0083] In S31, the global optimal solution is taken as the initial weight value of the BP neural network, and the BP neural network is trained until convergence is achieved to obtain a trained BP neural network. The PID control parameters are trained in combination with the trained BP neural network by using the PID control algorithm to obtain optimized PID control parameters. The specific steps are as follows:

[0084] ​S311, input the deviation test set into the trained BP neural network, collect the first iteration output result at the first iteration, set the error threshold as ω, record the first iteration output result and the size of the error threshold, the first iteration output result is the proportional coefficient, integral time coefficient and differential time coefficient of the incremental PID controller, and the control parameters of the first iteration incremental PID controller are calculated according to the control parameter calculation formula of the incremental PID controller at time d;

[0085] S312, at the second iteration, the control parameters of the second iteration incremental PID controller are calculated by adjusting the weights of the output layer and the hidden layer in the trained BP neural network, until the output result is less than the error threshold, the iteration is stopped, and the BP neural network model is obtained, at this time, the final output result is obtained, and the final output result is used as the optimized PID control parameter;

[0086] S32, the optimized PID control parameter corresponds to the optimized proportional coefficient, the optimized integral time coefficient and the optimized differential time coefficient in turn, the optimized PID control parameter is used as the input of the incremental PID controller, the incremental PID controller outputs the control signal, and the heat exchanger is controlled in the cooling cycle according to the heat exchanger automatic cooling cycle control model, and the specific steps are as follows:

[0087] S321, set the upper limit of the rated temperature of the heat exchanger as When the current temperature in the heat exchanger is greater than or equal to the rated temperature upper limit, at this time, the incremental PID controller receives the cooling signal of the temperature sensor, the control signal of the incremental PID controller is delayed according to the heat exchanger automatic cooling cycle control model, the cooling tower controls the injection of cooling water to cool the heat exchanger, and the incremental PID controller continues to monitor the heat exchanger to obtain the adjusted heat exchanger temperature;

[0088] S322, set the cooling water quality threshold and the cooling water flow rate threshold of the heat exchanger, when the current cooling water quality in the heat exchanger is greater than or equal to the cooling water quality threshold, reduce the cooling water flow rate, and open the heat exchanger drainage; when the adjusted heat exchanger temperature is greater than or equal to the rated temperature upper limit, increase the cooling water flow rate, when the current temperature in the heat exchanger is less than the rated temperature upper limit, stop the cooling process, otherwise, start the heat exchanger temperature abnormal alarm, and complete the heat exchanger cooling cycle control;

[0089] S4, train the GRU neural network to obtain a GRU neural network early warning model, output the heat exchanger temperature early warning value according to the heat exchanger related parameter data set and in combination with the GRU neural network early warning model, and realize heat exchanger temperature control early warning;

[0090] The S4 comprises the following steps:

[0091] S41. Obtain relevant parameters of the heat exchanger from previous years, including the heat exchanger temperature and outlet temperature from previous years. Assign these parameters to a time series with a time step of γ to obtain the data set of relevant parameters of the heat exchanger from previous years, G = {(p1,q1),(p2,q2),...,(p...}}. r ,q r )}, where p r q represents the heat exchanger temperature in the r-th year. r This represents the outlet temperature of the heat exchanger in the r-th year; the time series G′ corresponding to the set of heat exchanger related parameter data in previous years is G′={d1,d2,d3,..,d r}, where d r This represents the time of the r-th historical heat exchanger-related parameter data; the GRU neural network early warning model is trained using the historical heat exchanger-related parameter data set, and the specific steps are as follows:

[0092] S411. Normalize the data set of heat exchanger-related parameters from previous years to obtain a normalized parameter data set. Set a time window and divide the normalized parameter data set into a normalized parameter data segment set. Divide the normalized parameter data segment set into a sample training set and a sample test set. Set the GRU neural network to use forward propagation to obtain predicted values ​​and use the Adam algorithm to optimize the weights. Input the sample training set into the GRU neural network sequentially until convergence, and obtain a trained GRU neural network.

[0093] S412. Then, the sample test set is sequentially input into the trained GRU neural network, and a precision threshold is set. When the output prediction result is less than the precision threshold, the iteration is stopped, and the GRU neural network early warning model is obtained.

[0094] S42. Obtain heat exchanger related parameter data such as heat exchanger temperature and heat exchanger outlet temperature from the heat exchanger related parameter data set to obtain a heat exchanger early warning dataset. After normalizing the heat exchanger early warning dataset, input it into the GRU neural network early warning model to output the heat exchanger temperature early warning value. Set an early warning threshold. When the heat exchanger temperature early warning value is greater than the early warning threshold, the heat exchanger temperature is greater than or equal to the upper limit of the rated temperature after time step γ. Then, perform cooling treatment on the heat exchanger to realize heat exchanger temperature control early warning.

[0095] Example 2

[0096] This invention also discloses a system for an automatic cooling cycle control method for heat exchangers based on the Internet of Things, specifically including: a cooling model establishment module, a neural network parameter optimization module, a cooling cycle control module, and a heat exchanger temperature control early warning module;

[0097] The cooling model establishing module is configured to establish an automatic cooling cycle control model of the heat exchanger using heat exchanger related parameter data;

[0098] The neural network parameter optimization module is configured to find initial weights of the BP neural network using the improved magpie optimization algorithm;

[0099] The cooling cycle control module is configured to control the cooling cycle of the heat exchanger using the optimized PID control parameters and the cooling model;

[0100] The heat exchanger temperature control early warning module is configured to establish a neural network early warning model and output a heat exchanger temperature early warning value.

[0101] In the description of the present specification, the description of the terms "one embodiment", "an example", "a specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the invention. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0102] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details and limit the invention to the specific embodiments described. Obviously, many modifications and changes can be made according to the content of the present specification. The present specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can well understand and utilize the invention.

Claims

1. An automatic cooling cycle control method for a heat exchanger based on the Internet of Things, characterized by, It comprises the following steps: S1, obtaining heat exchanger related parameters, forming a heat exchanger related parameter data set, and establishing a heat exchanger automatic cooling cycle control model according to the heat exchanger related parameter data set; S2, establishing a BP neural network, introducing chaos mapping and dynamic weight to improve the optimization algorithm, obtaining an improved optimization algorithm, using the improved optimization algorithm to find the initial weight of the BP neural network, and obtaining the global optimal solution; S3, using the global optimal solution as the initial weight of the BP neural network, combining the PID control algorithm, training the PID control parameters, obtaining the optimized PID control parameters, using the optimized PID control parameters and the heat exchanger automatic cooling cycle control model to control the heat exchanger, and completing the cooling cycle control; S4, training the GRU neural network to obtain a GRU neural network early warning model, outputting a heat exchanger temperature early warning value according to the heat exchanger related parameter data set and the GRU neural network early warning model, and realizing heat exchanger temperature control early warning.

2. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 1, characterized in that, The S1 comprises the following steps: S11, obtaining heat exchanger related parameters, forming a heat exchanger related parameter data set, and establishing a heat exchanger heat balance formula; S12, Laplace transform is performed on the heat exchanger heat balance formula, and impedance is calculated, and the heat exchanger automatic cooling cycle control model is obtained.

3. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 2, characterized in that, The S2 comprises the following steps: S21, setting the BP neural network to comprise an input layer, a hidden layer and an output layer, using an incremental PID controller to adjust the temperature control system of the heat exchanger, the incremental PID controller comprising a controller input and a controller output, the deviation between the controller input and the controller output being a control deviation, the control deviation being input to the input layer of the BP neural network, the proportional coefficient, integral time coefficient and differential time coefficient of the incremental PID controller being output from the output layer of the BP neural network, and the BP neural network being established; S22, obtaining the control deviation of the incremental PID controller to obtain a control deviation set, dividing the control deviation set into a deviation training set and a deviation test set, inputting the deviation training set into the BP neural network, and using the output error of the BP neural network as a fitness function; introducing chaos mapping and dynamic weight to improve the optimization algorithm, obtaining an improved optimization algorithm, and using the improved optimization algorithm to optimize the initial weight of the BP neural network to obtain the global optimal solution.

4. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 3, characterized in that, The S22 comprises the following steps: S221, regarding the BP neural network as a search space, there are blue magpie populations in the search space, the blue magpie individuals in the blue magpie population represent the weights of the BP neural network, the positions of the blue magpie individuals in the blue magpie population are initialized using chaos mapping, and the blue magpie population initialization is completed; after the blue magpie population initialization, the blue magpie population enters the foraging stage, and the current iteration number is set; setting the food position and the maximum iteration number, dividing the blue magpie population into a plurality of blue magpie groups, setting the number of blue magpie groups, and updating the positions of the blue magpie individuals according to the blue magpie groups at this time; Calculate the fitness function value corresponding to the current iteration of the individual position of the red-billed blue magpie, and take the red-billed blue magpie individual position where the current best fitness function value is located as the current best weight value of the BP neural network, to obtain the current optimal solution; S222, introduce dynamic weights, give the red-billed blue magpie group weights, denoted as weight set, the red-billed blue magpie individual moves following the red-billed blue magpie group, and the dynamic weight is used to update the red-billed blue magpie individual position; Screen the best fitness function value to obtain the first-stage red-billed blue magpie population; repeat the iteration until the current iteration reaches the maximum iteration, stop the iteration, and obtain the final red-billed blue magpie population, the red-billed blue magpie individual position corresponding to the best fitness function value in the final red-billed blue magpie population is recorded as the global optimal solution, and the global optimal solution is the initial weight value of the BP neural network.

5. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 4, characterized in that, The S3 comprises the following steps: S31, taking the global optimal solution as the initial weight value of the BP neural network, training the BP neural network until convergence, and obtaining the trained BP neural network; Using the PID control algorithm and combining the trained BP neural network to train the PID control parameters to obtain the optimized PID control parameters; S32, the optimized PID control parameters correspond to the optimized proportional coefficient, the optimized integral time coefficient and the optimized differential time coefficient in turn, and the optimized PID control parameters are taken as the input of the incremental PID controller, and the incremental PID controller outputs the control signal, and the heat exchanger is controlled in the cooling cycle according to the heat exchanger automatic cooling cycle control model.

6. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 5, characterized in that, The S31 comprises the following steps: S311, input the deviation test set into the trained BP neural network, collect the first iteration output result at the first iteration, set the error threshold to ω, record the size of the first iteration output result and the error threshold, the first iteration output result is the proportional coefficient, the integral time coefficient and the differential time coefficient of the incremental PID controller, and the control parameters of the incremental PID controller at the first iteration are calculated according to the control parameter calculation formula of the incremental PID controller at time d; S312, at the second iteration, the control parameters of the incremental PID controller at the second iteration are calculated by adjusting the weights of the output layer and the hidden layer in the trained BP neural network until the output result is less than the error threshold, the iteration is stopped, and the BP neural network model is obtained, and the final output result is obtained at this time, and the final output result is taken as the optimized PID control parameter.

7. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 6, characterized in that, The S4 comprises the following steps: S41, obtain the relevant parameters of the heat exchanger in previous years and give the time sequence, the time step is γ, obtain the relevant parameter data set of the heat exchanger in previous years and the corresponding time sequence; the GRU neural network early warning model is trained using the relevant parameter data set of the heat exchanger in previous years; S42, obtain the heat exchanger early warning data set in the heat exchanger related parameter data set, input the normalized heat exchanger early warning data set into the GRU neural network early warning model, output the heat exchanger temperature early warning value, and realize the heat exchanger temperature control early warning.

8. The heat exchanger automatic cooling cycle control method based on the Internet of Things according to claim 7, characterized in that, The S41 comprises the following steps: S411、normalizing the heat exchanger related parameter data set of the previous years, obtaining a normalized parameter data set, setting a time window, dividing the normalized parameter data set using the time window, obtaining a normalized parameter data segment set, dividing the normalized parameter data segment set into a sample training set and a sample test set; setting the GRU neural network to obtain a predicted value using forward propagation, optimizing the weight using the Adam algorithm, inputting the sample training set into the GRU neural network in turn for training until convergence, obtaining a trained GRU neural network; S412、inputting the sample test set into the trained GRU neural network in turn, setting a precision threshold, stopping iteration when the output prediction result is less than the precision threshold, obtaining a GRU neural network early warning model.

9. A system for implementing the heat exchanger automatic cooling cycle control method based on the Internet of Things according to any one of claims 1-8, characterized in that, Specifically includes: a cooling model establishment module, a neural network parameter optimization module, a cooling cycle cooling control module and a heat exchanger temperature control early warning module; The cooling model establishment module is used to establish a heat exchanger automatic cooling cycle control model using heat exchanger related parameter data; The neural network parameter optimization module is used to find the initial weight of the BP neural network using the improved magpie robin optimization algorithm; The cooling cycle cooling control module is used to control the cooling cycle of the heat exchanger using the optimized PID control parameters and the cooling model; The heat exchanger temperature control early warning module is used to establish a neural network early warning model and output a heat exchanger temperature early warning value.

Images