Extreme learning machine fuzzy PID control method and system for central air conditioning water system

Abstract

The present invention discloses a fuzzy PID control method and system of an extreme learning machine for a central air-conditioning water system, and relates to the technical field of central air-conditioning control. The method comprises: determining the PID initial parameters of a fuzzy PID controller; using the output values ​​at different times and the output data of the central air-conditioning water system as input data of the extreme learning machine, and using the output data of the central air-conditioning water system at the corresponding time lag as output data for training; predicting the output data of the central air-conditioning water system with time lag according to the trained extreme learning machine; adjusting the PID initial parameters of the fuzzy PID controller according to fuzzy rules based on the predicted values ​​and the set temperature to obtain the adjusted fuzzy PID controller; controlling the central air-conditioning water system according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature. The present invention can improve the energy-saving effect of the central air-conditioning water system.

Description

Technical Field

[0001] The present invention relates to the technical field of central air-conditioning control, and in particular to an extreme learning machine fuzzy PID control method and system for a central air-conditioning water system. Background Art

[0002] Facing the major demand of accelerating the promotion of green and low-carbon development and realizing the "dual carbon" goal as soon as possible, it emphasizes "adhering to the principle of energy conservation priority and deepening energy conservation in industries, transportation, construction and other fields and public institutions". With the rapid development of China's economy, more and more high-rise buildings are rising from the ground, and the energy consumption of these buildings is also increasing in the proportion of total social energy consumption. Among them, the energy consumption ratio of central air-conditioning systems in these buildings is as high as 40% to 60%. People's daily life and industrial production are inseparable from central air-conditioning. Therefore, studying effective energy-saving control methods for central air-conditioning systems is one of the keys to solving the problem of energy conservation and consumption reduction in China's construction field. Due to the characteristics of nonlinearity, strong coupling, and large hysteresis of the central air-conditioning system, the optimization of a single device can only reduce its own energy consumption, but will increase the energy consumption of other equipment, and cannot effectively reduce the energy consumption of the overall system. The water system in the central air-conditioning system mainly includes the chilled water system and the cooling water system. Most air conditioning water systems still use traditional control methods such as constant flow control and PID control. Although they can reduce energy consumption to a certain extent, in the face of frequent switching of set temperatures and changes in ambient environmental factors, multiple iterations are required in the entire PID control process to output the final stable feedback value of the PID controller, which is time-consuming. In addition, the large deviation between the feedback value in the iterative process and the final stable feedback value may cause overshoot and oscillation. Due to the problems of overshoot, oscillation, long transition time, etc., which easily lead to energy waste, it is difficult for the air conditioning system to achieve a good energy-saving effect. Summary of the invention

[0003] The purpose of the present invention is to provide an extreme learning machine fuzzy PID control method and system for a central air-conditioning water system to solve the problem of poor energy-saving effect of the central air-conditioning water system.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] An extreme learning machine fuzzy PID control method for a central air conditioning water system, comprising:

[0006] Based on the set temperature of the central air-conditioning water system, the PID initial parameters of the fuzzy PID controller are determined by using the PID parameter empirical tuning method.

[0007] Obtain the output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined, the output data of the central air-conditioning water system at the corresponding time, and the output data of the central air-conditioning water system with a time lag at the corresponding time; the output data of the central air-conditioning water system is the measured temperature;

[0008] The output values ​​at different times and the output data of the central air-conditioning water system at the corresponding time are used as the input data of the extreme learning machine, and the output data of the central air-conditioning water system with a time lag at the corresponding time is used as the output data of the extreme learning machine to train the extreme learning machine, thereby obtaining a trained extreme learning machine;

[0009] The time-delayed output data of the central air-conditioning water system is predicted based on the trained extreme learning machine to obtain the predicted value;

[0010] Based on the predicted value and the set temperature, adjusting the PID initial parameters of the fuzzy PID controller according to fuzzy rules to obtain an adjusted fuzzy PID controller;

[0011] The central air-conditioning water system is controlled according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature.

[0012] Optionally, based on the set temperature of the central air-conditioning water system, a PID parameter empirical tuning method is used to determine the PID initial parameters of the fuzzy PID controller, specifically including:

[0013] Based on the set temperature of the central air-conditioning water system, the PID parameter empirical tuning method is adopted to adjust the proportional coefficient, integral coefficient and differential coefficient in turn to determine the initial proportional coefficient, initial integral coefficient and initial differential coefficient of the fuzzy PID controller.

[0014] Optionally, the training process of the extreme learning machine specifically includes:

[0015] The output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined and the output data of the central air-conditioning water system at the corresponding time are stored in the database as the input data set of the extreme learning machine; the output data of the central air-conditioning water system with a time lag at the corresponding time is stored in the database as the output data set of the extreme learning machine;

[0016] An input data set and an output data set of the extreme learning machine are extracted from a database; the input data set and the output data set are input into the extreme learning machine, and the extreme learning machine is trained offline to obtain a trained extreme learning machine.

[0017] Optionally, the extreme learning machine is:

[0018]

[0019] Among them, β i is the output weight corresponding to the i-th hidden layer neuron; G(·) is the activation function; W i is the input weight corresponding to the i-th hidden layer neuron; X j is the jth input data; b i is the bias corresponding to the i-th hidden layer neuron; L is the number of hidden layer neurons; o j For the input data X j The corresponding output data; N is the number of input data in the training data set.

[0020] Optionally, the output data of the central air conditioning water system with time lag is predicted according to the trained extreme learning machine to obtain a predicted value, which specifically includes:

[0021] The output value of the fuzzy PID controller after the initial parameters are determined at the current moment and the output data of the central air-conditioning water system at the current moment are input into the extreme learning machine to predict the time-delayed output data of the central air-conditioning water system to obtain the predicted value.

[0022] Optionally, based on the predicted value and the set temperature, adjusting the PID initial parameters of the fuzzy PID controller according to fuzzy rules to obtain an adjusted fuzzy PID controller specifically includes:

[0023] Determining a deviation and a deviation change rate according to the predicted value and the set temperature;

[0024] Determining adjusted PID parameters according to the deviation and the deviation change rate through fuzzy rules;

[0025] A tuned fuzzy PID controller is obtained based on the tuned PID parameters.

[0026] An extreme learning machine fuzzy PID control system for a central air conditioning water system, comprising:

[0027] An initial parameter determination module is used to determine the PID initial parameters of the fuzzy PID controller based on the set temperature of the central air-conditioning water system by using the PID parameter empirical tuning method;

[0028] A data acquisition module is used to obtain the output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined, the output data of the central air-conditioning water system at the corresponding time, and the output data of the central air-conditioning water system with a time lag at the corresponding time; the output data of the central air-conditioning water system is the measured temperature;

[0029] A training module is used to use the output values ​​at different times and the output data of the central air-conditioning water system at the corresponding time as input data of the extreme learning machine, and use the output data of the central air-conditioning water system with a time lag at the corresponding time as the output data of the extreme learning machine to train the extreme learning machine, so as to obtain a trained extreme learning machine;

[0030] A prediction module is used to predict the time-delayed output data of the central air-conditioning water system according to the trained extreme learning machine to obtain a predicted value;

[0031] A PID parameter adjustment module, used for adjusting the PID initial parameters of the fuzzy PID controller according to fuzzy rules based on the predicted value and the set temperature to obtain an adjusted fuzzy PID controller;

[0032] The control module is used to control the central air-conditioning water system according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature.

[0033] Optionally, the training module specifically includes:

[0034] A storage unit is used to store the output values ​​of the fuzzy PID controller at different times during the operation process after the initial parameters are determined and the output data of the central air-conditioning water system at the corresponding time as the input data set of the extreme learning machine in the database; and store the output data of the central air-conditioning water system at the corresponding time with a time lag as the output data set of the extreme learning machine in the database;

[0035] An extraction unit, used to extract an input data set and an output data set of the extreme learning machine from a database;

[0036] The offline training unit inputs the input data set and the output data set into the extreme learning machine, performs offline training on the extreme learning machine, and obtains a trained extreme learning machine.

[0037] Optionally, the prediction module specifically includes:

[0038] The prediction unit is used to input the output value of the fuzzy PID controller after the initial parameters are determined at the current moment and the output data of the central air-conditioning water system at the current moment into the extreme learning machine to predict the delayed output data of the central air-conditioning water system to obtain a predicted value.

[0039] Optionally, the PID parameter adjustment module specifically includes:

[0040] a deviation and deviation change rate determination unit, configured to determine the deviation and the deviation change rate according to the predicted value and the set temperature;

[0041] A PID parameter adjustment unit, used to determine an adjusted PID parameter according to the deviation and the deviation change rate through a fuzzy rule;

[0042] The adjusted fuzzy PID controller determining unit is used to obtain an adjusted fuzzy PID controller based on the adjusted PID parameters.

[0043] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0044] The present invention provides an extreme learning machine fuzzy PID control method and system for a central air-conditioning water system. An extreme learning machine is applied to the prediction of a nonlinear system. The extreme learning machine is used to predict the time-delayed output data of the central air-conditioning water system, namely, a predicted value. The predicted value is used as a feedback value in a fuzzy PID controller closed-loop system. The feedback value and a set temperature are used as inputs of the fuzzy PID controller to offset the feedback value generated by the fuzzy PID controller itself, thereby reducing the problem of decreased control effect due to feedback value interference, time delay and the like. The final PID control parameters are obtained based on the feedback value and the set temperature. The final PID control parameters can be obtained without multiple iterations, effectively reducing the over-adjustment time of the central air-conditioning water system, so that the central air-conditioning water system can reach an ideal operating state more quickly as the environment changes. Moreover, since the final PID control parameters are directly obtained, the problem of overshoot and oscillation caused by excessive deviation between the feedback value in the iteration process and the final stable feedback value is avoided, thereby achieving a control effect that is both fast and accurate with almost no overshoot, thereby improving the energy-saving effect of the central air-conditioning water system. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A flow chart of the extreme learning machine fuzzy PID control method for a central air conditioning water system provided by the present invention;

[0047] Figure 2 A schematic diagram of the process of the extreme learning machine fuzzy PID control method for a central air conditioning water system provided by the present invention;

[0048] Figure 3 A schematic diagram of the principle of the extreme learning machine fuzzy PID control method for a central air-conditioning water system provided by the present invention;

[0049] Figure 4 A schematic diagram of the extreme learning machine neural network structure provided by the present invention;

[0050] Figure 5 A Simulink simulation diagram of the extreme learning machine fuzzy PID control method for the central air conditioning water system provided by the present invention;

[0051] Figure 6 A comparison diagram of step response curves of four control methods for the central air conditioning water system provided by the present invention;

[0052] Figure 7 A comparison diagram of step response error curves of four control methods for the central air conditioning water system provided by the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] The purpose of the present invention is to provide an extreme learning machine fuzzy PID control method and system for a central air-conditioning water system to solve the problem of poor energy-saving effect of the central air-conditioning water system.

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] The present invention aims at the problem of high overall energy consumption of the central air-conditioning water system, and proposes a method based on extreme learning machine prediction of feedback value and fusion of fuzzy PID rules to achieve energy-saving control of the central air-conditioning water system. The extreme learning machine is applied to the prediction of nonlinear systems. The improved controller uses the extreme learning machine as an intelligent prediction model in the feedback loop, and effectively integrates it with the adaptive fuzzy PID control algorithm in the control loop, thereby avoiding or overcoming the adverse effects of feedback signal interference and lag. The extreme learning machine fusion fuzzy PID controller can not only ensure rapidity, but also ensure control accuracy, achieve no overshoot and reduce transition adjustment time, so that the central air-conditioning water system can reach the ideal operating state faster as the environment changes, thereby obtaining a certain degree of energy-saving control effect and saving the energy consumption of the central air-conditioning water system.

[0057] Since most of the equipment in the central air-conditioning water system is a high-order system that is difficult to calculate and process, a low-order approximation method is generally used to establish a model of the central air-conditioning system. The second-order function model contains two time constants, which can be flexibly adjusted and is closer to a high-order system. Therefore, the central air-conditioning water system control model can be approximately regarded as a second-order delayed model, and this control model has high accuracy and can fully meet the needs of simulation research. Therefore, the present invention uses a delayed second-order model transfer function to approximate the central air-conditioning water system control model, that is, the cooling water temperature difference model, and conducts simulation research based on this. The second-order model transfer function is shown in Formula 1.

[0058]

[0059] Among them, G'(·) is the transfer function of the cooling water temperature difference model; s is the complex variable of Laplace transform; and e is the delay operator.

[0060] like Figure 1 As shown, the present invention discloses an extreme learning machine fuzzy PID control method for a central air conditioning water system, comprising:

[0061] Step 1: Based on the set temperature of the central air-conditioning water system, the PID parameter empirical tuning method is used to determine the PID initial parameters of the fuzzy PID controller.

[0062] The central air conditioning water system input adopts a step signal, and the return water temperature of the cooling water pump main pipe is set to 25°C. The unit step response of the central air conditioning water system input is observed by an oscilloscope. Specifically, based on the set temperature of 25°C of the central air conditioning water system, the PID parameter empirical tuning method is used to adjust the proportional coefficient, integral coefficient and differential coefficient in turn to determine the initial proportional coefficient K of the fuzzy PID controller. p , initial integration coefficient K i and the initial differential coefficient K d Slowly increase the proportional coefficient, the return water temperature of the cooling water pump main pipe of the central air conditioning water system increases rapidly. When the overshoot is small and the central air conditioning water system responds quickly, stop increasing the proportional coefficient and determine the initial proportional coefficient K p ; Then introduce the time constant T i Eliminate the steady-state error of the central air conditioning water system and appropriately reduce the time constant T i , that is, gradually enhance the integral action, and determine the time constant T when the deviation e(k=∞)=0 is satisfied i The value of the initial integral coefficient K i Finally, by analyzing the change rate of the central air conditioning water system deviation signal, differential regulation is introduced to reduce the overshoot of the system, and when the adjustment time is short, the initial differential coefficient K is determined. dAs a specific embodiment, the initial proportional coefficient K of the fuzzy PID controller set by the present invention is p is 0.141, initial integral coefficient K i is 0.00235 and the initial differential coefficient K d is 0.5.

[0063] Step 2: Obtain the output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined, the output data of the central air-conditioning water system at the corresponding time, and the output data of the central air-conditioning water system with a time lag at the corresponding time. The output data of the central air-conditioning water system is the measured temperature.

[0064] Since traditional neural networks are greatly affected by the learning rate and have a slow training speed, Professor Huang Guangbin of Nanyang Technological University has conducted in-depth research on neural networks and proposed a new feedforward neural network, the Extreme Learning Machine (ELM). It can be seen that this algorithm has the advantages of arbitrary selection of input weights and arbitrary setting of the number of hidden layer nodes. It only needs to calculate the output weight matrix of the hidden layer to complete the training of the feedforward neural network. The entire feedforward neural network training process only needs one calculation to obtain the relevant parameters, which greatly reduces the amount of calculation. At the same time, the algorithm also has good generalization ability and strong applicability, and can be widely used in many fields.

[0065] The extreme learning machine is trained based on the input data and output data of the constructed fuzzy PID controller at different times and is a nonlinear prediction model. The ELM neural network contains n input layer neurons, L hidden layer neurons and m output layer neurons. The specific structure is as follows Figure 4 shown.

[0066] After the fuzzy PID controller completes the parameter setting, a more accurate and effective ELM training data set can be obtained. Training the ELM based on a more accurate and effective ELM training data set enables the ELM to better learn, understand and predict the dynamic characteristics of the control system, thereby improving the performance of the closed-loop control of the entire central air-conditioning water system. Specifically, the historical data of the fuzzy PID controller during operation after the initial parameters are determined (i.e., the output values ​​at different times and the output data of the central air-conditioning water system at the corresponding times) is stored in the database as the input data set of the extreme learning machine; the output data of the central air-conditioning water system with a time lag at the corresponding time is stored in the database as the output data set of the extreme learning machine. The input data set and output data set of the extreme learning machine are extracted from the database as a more accurate and targeted ELM training data set (X i ,t i ), where X iis the i-th input data in the ELM input data set, X i =[x i1 ,x i2 ,...,x in ] T ∈R n ;t i The ELM output data set is concentrated with the input data X i The corresponding output data, t i =[t i1 ,t i2 ,...,t im ] T ∈R m .

[0067] Step 3: Use the output values ​​at different times and the output data of the central air-conditioning water system at the corresponding time as the input data of the extreme learning machine, and use the output data of the central air-conditioning water system with time lag at the corresponding time as the output data of the extreme learning machine to train the extreme learning machine to obtain a trained extreme learning machine.

[0068] The input data set and the output data set are input into the extreme learning machine, and the extreme learning machine is trained offline to obtain a trained extreme learning machine.

[0069] The ELM neural network structure can be expressed as:

[0070]

[0071] Among them, β i is the output weight corresponding to the i-th hidden layer neuron; G(·) is the activation function, which is a nonlinear piecewise continuous function that satisfies the universal approximation ability theorem. The commonly used one is the Sigmoid function, as shown in formula (3); W i is the input weight, which can be expressed as W i =[w i1 ,w i2 ,...,w in ] T ;X j is the jth input data; b i is the bias of the i-th hidden layer neuron; L is the number of hidden layer neurons in the ELM neural network; o j For the input data X j The corresponding output data; N is the number of input data in the training data set.

[0072]

[0073] Specifically, the output weight β corresponding to the i-th hidden layer neuron is obtained using formulas (2) and (3): i, the bias b of the i-th hidden layer neuron i And the output data o corresponding to the jth input data j . Among them, the number of neurons in the hidden layer is L=10.

[0074] When the hidden layer nodes meet the requirements, ELM can approximate any function. Generally, the hidden layer neurons are set larger than the input neurons to give it better generalization ability and minimize the error, that is:

[0075]

[0076] That is, there is β i , W i and b i So that the following relationship holds:

[0077]

[0078] In addition, the matrix form of the ELM neural network can be expressed as:

[0079] Hβ=T (6)

[0080] Among them, H is the output matrix of hidden layer neurons; β is the output weight matrix; T is the expected output matrix.

[0081]

[0082]

[0083]

[0084] To find a single hidden layer neural network, let:

[0085]

[0086] The above formula (10) can be equivalent to minimizing the loss function E:

[0087]

[0088] Based on formula (6) to formula (9), formula (12) can be obtained as follows:

[0089]

[0090] As a specific embodiment, when formula (12) is obtained, the output matrix H, output weight matrix β and expected output matrix T of the hidden layer neurons can also be inversely obtained according to formula (6).

[0091] In the ELM neural network, if the input weights and the bias of the hidden layer neurons are determined, the output matrix of the hidden layer can be determined, which can be simplified into a linear relationship, as shown in formula (6). It can be seen that the least squares solution of the output weights can be calculated by formula (13).

[0092]

[0093] Among them, H + is the pseudo-inverse of the output weight matrix, which can be obtained by singular value decomposition; T is the expected output matrix, which can be obtained by formula (6); is the least squares solution of the output weights.

[0094] From the above formula (13), we can see that ELM not only ensures learning accuracy, but also has good applicability, strong generalization ability and fast solution speed.

[0095] like Figure 5 As shown, the present invention also provides a Simulink simulation diagram based on the extreme learning machine fuzzy PID control method. The general structure function of the prediction model of the nonlinear controlled object (central air conditioning water system) is as follows:

[0096]

[0097] The system model structure of the nonlinear controlled object can be described as follows:

[0098] y(k+1)=f(y(k-1),y(k-2),...,y(kn),u(k),u(k-1),...,u(km)) (15)

[0099] Among them, y(k) is the output of the central air-conditioning water system; u(k) is the input of the central air-conditioning water system; f(·) is the nonlinear function of the central air-conditioning water system established based on the neural network.

[0100] Let the input of the neural network of the central air conditioning water system be:

[0101] P=[y(k),...,y(kn),u(k),...u(km)] (16)

[0102] Then the neural network estimation model of the central air conditioning water system is:

[0103]

[0104] in, is the output of the neural network; is an estimate of f(·).

[0105] Step 4: Predict the time-delayed output data of the central air-conditioning water system according to the trained extreme learning machine to obtain the predicted value.

[0106] The output value u(k) of the fuzzy PID controller after the initial parameters are determined at the current moment and the output data y(k) of the central air-conditioning water system at the current moment are input into the extreme learning machine as the input data of the extreme learning machine at time k. The output data of the central air-conditioning water system with time lag is predicted to obtain the output data y(k+d) of the central air-conditioning water system at time k+d, that is, the predicted value, where d is the time lag, d=30.

[0107] Step 5: Based on the predicted value and the set temperature, the PID initial parameters of the fuzzy PID controller are adjusted according to fuzzy rules to obtain an adjusted fuzzy PID controller.

[0108] Determine the deviation e and the deviation change rate e according to the predicted value y(k+d) and the set temperature r(k) c Through fuzzy rules, according to the deviation e and the deviation change rate e c An adjusted PID parameter is determined, and an adjusted fuzzy PID controller is obtained based on the adjusted PID parameter.

[0109] As a specific embodiment, according to the deviation e between the predicted value y(k+d) and the set temperature r(k) and the change rate e c , and refer to the fuzzy rules to adaptively adjust the parameters of the fuzzy PID controller. Sampling the controlled object according to the sampling period T, the control quantity (measured temperature of cooling water) of the fuzzy PID control based on the extreme learning machine prediction is obtained by the following formula:

[0110]

[0111] Among them, u(k) is the output of the fuzzy PID controller at the kth sampling moment; e(k) is the deviation between the actual output of the controlled object at the kth sampling moment and the target output; e(i) is the deviation between the actual output of the controlled object at the ith sampling moment and the target output; 0≤i≤k; through the formula and Find the integral coefficient; Δk p , Δk i , Δk d K p , K i and K d increment.

[0112] The fuzzification process is the process of fuzzifying the input of the fuzzy controller to obtain the fuzzy value of the input. The fuzzification process requires two inputs, one is the deviation value e between the output of the controlled object and the target value (set temperature), and the other is the deviation change rate e. c , e c It is the ratio of the deviation value e(k) between the output of the controlled object and the target value at the current sampling moment to the difference value e(k-1) between the output of the controlled object and the target value at the previous moment before the current moment.

[0113] Determine the goal of fuzzifying the two input quantities, that is, define the two input quantities into fuzzy subsets, each fuzzy subset is represented by a fuzzy language value. c Defined into 7 fuzzy subsets {NB, NM, NS, ZO, PS, PM, PB}, where NB, NM, NS, ZO, PS, PM and PB are fuzzy language values, representing negative large, negative medium, negative small, zero, positive small, positive medium and positive large respectively. The two input quantities are defined into fuzzy subsets by using Gaussian membership function, as shown in formula (19).

[0114]

[0115] Among them, x is the input variable, a and b are parameters of different membership functions. By adjusting the parameters a and b, the shape of the function can be changed to adapt to the control requirements of different systems and different input variables in different value ranges.

[0116] Refer to the fuzzy control parameters Δk respectively p , Δk i and Δk d The two-dimensional control table of e and e is shown in Table 1, Table 2 and Table 3. c The fuzzy quantity can find the corresponding fuzzy control parameter Δk p , Δk i , Δk d The value of .

[0117] Table 1 Fuzzy control parameters Δk p Two-dimensional control table

[0118]

[0119] Table 2 Fuzzy control parameters Δk i Two-dimensional control table

[0120]

[0121] Table 3 Fuzzy control parameters Δk d Two-dimensional control table

[0122]

[0123] Step 6: Control the central air-conditioning water system according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature.

[0124] According to e and e c The fuzzy quantity is used to find the corresponding fuzzy control parameter Δk p , Δk i , Δk d After taking the value of p , Δk i , Δk d The value of is taken as the final PID control parameter of the fuzzy PID controller and the central air-conditioning water system is controlled so that the central air-conditioning water system reaches the set temperature and achieves a fast response and stable control effect.

[0125] In summary, if Figure 2 and Figure 3 As shown, the control system architecture of the present invention can be divided into two main components, which work together to achieve fast, accurate and efficient energy-saving control. The present invention first introduces the offline trained extreme learning machine as a nonlinear prediction model into the feedback control system, which is responsible for receiving input data from the controlled object and predicting the current output based on its previous training data. After the real-time predicted output is compared with the set temperature, it will be passed to the fuzzy PID controller. At time k, the input u(k) and output y(k) of the controlled object are predicted by the extreme learning machine to obtain the output y(k+d) at time k+d and compared with the set temperature r(k) to obtain the deviation e; secondly, according to the deviation e and the deviation change rate e c According to the fuzzy rules, the parameters of the fuzzy PID controller are adjusted to enable the entire closed-loop control system to have a faster, more accurate response and a stable control effect. Therefore, based on the fuzzy PID controller, the present invention uses an extreme learning machine as a nonlinear prediction model, and uses the extreme learning machine to predict the feedback signal to overcome the influence of interference, time lag and other factors on the control performance of the closed-loop control system.

[0126] The primary task of the fuzzy PID controller is to receive the predicted output from the extreme learning machine and generate corresponding control signals to adjust the controlled object to achieve better control results. Its uniqueness lies in the adaptability of the fuzzy PID controller. It can continuously adjust its output parameters based on the parameter initialization value according to the deviation and its rate of change between the predicted output and the set temperature according to the fuzzy rules to achieve better control performance. The ELM nonlinear prediction model can better reflect the dynamic characteristics of the controlled object. At the same time, with the cooperation of the fuzzy PID adaptive adjustment, it ensures that the central air-conditioning cooling water system can still obtain fast and accurate control effects in the presence of adverse factors such as interference and time lag.

[0127] After adjusting the parameters at the current moment, continue to the next moment and restart the prediction.

[0128] Figure 6 This is a comparison diagram of the step response curves of the four control methods for the central air conditioning water system provided by the present invention. Figure 7 This is a comparison chart of the step response error curves of the four control methods for the central air conditioning water system provided by the present invention. Figure 6 and Figure 7 It can be seen that different control methods have different adjustment times for the set temperature. Traditional PID control, fuzzy PID control and single neuron control methods all have different degrees of overshoot. The transition time to the initial set temperature of 25℃ ±5% steady state takes 3-5 minutes, while the extreme learning machine fuzzy PID control method has a good control effect, and the transition time takes only 1 minute to enter the steady state set temperature ±5%; from the adjustment state, the initial adjustment amount of PID control and other algorithms exceeds the initial temperature, increasing energy consumption. From this point of view, it can be seen that the use of extreme learning machine fuzzy PID for feedback loop nonlinear prediction can make the control system run faster, more stable and more accurate, and has a good control effect on the electrical energy saving of the air conditioning water system.

[0129] like Figure 6 and Figure 7 As shown in the figure, by comparing the control effects of the fuzzy PID control predicted by the extreme learning machine and several other control algorithms, it can be clearly seen that this new prediction algorithm has a faster response speed and more stable and accurate control than the traditional fuzzy PID algorithm and single neuron algorithm.

[0130] Compared with the existing central air-conditioning water system control method technology, the central air-conditioning water system extreme learning machine fuzzy PID control method provided by the present invention has the following advantages:

[0131] (1) Different from the traditional single closed-loop negative feedback control system, the present invention adopts an extreme learning machine to predict the dynamic changes, interference and hysteresis of the controlled object, thereby improving the accuracy of the entire closed-loop control system.

[0132] (2) Compared with traditional PID, single neuron PID, fuzzy PID and other control methods, the extreme learning machine fuzzy PID control method of the central air-conditioning water system of the present invention has the advantages of faster response, smaller overshoot, and shorter adjustment time, avoiding energy waste caused by overshoot, oscillation and long transition time, and has a certain degree of energy-saving control effect.

[0133] The present invention also provides an extreme learning machine fuzzy PID control system for a central air conditioning water system, comprising:

[0134] The initial parameter determination module is used to determine the PID initial parameters of the fuzzy PID controller based on the set temperature of the central air-conditioning water system by adopting the PID parameter empirical tuning method.

[0135] The data acquisition module is used to obtain the output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined, the output data of the central air-conditioning water system at the corresponding time, and the output data of the central air-conditioning water system with a time lag at the corresponding time; the output data of the central air-conditioning water system is the measured temperature.

[0136] The training module is used to use the output values ​​at different times and the output data of the central air-conditioning water system at the corresponding time as the input data of the extreme learning machine, and use the output data of the central air-conditioning water system with time lag at the corresponding time as the output data of the extreme learning machine to train the extreme learning machine to obtain a trained extreme learning machine.

[0137] The prediction module is used to predict the time-delayed output data of the central air-conditioning water system according to the trained extreme learning machine to obtain the predicted value.

[0138] The PID parameter adjustment module is used to adjust the PID initial parameters of the fuzzy PID controller according to fuzzy rules based on the predicted value and the set temperature to obtain an adjusted fuzzy PID controller.

[0139] The control module is used to control the central air-conditioning water system according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature.

[0140] Optionally, the training module specifically includes:

[0141] The storage unit is used to store the output values ​​of the fuzzy PID controller at different times during the operation after the initial parameters are determined and the output data of the central air-conditioning water system at the corresponding time as the input data set of the extreme learning machine in the database; and store the output data of the central air-conditioning water system with a time lag at the corresponding time as the output data set of the extreme learning machine in the database.

[0142] The extraction unit is used to extract the input data set and the output data set of the extreme learning machine from the database.

[0143] The offline training unit inputs the input data set and the output data set into the extreme learning machine, performs offline training on the extreme learning machine, and obtains a trained extreme learning machine.

[0144] Optionally, the prediction module specifically includes:

[0145] The prediction unit is used to input the output value of the fuzzy PID controller after the initial parameters are determined at the current moment and the output data of the central air-conditioning water system at the current moment into the extreme learning machine to predict the delayed output data of the central air-conditioning water system to obtain a predicted value.

[0146] Optionally, the PID parameter adjustment module specifically includes:

[0147] The deviation and deviation change rate determination unit is used to determine the deviation and the deviation change rate according to the predicted value and the set temperature.

[0148] The PID parameter adjustment unit is used to determine the adjusted PID parameter according to the deviation and the deviation change rate through fuzzy rules.

[0149] The adjusted fuzzy PID controller determining unit is used to obtain an adjusted fuzzy PID controller based on the adjusted PID parameters.

[0150] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0151] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A central air conditioning water system extreme learning machine fuzzy PID control method, characterized in that: include: Based on the set temperature of the central air-conditioning water system, the PID initial parameters of the fuzzy PID controller are determined by using the PID parameter empirical tuning method. Obtain the output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined, the output data of the central air-conditioning water system at the corresponding time, and the output data of the central air-conditioning water system with a time lag at the corresponding time; the output data of the central air-conditioning water system is the measured temperature; The output values ​​at different times and the output data of the central air-conditioning water system at the corresponding time are used as the input data of the extreme learning machine, and the output data of the central air-conditioning water system with a time lag at the corresponding time is used as the output data of the extreme learning machine to train the extreme learning machine, thereby obtaining a trained extreme learning machine; The time-delayed output data of the central air-conditioning water system is predicted based on the trained extreme learning machine to obtain the predicted value; Based on the predicted value and the set temperature, adjusting the PID initial parameters of the fuzzy PID controller according to fuzzy rules to obtain an adjusted fuzzy PID controller; The central air-conditioning water system is controlled according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature.

2. The extreme learning machine fuzzy PID control method for a central air conditioning water system according to claim 1, characterized in that: Based on the set temperature of the central air-conditioning water system, the PID parameter empirical tuning method is used to determine the PID initial parameters of the fuzzy PID controller, including: Based on the set temperature of the central air-conditioning water system, the PID parameter empirical tuning method is adopted to adjust the proportional coefficient, integral coefficient and differential coefficient in turn to determine the initial proportional coefficient, initial integral coefficient and initial differential coefficient of the fuzzy PID controller.

3. The extreme learning machine fuzzy PID control method for a central air conditioning water system according to claim 2, characterized in that: The training process of the extreme learning machine specifically includes: The output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined and the output data of the central air-conditioning water system at the corresponding time are stored in the database as the input data set of the extreme learning machine; the output data of the central air-conditioning water system with a time lag at the corresponding time is stored in the database as the output data set of the extreme learning machine; Extracting an input data set and an output data set of the extreme learning machine from a database; The input data set and the output data set are input into the extreme learning machine, and the extreme learning machine is trained offline to obtain a trained extreme learning machine.

4. The extreme learning machine fuzzy PID control method for a central air conditioning water system according to claim 1, characterized in that: The extreme learning machine is: Among them, β i is the output weight corresponding to the i-th hidden layer neuron; G(·) is the activation function; W i is the input weight corresponding to the i-th hidden layer neuron; X j is the jth input data; b i is the bias corresponding to the i-th hidden layer neuron; L is the number of hidden layer neurons; o j For the input data X j The corresponding output data; N is the number of input data in the training data set.

5. The extreme learning machine fuzzy PID control method for a central air conditioning water system according to claim 4, characterized in that: The time-delayed output data of the central air-conditioning water system is predicted based on the trained extreme learning machine to obtain the predicted value, including: The output value of the fuzzy PID controller after the initial parameters are determined at the current moment and the output data of the central air-conditioning water system at the current moment are input into the extreme learning machine to predict the time-delayed output data of the central air-conditioning water system to obtain the predicted value.

6. The extreme learning machine fuzzy PID control method for a central air conditioning water system according to claim 5, characterized in that: Based on the predicted value and the set temperature, adjusting the PID initial parameters of the fuzzy PID controller according to the fuzzy rule to obtain the adjusted fuzzy PID controller specifically includes: Determining a deviation and a deviation change rate according to the predicted value and the set temperature; Determining adjusted PID parameters according to the deviation and the deviation change rate through fuzzy rules; A tuned fuzzy PID controller is obtained based on the tuned PID parameters.

7. A central air conditioning water system extreme learning machine fuzzy PID control system, characterized in that: include: An initial parameter determination module is used to determine the PID initial parameters of the fuzzy PID controller based on the set temperature of the central air-conditioning water system by using the PID parameter empirical tuning method; A data acquisition module is used to obtain the output values ​​of the fuzzy PID controller at different times during operation after the initial parameters are determined, the output data of the central air-conditioning water system at the corresponding time, and the output data of the central air-conditioning water system with a time lag at the corresponding time; the output data of the central air-conditioning water system is the measured temperature; A training module is used to use the output values ​​at different times and the output data of the central air-conditioning water system at the corresponding time as input data of the extreme learning machine, and use the output data of the central air-conditioning water system with a time lag at the corresponding time as the output data of the extreme learning machine to train the extreme learning machine, so as to obtain a trained extreme learning machine; A prediction module is used to predict the time-delayed output data of the central air-conditioning water system according to the trained extreme learning machine to obtain a predicted value; A PID parameter adjustment module, used for adjusting the PID initial parameters of the fuzzy PID controller according to fuzzy rules based on the predicted value and the set temperature to obtain an adjusted fuzzy PID controller; The control module is used to control the central air-conditioning water system according to the adjusted fuzzy PID controller so that the central air-conditioning water system reaches the set temperature.

8. The extreme learning machine fuzzy PID control system for a central air conditioning water system according to claim 7, characterized in that: The training module specifically includes: A storage unit is used to store the output values ​​of the fuzzy PID controller at different times during the operation process after the initial parameters are determined and the output data of the central air-conditioning water system at the corresponding time as the input data set of the extreme learning machine in the database; and store the output data of the central air-conditioning water system at the corresponding time lag as the output data set of the extreme learning machine in the database; An extraction unit, used to extract an input data set and an output data set of the extreme learning machine from a database; The offline training unit inputs the input data set and the output data set into the extreme learning machine, performs offline training on the extreme learning machine, and obtains a trained extreme learning machine.

9. The extreme learning machine fuzzy PID control system for a central air conditioning water system according to claim 7, characterized in that: The prediction module specifically includes: The prediction unit is used to input the output value of the fuzzy PID controller after the initial parameters are determined at the current moment and the output data of the central air-conditioning water system at the current moment into the extreme learning machine to predict the delayed output data of the central air-conditioning water system to obtain a predicted value.

10. The extreme learning machine fuzzy PID control system for a central air conditioning water system according to claim 7, characterized in that: The PID parameter adjustment module specifically includes: a deviation and deviation change rate determination unit, configured to determine the deviation and the deviation change rate according to the predicted value and the set temperature; A PID parameter adjustment unit, used for determining an adjusted PID parameter according to the deviation and the deviation change rate through a fuzzy rule; The adjusted fuzzy PID controller determining unit is used to obtain an adjusted fuzzy PID controller based on the adjusted PID parameters.

Images