A new laboratory remote monitoring and operation maintenance management system based on the Internet of Things

Abstract

The present invention discloses a new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things, which belongs to the field of laboratory monitoring and control technology. It is equipped with a parameter acquisition module, a control device, a parameter prediction module, an adjustment analysis module, and a PID controller. The parameter prediction module uses a trained cyclic network model to predict the subsequent change curve of the reference value of the controlled parameter in the laboratory based on the reference value curve of the current controlled parameter; the adjustment analysis module sets and corrects the reference period T1, and calculates the reference value of the controlled parameter at the time point t0+T1 as the final reference value based on the subsequent change curve of the predicted controlled parameter reference value. In the present invention, by calculating the deviation rate of the numerical curve of the controlled parameter value under different reference periods T1 and the subsequent change curve of the controlled parameter reference value, the reference period T1 with the highest accuracy is screened out as the final reference period value, thereby improving the monitoring and control accuracy of the controlled parameters of the internal environment of the laboratory.

Description

Technical Field

[0001] The present invention relates to the technical field of laboratory monitoring, and in particular to a novel laboratory remote monitoring and operation maintenance management system based on the Internet of Things. Background Art

[0002] Laboratory items and equipment are usually stored in laboratories. Some flammable substances and precious materials may be present in these items and equipment. In order to ensure the safety of the laboratory, it is necessary to monitor the laboratory or have laboratory supervisors inspect the laboratory. Laboratory monitoring equipment generally uses cameras to view the internal environment of the laboratory. However, since the monitoring server cannot observe the laboratory image at all times, it is impossible to take corresponding emergency measures in time according to changes in the internal environment of the laboratory.

[0003] To solve the above problems, the existing technology usually installs specific temperature sensors, humidity sensors and other structures at different locations in the laboratory to monitor the internal environment of the laboratory. In addition, the internal environment of the laboratory is regulated by corresponding control equipment such as fans and air conditioners to achieve the best experimental environment or storage environment for experimental items and equipment inside the laboratory.

[0004] Taking the internal temperature parameters of the laboratory as an example, the temperature in the laboratory is detected by a temperature sensor, and then the temperature change curve is set according to the required temperature of the laboratory. For example, the temperature target value is set to 25°C during the experiment, 28°C after the experiment, and 23°C at night to ensure that the internal environment of the laboratory in different time periods adapts to the different states inside the laboratory. The temperature in the laboratory is controlled by the control equipment driven by the target value. However, during the control process, the parameter control process is affected by the environmental factors and control rate in the laboratory. The parameter control process cannot ensure that its own control rate can match the required state of the environment, resulting in a large deviation between the internal environmental parameters of the laboratory and the actual parameter target values ​​to be adjusted, and the accuracy of the internal environmental monitoring and control of the laboratory is low. Summary of the invention

[0005] To this end, the present invention provides a novel laboratory remote monitoring and operation and maintenance management system based on the Internet of Things, which effectively solves the technical problems in the prior art that the control process is affected by the environmental factors and control rate in the laboratory, the parameter control process cannot ensure that its own control rate can match the required state of the environment, resulting in a large deviation between the internal environmental parameters of the laboratory and the actual target values ​​of the parameters to be adjusted, and the accuracy of the internal environmental monitoring and control of the laboratory is low.

[0006] In order to solve the above technical problems, the present invention specifically provides the following technical solutions: a new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things, having:

[0007] Parameter collection module, used to collect laboratory 0 The value of the controlled parameter at a point in time;

[0008] Control equipment, used to control the values ​​of controlled parameters in the laboratory;

[0009] A parameter prediction module introduces a cyclic network model, wherein the parameter prediction module obtains several groups of historical controlled parameter reference values ​​as training sets and trains the cyclic network model based on the training sets, and uses the trained cyclic network model to predict subsequent change curves of the controlled parameter reference values ​​in the laboratory based on the reference value curve of the current controlled parameter;

[0010] Adjust the analysis module, set and modify the reference period T 1 , based on the subsequent change curve of the predicted reference value of the controlled parameter, calculate t 0 +T 1 The reference value of the controlled parameter at a time point is used as the final reference value;

[0011] A PID controller is connected in series with the parameter acquisition module and the control device, and the PID controller controls the control device to adjust the value of the controlled parameter in the laboratory based on the comparison between the controlled parameter data and the final reference value;

[0012] Among them, in the reference period T 1 During the correction process, the reference period T is initialized 1 , calculate t according to the subsequent change curve of the predicted reference value of the controlled parameter 0 +T 1 The reference value of the controlled parameter at the time point is used as the basis for simulating the control cycle T 2 The numerical curve of the controlled parameter value in the control parameter is calculated, and the deviation rate between the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value is calculated. The above process is repeated to measure the deviation rate between the controlled parameter value and the subsequent change curve of the controlled parameter reference value in different reference periods T 1 The accuracy of the controlled parameter value is used to select the reference period T with the highest accuracy. 1 As the current regulation cycle T 2 The final reference period value.

[0013] Further,

[0014] Simulation control cycle T 2 The numerical curve of the internal controlled parameter value includes the following steps:

[0015] Establish the state transfer matrix of the laboratory parameter control system:

[0016]

[0017] In the formula, S represents the reference value of the controlled parameter, S 1is the value of the corresponding parameter outside the laboratory, S 2 is the value of the controlled parameter in the laboratory, β represents the relationship between the value of the controlled parameter in the laboratory and the reference value of the controlled parameter, τ represents the power output power of the control equipment, and μ represents the relationship between the voltage load of the control equipment and the value of the controlled parameter;

[0018] The state transfer matrix of the laboratory parameter control system is used to establish a laboratory parameter change rate model in combination with the changes in the values ​​of the controlled parameters:

[0019]

[0020] Where L(t) represents the sum of environmental variables, and V(p) represents the function of regulating the output of the device.

[0021] Further,

[0022] Calculation of t based on laboratory parameter change rate model 0 The parameter change rate at a time point, combined with t 0 The value of the controlled parameter at a time point is plotted backwards over the control period T 2 Numerical curve of the internal controlled parameter value.

[0023] Further,

[0024] The calculation process of the deviation rate specifically includes the following steps:

[0025] The current control cycle T is calculated based on the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value. 2 The difference between the controlled parameter value and the controlled parameter reference value corresponding to each control time node;

[0026] Set the upper limit of the difference;

[0027] Calculate the control cycle T 2 The deviation rate a% is obtained by calculating the ratio of the number of differences exceeding the upper limit of the difference to the total number of differences.

[0028] Further,

[0029] Take (1-a%) as the accuracy of the controlled parameter value in the current reference period, and select the reference period T with the highest accuracy in turn. 1 ;

[0030] In each deviation rate calculation process, if the reference period T with the highest accuracy rate is finally selected 1 If it exceeds one, reset the upper limit of the difference and recalculate the deviation rate and accuracy.

[0031] Further,

[0032] The specific implementation steps of the PID controller for remote parameter monitoring include:

[0033] Constructing PID control model:

[0034]

[0035] In the formula, u(t) is the control quantity, e(t) is the error signal, K p , K i , K d They are proportional coefficient, integral coefficient, differential coefficient, K p e(t) is the proportional term, is the integral term, is the differential term;

[0036] The network performance indicators are calculated using the RBF neural network:

[0037] In the formula, J represents the network performance index, Y represents the RBF neural network output, and h i represents the Gaussian function, y(k) represents the center of the kth radial basis function, y m (k) represents the center of the mth radial basis function, s i represents the i-th perceptual variable, c i represents the center of the i-th radial basis function;

[0038] Calculate the change sensitivity of the controlled parameters based on the Jacobian matrix, and adjust the parameters of the PID control model according to the change sensitivity and network performance indicators;

[0039] The sensitivity is expressed as w represents the parameter adjustment amount, C represents the parameter adjustment ratio, g represents the momentum factor, and η represents the learning factor.

[0040] Further,

[0041] The parameters of the adjusted PID control model are

[0042] ΔK p , ΔK i , ΔK d Respectively represent K p , K i , K d The adjustment amount, -η p 、-η i 、-η d Represent the learning rates of proportion, integration, and differentiation respectively.

[0043] Further,

[0044] The construction and training process of the recurrent network model includes the following steps:

[0045] Create a three-layer neural network consisting of an input layer, a hidden layer, and an output layer;

[0046] Set the training function, connection function and output function of the hidden layer, connection layer and output layer;

[0047] Set the network's expected minimum error, maximum number of iterations, and learning rate;

[0048] Input the training set into the recurrent network model and train it;

[0049] Adjust the weights of each layer in the recurrent network model according to the error;

[0050] Determine whether the recurrent network model has converged, and the training is completed after convergence;

[0051] Among them, when the error is less than the minimum expected error, the algorithm converges;

[0052] When the maximum number of iterations is reached, the algorithm ends.

[0053] Further,

[0054] Inputting the reference values ​​of the controlled parameters within the time period into the trained recurrent network model in sequence, and outputting subsequent predicted values ​​of the reference values ​​of the controlled parameters;

[0055] Based on the subsequent predicted values ​​of the reference values ​​of the controlled parameters, a subsequent change curve of the reference values ​​of the controlled parameters in the laboratory is drawn.

[0056] Further,

[0057] In t 0 Before the time point parameter adjustment work is carried out, obtain t 0 The actual value of the reference value of the controlled parameter at a certain time point is input into the trained recurrent network model to predict the subsequent change curve of the reference value of the controlled parameter.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] In the present invention, the subsequent change curve of the reference value of the controlled parameter is predicted by the cyclic network model, and a certain reference period T is set. 1 , calculate t on the subsequent change curve of the reference value of the controlled parameter 0 +T 1 The reference value of the controlled parameter at the time point is used as the final reference value, and the value of the controlled parameter in the laboratory is regulated according to the final reference value. 1The deviation rate between the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value is used to calculate the deviation rate between the controlled parameter value and the subsequent change curve of the controlled parameter reference value in different reference periods T. 1 The accuracy of the controlled parameter value is used to select the reference period T with the highest accuracy. 1 As the final reference period value, realize the reference period T 1 The correction of the value is carried out in the reference period T 1 The final reference value obtained is the optimal reference value after training and analysis, which can enable the controlled parameter value to change in a state closest to the reference value curve of the controlled parameter during the controlled parameter regulation process, thereby improving the monitoring and regulation accuracy of the controlled parameters in the laboratory internal environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0061] Figure 1 A structural block diagram of a novel laboratory remote monitoring and operation and maintenance management system based on the Internet of Things provided by an embodiment of the present invention;

[0062] Figure 2 is the control period T in the embodiment of the present invention 2 A numerical curve diagram of the controlled parameter value when the reference value of the internally predicted controlled parameter remains unchanged;

[0063] Figure 3 is the control period T in the embodiment of the present invention 2 A numerical curve diagram of the controlled parameter value when the reference value of the internally predicted controlled parameter changes;

[0064] Figure 4 It is a schematic diagram of a subsequent change curve of a reference value of a controlled parameter and a numerical curve of a numerical value of the controlled parameter in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] 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.

[0066] like Figure 1As shown, the present invention provides a novel laboratory remote monitoring and operation maintenance management system based on the Internet of Things, which is equipped with a parameter acquisition module, a control device, a parameter prediction module, a regulation analysis module and a PID controller.

[0067] The parameter acquisition module is used to collect the laboratory 0 The value of the controlled parameter at a time point. If the controlled parameter is temperature, the parameter acquisition module uses a temperature sensor. If the controlled parameter is humidity, the corresponding parameter acquisition module uses a humidity sensor.

[0068] The control equipment is used to control the values ​​of the controlled parameters in the laboratory, and control equipment such as fans and air conditioners can be used.

[0069] The parameter prediction module introduces a recurrent network model. The parameter prediction module obtains several groups of historical controlled parameter reference values ​​as training sets and trains the recurrent network model based on the training sets. The trained recurrent network model is used to predict the subsequent change curve of the controlled parameter reference value in the laboratory based on the reference value curve of the current controlled parameter.

[0070] Adjust the analysis module to set and modify the reference period T 1 , based on the subsequent change curve of the predicted reference value of the controlled parameter, calculate t 0 +T 1 The reference value of the controlled parameter at a time point is used as the final reference value;

[0071] The PID controller is connected in series with the parameter acquisition module and the control device. The PID controller controls the control device to adjust the value of the controlled parameter in the laboratory based on the comparison between the controlled parameter data and the final reference value;

[0072] Among them, in the reference period T 1 During the correction process, the reference period T is initialized 1 , calculate t according to the subsequent change curve of the predicted reference value of the controlled parameter 0 +T 1 The reference value of the controlled parameter at a time point is used to simulate the control cycle T 2 The numerical curve of the controlled parameter value in the control parameter is calculated, and the deviation rate between the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value is calculated. The above process is repeated to measure the deviation rate between the controlled parameter value and the subsequent change curve of the controlled parameter reference value in different reference periods T 1 The accuracy of the controlled parameter value is used to select the reference period T with the highest accuracy. 1 As the current regulation cycle T 2 The final reference period value.

[0073] In the present invention, the subsequent change curve of the reference value of the controlled parameter is predicted by the cyclic network model, and a certain reference period T is set. 1, calculate t on the subsequent change curve of the reference value of the controlled parameter 0 +T 1 The reference value of the controlled parameter at the time point is used as the final reference value, and the value of the controlled parameter in the laboratory is regulated according to the final reference value. 1 The deviation rate between the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value is used to measure the deviation rate between the controlled parameter value and the subsequent change curve of the controlled parameter reference value in different reference periods T. 1 The accuracy of the controlled parameter value is used to select the reference period T with the highest accuracy. 1 As the final reference period value, realize the reference period T 1 The correction of the value is carried out in the reference period T 1 The final reference value obtained is the optimal reference value after training and analysis, which can enable the controlled parameter value to change in a state closest to the reference value curve of the controlled parameter during the controlled parameter regulation process, thereby improving the monitoring and regulation accuracy of the controlled parameters in the laboratory internal environment.

[0074] In the present invention, the controlled parameter value refers to the value of the controlled parameter, such as the temperature value, humidity value, etc., and the controlled parameter reference value refers to the reference value of the controlled parameter, and the controlled parameter can be regulated according to the reference value.

[0075] In the present invention, the simulation control cycle T 2 The numerical curve of the internal controlled parameter value includes the following steps:

[0076] Establish the state transfer matrix of the laboratory parameter control system:

[0077]

[0078] In the formula, S represents the reference value of the controlled parameter, S 1 is the value of the corresponding parameter outside the laboratory, S 2 is the value of the controlled parameter in the laboratory, β represents the relationship between the value of the controlled parameter in the laboratory and the reference value of the controlled parameter, τ represents the power output power of the control equipment, and μ represents the relationship between the voltage load of the control equipment and the value of the controlled parameter;

[0079] The state transfer matrix of the laboratory parameter control system is used to establish a laboratory parameter change rate model in combination with the changes in the values ​​of the controlled parameters:

[0080]

[0081] Where L(t) represents the sum of environmental variables, and V(p) represents the function of regulating the output of the device.

[0082] Calculation of t based on laboratory parameter change rate model 0The parameter change rate at a time point, combined with t 0 The value of the controlled parameter at a time point is plotted backwards over the control period T 2 Numerical curve of the internal controlled parameter value.

[0083] In the above embodiments, the numerical values ​​of the corresponding parameters outside the laboratory, the relevant parameters of the controlled parameter numerical control equipment in the laboratory, and the sum of the environmental variables in the laboratory are introduced, and the influence of the indoor and outdoor environment and the operating rate of the control equipment itself on the laboratory parameter change rate is considered, so as to construct a laboratory parameter change rate model, and t can be calculated according to the laboratory parameter change rate model. 0 The parameter change rate at a time point, combined with t 0 The value of the controlled parameter at a time point is plotted backwards over the control period T 2 Numerical curve of the internal controlled parameter value.

[0084] In the process of actual control equipment, the value of the controlled parameter generally changes at a certain parameter change rate. Assuming the control cycle T 2 The reference value of the controlled parameter of the internal prediction has not changed, such as Figure 2 As shown, after knowing t 0 The parameter change rate at the time point and t 0 After obtaining the controlled parameter value at a certain time point, a straight line with a specific starting point and a specific slope can be drawn backwards, which is used as the control period T 2 Numerical curve of the internal controlled parameter value.

[0085] In addition, due to the regulation cycle T 2 The reference value of the controlled parameter predicted in the internal control cycle T may change. 2 The reference value of the controlled parameter for the prediction changes, which results in a change in the laboratory parameter conversion rate, such as Figure 3 As shown, before the time point when the change occurs, t 0 The parameter change rate at the time point and t 0 The controlled parameter value at the time point is drawn backward with an inclined line segment, and after the time point when the change occurs, the slope of the curve is adjusted according to the recalculated laboratory parameter change rate, and the inclined straight line is continued to be drawn backward.

[0086] The plotting of the numerical curves of the above two controlled parameter values ​​is carried out when other conditions inside and outside the laboratory remain unchanged. However, in the actual application process, since this process is a process of simulating the process of plotting the numerical curves of the controlled parameter values, the values ​​of the corresponding parameters outside the laboratory and other data with little correlation can be assumed to remain unchanged, and the controlled parameter value S in the laboratory 2 It will inevitably change with the regulation process, so in the regulation cycle T 2The change rate of the laboratory parameters changes in real time, so the control cycle T can be drawn in the simulation. 2 In the process of the numerical curve of the internal controlled parameter value, the slope is continuously adjusted according to the calculated parameter change rate value to form an adaptive simulation numerical curve, which is more consistent with the change process of the actual controlled parameter value.

[0087] When drawing the control cycle T 2 After obtaining the numerical curve of the internal controlled parameter value, it is necessary to calculate the deviation rate between the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value.

[0088] The calculation process of the deviation rate specifically includes the following steps:

[0089] Based on the numerical curve of the controlled parameter value, calculate the current control cycle T 2 The difference between the controlled parameter value and the controlled parameter reference value corresponding to each control time node;

[0090] Set the upper limit of the difference;

[0091] Calculate the control cycle T 2 The deviation rate a% is obtained by calculating the ratio of the number of differences exceeding the upper limit of the difference to the total number of differences.

[0092] by Figure 4 As an example, the dotted straight line at the top is the subsequent change curve of the reference value of the controlled parameter, and the solid curve at the bottom is the numerical curve of the simulated controlled parameter value. The vertical distance between the two lines corresponding to each control time node (the position of the hollow point in the figure) on the numerical curve of the controlled parameter value is calculated, that is, the difference between the controlled parameter value and the controlled parameter reference value corresponding to each control time node is calculated, and the upper limit value y of the difference is set. Assuming that the control cycle T 2 The difference exceeding the upper limit of the difference is the difference at the first point, and the deviation rate is the ratio of the number of differences exceeding the upper limit of the difference to the total number of differences, resulting in a deviation rate of 20%.

[0093] The above process is an example of the calculation process of the deviation rate. In actual application, in the control cycle T 2 When a monitoring time node is reached, the current time point is taken as t 0 , calculate t according to the subsequent change curve of the predicted reference value of the controlled parameter 0 +T 1 The reference value of the controlled parameter at a time point is used to calculate the laboratory parameter change rate, and the numerical curve corresponding to the simulated controlled parameter value changes.

[0094] Afterwards, (1-a%) is used as the accuracy of the controlled parameter value in the current reference period, and the reference period T with the highest accuracy is selected in turn. 1 According to the above example, 80% is taken as the accuracy of the controlled parameter value under the reference period.

[0095] According to the above steps, different reference periods T are calculated. 1 Different reference values ​​of controlled parameters are given below.

[0096] In each deviation rate calculation process, if the reference period T with the highest accuracy rate is finally selected 1 If more than one is exceeded, the upper limit of the difference is reset and the deviation rate and accuracy are recalculated until the reference period T with the highest accuracy is screened out. 1 It is used as a final reference value, and the control equipment in actual control performs control work based on the final reference value.

[0097] From the above implementation process, the remote monitoring process of the controlled parameters of the laboratory is as follows:

[0098] (1) Assume that at t 0 The reference value of the controlled parameter (the actual reference value of the controlled parameter) has been set at all times, which is input into the recurrent network model, combined with historical data to form a training set, and the subsequent change curve of the reference value of the controlled parameter is predicted;

[0099] (2) Set a reference period T 1 The value of the reference period T 1 The subsequent change curve of the reference value of the controlled parameter is calculated on the subsequent change curve of the reference value of the controlled parameter. 0 +T 1 Reference value of the controlled parameter at a time point;

[0100] Substitute the calculated reference value of the controlled parameter into the laboratory parameter change rate model to calculate t 0 The parameter change rate at a time point, combined with t 0 The controlled parameter value at a time point (the detected controlled parameter value) is plotted backwards against the control cycle T 2 Numerical curve of the internal controlled parameter value;

[0101] In the process of simulating the numerical curve of the controlled parameter value, time nodes with equal time intervals can be set. At the first time node, the numerical curve of the controlled parameter value is drawn backwards with the first calculated parameter change rate and the actual controlled parameter value. When the second time node is reached, the numerical curve is drawn again with this point as t 0 At the time point, calculate the subsequent change curve of the reference value of the controlled parameter t 0 +T 1The reference value of the controlled parameter at the time point is used to calculate the parameter change rate with the calculated reference value of the controlled parameter. The numerical curve of the controlled parameter value is drawn backward with the second calculated parameter change rate and the current curve endpoint position, and so on, until the control cycle T is drawn. 2 The numerical curve of all the controlled parameter values;

[0102] The above time node can be set as an actual control time node or other time node. The control time node refers to the time node of each control action in the actual control.

[0103] (3) Adjustment of reference period T 1 The value of and repeat step (2) to obtain multiple reference periods T 1 The numerical curve of the controlled parameter value corresponding to the value of ;

[0104] (4) Calculate each reference period T according to the steps 1 The deviation rate between the numerical curve of the controlled parameter value corresponding to the value of and the subsequent change curve of the controlled parameter reference value is obtained to obtain each reference period T 1 The accuracy of the numerical curve of the controlled parameter value corresponding to the value of ;

[0105] Adjust the upper limit of the difference until the numerical curve of the controlled parameter with the highest accuracy is screened out, and set its corresponding reference period T 1 The value of is taken as the final reference value;

[0106] (5) The PID controller uses the final reference value as the target value and controls the control equipment based on the current laboratory controlled parameter data to adjust the controlled parameter value in the laboratory.

[0107] In the above embodiment, only the control period T 2 The numerical curves of all the controlled parameter values ​​are screened and the control cycle T 2 It can also be adjusted according to the actual situation. If the controlled parameter fluctuates greatly during a certain period of time, the control cycle T can be shortened. 2 If the controlled parameter fluctuates slightly within a certain period of time, the control cycle T can be increased. 2 On the one hand, it can avoid the impact of excessive fluctuations on the results, and on the other hand, it can improve the screening accuracy by increasing the cycle.

[0108] In the prior art, at t 0 The reference value of the controlled parameter set at the time is usually the value that the controlled parameter in the laboratory is expected to reach. If the value is not reached when the next control time node arrives, the next control action will cause the overall change curve of the controlled parameter to deviate too much from the reference value curve of the controlled parameter. Therefore, in the present invention, the reference value of the controlled parameter set at t is not set.0 The reference value of the controlled parameter set at the moment is used as the expected value of the controlled parameter in the current laboratory, but 0 +T 1 The reference value of the controlled parameter corresponding to the prediction at the moment is taken as the target value. Taking the process of adjusting the value of the controlled parameter driven by the control device as an example, theoretically, t 0 +T 1 The reference value of the controlled parameter corresponding to the prediction at time t must be greater than that at t 0 The reference value of the controlled parameter set at the moment is increased by increasing the target value by increasing the reference value of the controlled parameter to speed up the control rate of the control device, so that the controlled parameter reaches the value in a more optimized state. The above method of changing the target value of the control device to change the control process will not cause the controlled parameter to be over-adjusted, because under continuous training and analysis, the controlled parameter adjustment reaches the initially set reference value of the controlled parameter and continues to adjust to the target value. Before the deviation is too large, it will enter the next control, and T 1 The value can be 0. At this time, the initially set controlled reference value is used as the expected value of the controlled parameter in the current laboratory. That is to say, in the screening of the most suitable reference period T 1 During the process, the initially set reference value of the controlled parameter itself will be used as the target value to simulate the numerical curve of the controlled parameter value, and compared with the numerical curves of the controlled parameter values ​​obtained when the reference value of the controlled parameter is increased in other cases, so as to obtain the target value with the highest accuracy and the lowest deviation rate, which is the final reference value.

[0109] In the present invention, the specific implementation steps of the PID controller for remote parameter monitoring include:

[0110] Constructing PID control model:

[0111]

[0112] In the formula, u(t) is the control quantity, e(t) is the error signal, K p , K i , K d They are proportional coefficient, integral coefficient, differential coefficient, K p e(t) is the proportional term, is the integral term, is the differential term;

[0113] The network performance indicators are calculated using the RBF neural network:

[0114] In the formula, J represents the network performance index, Y represents the RBF neural network output, and h i represents the Gaussian function, y(k) represents the center of the kth radial basis function, y m (k) represents the center of the mth radial basis function, si represents the i-th perceptual variable, c i represents the center of the i-th radial basis function;

[0115] Calculate the change sensitivity of the controlled parameters based on the Jacobian matrix, and adjust the parameters of the PID control model according to the change sensitivity and network performance indicators;

[0116] The sensitivity is expressed as w represents the parameter adjustment amount, C represents the parameter adjustment ratio, g represents the momentum factor, and η represents the learning factor.

[0117] The parameters of the adjusted PID control model are

[0118] ΔK p , ΔK i , ΔK d Respectively represent K p , K i , K d The adjustment amount, -η p 、-η i 、-η d Represent the learning rates of proportion, integration, and differentiation respectively.

[0119] The above steps complete the establishment of the PID control model and the adjustment of its parameters. In the specific implementation process, the performance indicators of the neural network can determine the adjustment of the PID control model parameters, thereby controlling the working rate of the entire model.

[0120] In the present invention, a parameter prediction module introduces a recurrent network model. The parameter prediction module obtains several groups of historical controlled parameter reference values ​​as training sets and trains the recurrent network model based on the training sets. The trained recurrent network model is used to predict the subsequent change curve of the controlled parameter reference value in the laboratory based on the reference value curve of the current controlled parameter.

[0121] The construction and training process of the recurrent network model specifically includes the following steps:

[0122] Create a three-layer neural network consisting of an input layer, a hidden layer, and an output layer;

[0123] Set the training function, connection function and output function of the hidden layer, connection layer and output layer;

[0124] Set the network's expected minimum error, maximum number of iterations, and learning rate;

[0125] Input the training set into the recurrent network model and train it;

[0126] Adjust the weights of each layer in the recurrent network model according to the error;

[0127] Determine whether the recurrent network model has converged, and the training is completed after convergence;

[0128] Among them, when the error is less than the minimum expected error, the algorithm converges;

[0129] When the maximum number of iterations is reached, the algorithm ends.

[0130] The reference values ​​of the controlled parameters within the time period are sequentially input into the trained cyclic network model, and the subsequent predicted values ​​of the reference values ​​of the controlled parameters are output. Based on the subsequent predicted values ​​of the reference values ​​of the controlled parameters, the subsequent change curve of the reference values ​​of the controlled parameters in the laboratory is drawn.

[0131] In the present invention, at t 0 Before the time point parameter adjustment work is carried out, obtain t 0 The actual value of the reference value of the controlled parameter at a time point is input into the trained recurrent network model to predict the subsequent change curve of the reference value of the controlled parameter, which can improve the prediction accuracy of the subsequent change curve of the reference value of the controlled parameter.

[0132] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. A new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things, characterized in that: have: Parameter acquisition module, used to collect the controlled parameter values ​​at time point t0 in the laboratory; Control equipment, used to control the values ​​of controlled parameters in the laboratory; A parameter prediction module introduces a cyclic network model, wherein the parameter prediction module obtains several groups of historical controlled parameter reference values ​​as training sets and trains the cyclic network model based on the training sets, and uses the trained cyclic network model to predict subsequent change curves of the controlled parameter reference values ​​in the laboratory based on the reference value curve of the current controlled parameter; The adjustment analysis module sets and modifies the reference period T1, and calculates the reference value of the controlled parameter at the time point t0+T1 as the final reference value based on the subsequent change curve of the predicted reference value of the controlled parameter; A PID controller is connected in series with the parameter acquisition module and the control device, and the PID controller controls the control device to adjust the value of the controlled parameter in the laboratory based on the comparison between the controlled parameter data and the final reference value; Wherein, in the correction process of the reference period T1, the reference period T1 is initialized, and the reference value of the controlled parameter at the time point t0+T1 is calculated according to the subsequent change curve of the predicted reference value of the controlled parameter, and the numerical curve of the controlled parameter value in the control period T2 is simulated based on the reference value of the controlled parameter, and the deviation rate between the numerical curve of the controlled parameter value and the subsequent change curve of the reference value of the controlled parameter is calculated, and the above process is repeated to measure the accuracy of the controlled parameter value under different reference periods T1 to select the reference period T1 with the highest accuracy as the final reference period value of the current control period T2; The numerical curve of the controlled parameter value in the simulation control period T2 includes the following steps: Establish the state transfer matrix of the laboratory parameter control system: In the formula, S represents the reference value of the controlled parameter, S1 is the value of the corresponding parameter outside the laboratory, S2 is the value of the controlled parameter in the laboratory, β represents the relationship between the controlled parameter value in the laboratory and the reference value of the controlled parameter, τ represents the power output power of the control equipment, and μ represents the relationship between the voltage load of the control equipment and the value of the controlled parameter; The state transfer matrix of the laboratory parameter control system is used to establish a laboratory parameter change rate model in combination with the changes in the values ​​of the controlled parameters: In the formula, L(t) represents the sum of environmental variables, and V(p) represents the function of regulating the output of the device; The calculation process of the deviation rate specifically includes the following steps: Based on the numerical curve of the controlled parameter value and the subsequent change curve of the controlled parameter reference value, the difference between the controlled parameter value and the controlled parameter reference value corresponding to each control time node in the current control cycle T2 is calculated; Set the upper limit of the difference; The ratio of the number of differences exceeding the upper limit of the difference within the control period T2 to the total number of differences is calculated to obtain the deviation rate a%.

2. According to the new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things in claim 1, it is characterized in that: The parameter change rate at time point t0 is calculated based on the laboratory parameter change rate model, and the numerical curve of the controlled parameter value within the control period T2 is drawn backward in combination with the controlled parameter value at time point t0.

3. The new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things according to claim 1 is characterized in that: Taking (1-a%) as the accuracy of the controlled parameter value in the current reference period, the reference period T1 with the highest accuracy is selected in turn; In each deviation rate calculation process, if the reference period T1 with the highest accuracy rate finally screened out exceeds one, the upper limit of the difference is reset and the deviation rate and accuracy rate are recalculated.

4. The new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things according to claim 1 is characterized in that: The construction and training process of the recurrent network model includes the following steps: Create a three-layer neural network consisting of an input layer, a hidden layer, and an output layer; Set the training function, connection function and output function of the hidden layer, connection layer and output layer; Set the network's expected minimum error, maximum number of iterations, and learning rate; Input the training set into the recurrent network model and train it; Adjust the weights of each layer in the recurrent network model according to the error; Determine whether the recurrent network model has converged, and the training is completed after convergence; Among them, when the error is less than the minimum expected error, the algorithm converges; When the maximum number of iterations is reached, the algorithm ends.

5. The new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things according to claim 4 is characterized in that: Inputting the reference values ​​of the controlled parameters within the time period into the trained recurrent network model in sequence, and outputting subsequent predicted values ​​of the reference values ​​of the controlled parameters; Based on the subsequent predicted values ​​of the reference values ​​of the controlled parameters, a subsequent change curve of the reference values ​​of the controlled parameters in the laboratory is drawn.

6. The new type of laboratory remote monitoring and operation maintenance management system based on the Internet of Things according to claim 5 is characterized in that: Before the parameter control work at time point t0 is performed, the actual value of the reference value of the controlled parameter at time point t0 is obtained and input into the trained recurrent network model to predict the subsequent change curve of the reference value of the controlled parameter.

Images