Drug calcining machine model prediction control system and method based on entropy weight method

Abstract

The invention discloses a medicine calcining machine model prediction control system and method based on an entropy weight method, and the method comprises the steps: building a nonlinear model based on the quality data, temperature data and electric heating machine real-time power in a medicinal material processing process, building a system state-space equation, and carrying out the linear processing of the system state-space equation, designing a state observer to observe the value of a first-order derivative of the real-time temperature of the medicinal material; performing discretization processing on the system state space equation, and constructing a discretization system; for a discretization system, predicting a system state in a period of time in the future by using a rolling optimization method; designing a quadratic form control law, and obtaining a weight coefficient of a system state and a weight coefficient of an input variable in the quadratic form control law by using an entropy weight method; and the predicted system state in a period of time in the future is substituted in, constraint conditions are set, a system input variable sequence is solved, and the first input variable in the solved sequence serves as the electric heating machine power at the next moment to be applied to the medicine forging machine.

Description

Technical Field

[0001] The invention relates to the field of system control and pharmaceutical manufacturing, and in particular to a model prediction control system and method for a medicine calcining machine based on an entropy weight method. Background Art

[0002] In the pharmaceutical industry, especially in the production of traditional Chinese medicine, temperature is an important environmental factor that affects the efficacy and quality of medicinal materials. Too high or too low temperature will have an adverse effect on the quality of traditional Chinese medicine. However, in the pharmaceutical industry, there are a large number of equipment and process links that require temperature control, such as calcining machines, dryers, reactors, etc.

[0003] Traditional methods of controlling the temperature of medicinal materials are to use constant temperature equipment or manual adjustment. These methods have problems such as inaccurate temperature control, slow response, and high energy consumption. The more advanced control method is PID control, but it lacks the ability to monitor and dynamically adjust temperature distribution in real time. In addition, these methods are difficult to adjust parameters, are sensitive to interference, and have performance limitations. In addition, PID controllers may be sensitive to noise and external disturbances, causing the system output to become unstable when facing high-frequency noise. Summary of the invention

[0004] The purpose of the present invention is to provide a model predictive control system and method for a medicine forging machine based on the entropy weight method, so as to solve the problems of inaccurate temperature control, slow response, high energy consumption, etc. when using a medicine forging machine for processing medicinal materials, and to improve the system control accuracy and efficiency.

[0005] In order to achieve the above tasks, the present invention adopts the following technical solutions:

[0006] A model predictive control system for a calcining machine based on an entropy weight method, comprising:

[0007] The medicinal material identification module is used to obtain the smell, shape and color data of the medicinal materials after they enter the medicinal material forging machine;

[0008] The processing information database includes two sub-databases; the first sub-database pre-stores information entries of medicinal materials for matching the processing time and processing temperature of the medicinal materials; the second sub-database pre-stores process entries of the medicinal materials for matching the initial power of the electric heating machine;

[0009] The real-time monitoring module includes three sub-monitoring modules; the first sub-monitoring module is used to collect quality data during the medicinal material processing, the second sub-monitoring module is used to collect temperature data at different positions during the medicinal material processing, and the third sub-monitoring module is used to collect the real-time power of the electric heater;

[0010] The model prediction module uses the entropy weight method-based model prediction control method of the calcining machine based on the medicinal material processing information, the real-time quality data, temperature data and the real-time power of the electric heater to predict and output the power of the electric heater in the calcining machine at the next moment;

[0011] The temperature control module is used to control the heating module of the electric heating machine to perform heating work with the power output at the next moment.

[0012] Furthermore, the medicinal material identification module includes an electronic nose for identifying the smell of medicinal materials and a high-definition camera for identifying the shape and color of medicinal materials; when the medicinal materials enter the calcining machine, the odor sampling operator and gas sensor array in the electronic nose identify the odor intensity and odor category of the medicinal materials, convert them into digital signals of the odor, and transmit them to the processing information database together with the shape and color obtained and identified by the high-definition camera.

[0013] Furthermore, each information entry in the first sub-database includes the type of medicinal material, the name of the medicinal material, the shape, color, smell, processing time and processing temperature; using the shape, smell and color collected by the medicinal material recognition module as search conditions, the first sub-database matches the stored information entries with the search conditions to find the corresponding medicinal material type and medicinal material name, and obtains the medicinal material processing time and processing temperature from the corresponding information entry;

[0014] Each process entry in the second sub-database includes the type of medicinal material and the initial power of the electric heater matching the type of medicinal material; the second sub-database selects the initial power of the electric heater in the corresponding process entry according to the type of medicinal material matched by the first sub-database, and transmits the initial power of the electric heater, processing time, and processing temperature as medicinal material processing information to the real-time monitoring module.

[0015] Furthermore, the first sub-monitoring module includes multiple quality sensors for detecting the real-time quality of medicinal materials placed on the heating medium; the second sub-monitoring module includes multiple temperature sensors for monitoring the real-time temperature of various positions of the calcining machine during the medicinal material processing, including the initial temperature of the air in the calcining machine, the real-time temperature of the heating medium, the real-time temperature of the medicinal materials, the real-time temperature of the air in the calcining machine, and the real-time temperature of the casing; the third sub-monitoring module is used to collect the real-time power of the electric heater, and transmit these quality data, temperature data, real-time power of the electric heater and medicinal material processing information to the model prediction module.

[0016] A model predictive control method for a calcining machine based on an entropy weight method, comprising:

[0017] A nonlinear model is constructed based on the quality data, temperature data and real-time power of the electric heating machine during the medicinal material processing; the nonlinear model is converted, the system state space equation is constructed, and the coefficient matrix of the system state space equation is determined;

[0018] The backstepping method is used to linearize the established system state space equation, and a state observer is designed to observe the first-order derivative of the real-time temperature of the medicinal materials.

[0019] The system state space equation obtained by linearization is discretized using a zero-order holder to construct a discretized system.

[0020] For the discretized system, a rolling optimization method is used to predict the system state for a period of time in the future;

[0021] A quadratic control law is designed, and the entropy weight method is used to obtain the weight coefficients of the system state and the weight coefficients of the input variables in the quadratic control law; the predicted system state for a period of time in the future is brought in, and constraints are set to solve the system input variable sequence, and the first input variable in the solved sequence is applied to the forging machine as the power of the electric heater at the next moment.

[0022] Furthermore, the nonlinear model is as follows:

[0023] ;

[0024] ;

[0025] in, is the real-time power of the electric heater, Working hours of electric heating machine; is the initial temperature of the air in the calciner; is the mass of the heating medium, is the specific heat capacity of the heating medium, The real-time temperature of the heating medium; To ensure the real-time quality of medicinal materials, is the specific heat capacity of the medicinal material, The real-time temperature of the medicinal materials; It is the real-time temperature of the air in the calcining machine; is the real-time temperature of the forging machine casing; , represents its first and second order derivatives; , , , All are composite functions;

[0026] Set the state variable ,in , the superscript T indicates transposition; the input variable u=p, the output variable y is the real-time temperature T2 of the medicinal material, and the system state space equation is established by the above nonlinear model:

[0027] ;

[0028] ;

[0029] in, is the first-order derivative of the system state variable, u is the input variable, y is the output variable; A is the system matrix, B is the control matrix, C is the output matrix or observation matrix, and D is the direct transfer matrix;

[0030] ;

[0031] .

[0032] Furthermore, the established system state space equation is linearized by backstepping method, and a state observer is designed to observe the value of the first-order derivative of the real-time temperature of the medicinal material, including:

[0033] Using the backstepping method we can get:

[0034] ;

[0035] Among them, T m It is the target temperature when processing medicinal materials, i.e., the processing temperature; is an adjustable parameter, , is the first and second order derivatives of the real-time temperature T2 of the medicinal material, then:

[0036] ;

[0037] Therefore, the linearized system state space equation can be expressed as:

[0038] ;

[0039] ;

[0040] Design a state observer to observe , let the state variable of the state observer be =[ , ],in ; The first-order derivative of is =[ , ], the result of the state observer is:

[0041] ;

[0042] ;

[0043] in, , is the state variable of the state observer , The estimated value of , is the estimated value of the first-order derivative of the state variable in the state observer, is the observer gain matrix;

[0044] Let the first-order derivative of the state variable x of the system state space equation be The error e between the first-order derivative of the state variable of the state observer tends to 0, and the state variable x of the system state space equation is solved.

[0045] Furthermore, the system state space equation obtained by linearization processing is discretized by using the zero-order holder to construct a discretized system, which is expressed as:

[0046] ;

[0047] Among them, k is the current time, is the value of the state variable x at time k, Enter the value of variable u at time k, represents the system state at time k+1 predicted at time k, , is the discrete system coefficient matrix, we have:

[0048] ;

[0049] ;

[0050] Among them, I is the identity matrix, A is the system matrix, B is the control matrix, and T is the sampling period.

[0051] Furthermore, the design of the quadratic control law includes:

[0052] The general form of designing the quadratic control law J is:

[0053] ;

[0054] Among them, Q is The weight coefficient of the system state at the moment, R is the weight coefficient of the input variable, and F is The weight coefficient of the system state at the moment; is the system state at time k+i predicted at time k, is the input variable at the k+ith moment predicted at the kth moment, i=1,2,…,N-1;

[0055] The constraints of the quadratic control law J are:

[0056] .

[0057] Furthermore, the entropy weight method is used to obtain the weight coefficients of the system state and the weight coefficients of the input variables in the quadratic control law, including:

[0058] Obtain the historical data of the currently processed medicinal materials, and extract the real-time temperature T2 and the derivative of the real-time temperature of the medicinal materials from time k+1 to k+N in the same prediction interval length and power p, and each data value of each historical data is taken as a sample;

[0059] Establish positive indicators for each sample :

[0060] ;

[0061] in, For sample a ij Positive indicator, min and max are for minimum and maximum value operations, a ij Represents the data value of the j-th historical data at the k+i-th moment;

[0062] Then, the positive index of each sample is calculated in turn The proportion of positive indicators at all times , the entropy value of the positive indicator corresponding to each type of historical data :

[0063] ;

[0064] ;

[0065] in, ;

[0066] Finally, we get the weight of each positive indicator of historical data. :

[0067] ;

[0068] in, ; The weight coefficient Q of the system state at a certain moment is a positive definite matrix, and each element on its diagonal is given by The weight coefficient R is a unit matrix, and its diagonal elements are ;

[0069] but The weight coefficient F of the system state at a moment is:

[0070] .

[0071] Compared with the prior art, the present invention has the following technical features:

[0072] 1. The present invention integrates modules such as medicinal material identification, real-time monitoring, and model prediction, maps the data of the power control and temperature control systems, and realizes the coupled control of power and temperature by constructing a nonlinear model; the model predictive control method is adopted to improve the robustness and stability of the temperature control system, and realize dynamic and precise control of the temperature.

[0073] 2. The present invention uses large model technology to compare the collected medicinal material property data with the historical data in the database, and uses electronic noses and high-definition cameras to replace manual identification and selection, thereby reducing labor costs, reducing working time, and realizing the automation of medicinal material identification.

[0074] 3. The present invention dynamically updates the weight coefficient of the linear quadratic control law, and applies entropy weight to obtain the weight coefficient of the system state and the system input; the real-time data in the database is processed in the same way, and the weight coefficient is continuously updated to make the prediction result closer to the expected trajectory, thereby achieving more accurate and rapid temperature control. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 is a structural diagram of the system of the present invention;

[0076] Figure 2 A schematic diagram of a model building and data analysis process in one embodiment of the method of the present invention;

[0077] Figure 3 The figure is a schematic diagram of the data processing flow in one embodiment of the method of the present invention. DETAILED DESCRIPTION

[0078] See attached Figure 1 The present invention provides a model predictive control system for a calcining machine based on an entropy weight method, comprising:

[0079] The medicinal material identification module includes an electronic nose for identifying the smell of medicinal materials and a high-definition camera for identifying the shape and color of medicinal materials. When the medicinal materials enter the calcining machine, the odor sampling operator and gas sensor array in the electronic nose identify the odor intensity and odor category of the medicinal materials, convert them into digital signals of the odor, and transmit them to the processing information database together with the shape and color obtained and identified by the high-definition camera.

[0080] The processing information database includes two sub-databases; the first sub-database pre-stores information entries of medicinal materials, each of which includes: medicinal material type, medicinal material name, shape, color, smell, processing time and processing temperature, etc.; using the shape, smell and color collected by the medicinal material recognition module as search conditions, the first sub-database matches the stored information entries with the search conditions to find the corresponding medicinal material type and medicinal material name, and obtains the medicinal material processing time t from the corresponding information entry g And processing temperature T m .

[0081] The second sub-database pre-stores the process entries of medicinal materials, each of which includes: the type of medicinal material and the initial power of the electric heating machine matching the type of medicinal material The second sub-database selects the initial power of the electric heater in the corresponding process entry according to the type of medicinal materials matched by the first sub-database , and the initial power of the electric heater , processing time t g , Processing temperature T m Transmitted to the real-time monitoring module as medicinal material processing information.

[0082] The real-time monitoring module includes three sub-monitoring modules, which are used to collect quality data, temperature data and real-time power of the electric heater during the medicinal material processing; the first sub-monitoring module includes multiple quality sensors, which are used to detect the real-time quality of the medicinal materials placed on the heating medium. ; The second sub-monitoring module includes multiple temperature sensors, which are installed in multiple positions such as the heating medium and the casing of the calcining machine, and are used to monitor the real-time temperature of various positions of the calcining machine during the medicinal material processing. The real-time temperatures that need to be monitored include the initial temperature T0 of the air in the calcining machine, the real-time temperature T1 of the heating medium, the real-time temperature T2 of the medicinal material, the real-time temperature T3 of the air in the calcining machine, the real-time temperature T4 of the casing, etc.; The third sub-monitoring module is used to collect the real-time power p of the electric heating machine;

[0083] These quality data, temperature data, real-time power of the electric heater and medicinal material processing information are transmitted to the model prediction module.

[0084] The model prediction module uses the entropy weight method-based model prediction control method of the calcining machine based on the medicinal material processing information, real-time quality data, temperature data, and real-time power of the electric heater to predict and output the power of the electric heater in the calcining machine at the next moment;

[0085] The temperature control module is used to control the heating module of the electric heating machine to perform heating work with the power output at the next moment.

[0086] In one embodiment of the present invention, taking keel processing as an example, an electronic nose is installed on the outside of the feed port of the calcining machine to identify the smell of the keel; a high-definition camera is arranged on the base of the electronic nose to identify the shape and color of the keel; a quality sensor is installed at the bottom of the crucible of the calcining machine to monitor changes in the quality of medicinal materials; temperature sensors are evenly and equidistantly arranged at the bottom of the crucible, the top of the crucible, the inside of the casing and the inside of the calcining machine, and two temperature sensors are arranged at each monitoring point to monitor temperature changes at different positions.

[0087] The model prediction control method of the calcining machine based on the entropy weight method of the model prediction module in this scheme has the following specific steps:

[0088] Step 1: construct a nonlinear model based on the quality data, temperature data and real-time power of the electric heater in the medicinal material processing process; transform the nonlinear model, construct the system state space equation, and determine the coefficient matrix of the system state space equation.

[0089] This step uses the heating and heat conduction mechanism to identify the system, build a nonlinear model, and transform the nonlinear model to obtain the state space equation. First, the system needs to be identified. Since the model of the calcining machine has less heat dissipation and a simple structure, the nonlinear model is established as follows:

[0090] ;

[0091] ;

[0092] in, is the real-time power of the electric heater, Working hours of electric heating machine; is the initial temperature of the air in the calciner; is the mass of the heating medium (in this solution, a heating plate is used for heating, so it is a constant value), is the specific heat capacity of the heating medium, The real-time temperature of the heating medium; To ensure the real-time quality of medicinal materials, is the specific heat capacity of the medicinal material, The real-time temperature of the medicinal materials; It is the real-time temperature of the air in the calcining machine; It is the real-time temperature of the forging machine casing.

[0093] The parameters in this scheme are superscript , Represents its first and second order derivatives, for example is the first-order derivative of the real-time temperature of the heating medium, is the first-order derivative of the real-time temperature of the medicinal material, The first derivative of the real-time temperature of the air inside the calciner, is the first-order derivative of the real-time temperature of the forging machine casing; , , , All are composite functions.

[0094] Set the state variable ,in ; Input variable u=p, output variable y is the real-time temperature T2 of the medicinal material, and the system state space equation is established by the above nonlinear model:

[0095] ;

[0096] ;

[0097] in, is the first-order derivative of the system state variable, u is the input variable, and y is the output variable; the superscript T indicates transposition.

[0098] according to , , , The coefficients of and the general form of the state-space equations determine the coefficient matrix of the system state-space equations:

[0099] Among them, A is the system matrix, B is the control matrix, C is the output matrix or observation matrix, and D is the direct transfer matrix:

[0100] ;

[0101] ;

[0102] in, for The second-order derivative of for The second derivative of .

[0103] Step 2: For the established system state space equation, the backstepping method is used to linearize it, and a state observer is designed to observe the value of the first-order derivative of the real-time temperature of the medicinal material.

[0104] To make the state space equation of the system asymptotically stable, it is necessary to obtain the Lyapunov function, which can be obtained by backstepping:

[0105] ;

[0106] Among them, T m It is the target temperature when processing medicinal materials, i.e., the processing temperature; is an adjustable parameter, , is the first and second order derivatives of the real-time temperature T2 of the medicinal material, then:

[0107] ;

[0108] Therefore, the linearized system state space equation can be expressed as:

[0109] ;

[0110] ;

[0111] By designing the above backstepping method, the state space equation is linearized.

[0112] In the above formula, The specific values ​​and performance cannot be directly observed, which makes the above state space equation unable to calculate the result. It is necessary to design a state observer to observe .

[0113] Assume that the state variable of the state observer is =[ , ],in ; The output variable is ; above is a disturbance term that will cancel out the input or can be considered as the initial state, then there is an initial state space equation applied to the state observer:

[0114] ;

[0115] ;

[0116] remember The first-order derivative of is =[ , ].

[0117] Through the design and calculation of the above state observer, the final observer result can be obtained as follows:

[0118] ;

[0119] ;

[0120] in, , is the state variable of the state observer , The estimated value of , is the estimated value of the first-order derivative of the state variable in the state observer, is the observer gain matrix.

[0121] Let the first-order derivative of the state variable x of the system state space equation be The error e between the first-order derivative of the state variable of the state observer tends to 0, and the state variable x of the system state space equation is solved to obtain , used in subsequent calculations:

[0122] .

[0123] Step 3: Use a zero-order holder to discretize the system state space equation obtained by linearization to construct a discretized system.

[0124] Since the system state space equation obtained in step 2 is continuous, in order to simplify the model, improve model stability, reduce the risk of overfitting and improve computational efficiency, the linearized state space equation is discretized using a zero-order holder to obtain a discretized system:

[0125] ;

[0126] Among them, k is the current time, is the value of the state variable x at time k, Enter the value of variable u at time k, represents the system state at time k+1 predicted at time k, , is the discrete system coefficient matrix, we have:

[0127] ;

[0128] ;

[0129] Where I is the identity matrix, is the time constant, is a natural constant, A is the system matrix, B is the control matrix, and T is the sampling period.

[0130] Step 4: For the discretized system, use the rolling optimization method to predict the system state for a period of time in the future.

[0131] The system state at a certain time in the future to The expression is as follows:

[0132] ; ; ;…… ;

[0133] in, is the initial system state at time k, whose value is the state variable x solved in step 2; The system state at time k+N predicted at time k; is the input variable at time k, and its value is the real-time power p of the electric heating machine. is the input variable at time k+N-1 predicted at time k, and N is the length of the prediction interval.

[0134] Step 5, design a quadratic control law, and use the entropy weight method to obtain the weight coefficient of the system state and the weight coefficient of the input variable in the quadratic control law; bring in the predicted system state for a period of time in the future, set constraints, solve the system input variable sequence, and use the first input variable in the solved sequence as the power of the electric heater at the next moment to apply to the forging machine.

[0135] First, the general form of the quadratic control law J is designed:

[0136] ;

[0137] Among them, Q is The weight coefficient of the system state at the moment, R is the weight coefficient of the input variable, and F is The weight coefficient of the system state at the moment; is the system state at time k+i predicted by time k obtained in step 4, is the input variable at time k+i predicted at time k, i=1,2,…,N-1, and the superscript T indicates transposition.

[0138] The temperature fluctuation range of the calcining machine is usually designed to be within ±1°C to ensure the stability and accuracy of the calcining process; for the above quadratic control law J, its constraints are:

[0139] ;

[0140] For the above quadratic control law J, The weight coefficient Q of the system state at the moment, The weight coefficient F of the system state at the moment and the weight coefficient R of the input variable are obtained by the entropy weight method, as follows:

[0141] Obtain the historical data of the currently processed medicinal materials, and extract the real-time temperature T2 and the derivative of the real-time temperature of the medicinal materials from time k+1 to k+N in the same prediction interval length The three historical data of power p are taken as a sample, and are represented in a tabular form with N rows and 3 columns:

[0142]

[0143] The sample a in the i-th (i=1,2,…,N) row and j-th (j=1,2,3) column ij is the data value of the jth historical data at the k+ith moment; a positive indicator is established for each sample :

[0144] ;

[0145] in, For sample a ij Positive indicator, min and max are operations for finding minimum and maximum values.

[0146] Then, the positive index of each sample is calculated in turn The proportion of positive indicators at all times , the entropy value of the positive indicator corresponding to each type of historical data :

[0147] ;

[0148] ;

[0149] in, ;

[0150] Finally, we get the weight of each positive indicator of historical data. :

[0151] ;

[0152] in, The weight coefficient Q of the system state at a certain moment is a positive definite matrix, and each element on its diagonal is given by The weight coefficient R is a unit matrix, and its diagonal elements are .

[0153] but The weight coefficient F of the system state at a moment is:

[0154] ;

[0155] The first N-1 values ​​of the system state in the future period obtained in step 4, that is, to As , the Nth value As Substitute it into the quadratic control law J, take the derivative of the quadratic control law J, and set the derivative to 0, and the input variable sequence of the system can be obtained: , i=1…N-1.

[0156] The obtained system input variables are the combination matrix of the system input variables u from the kth moment to the k+N-1th moment, and the first input variable is taken The power p of the electric heater at the next moment is applied to the forging machine, and the electric heater is controlled to perform heating operation at this power.

[0157] When the present invention is applied in a specific way, the working process is as follows:

[0158] At the initial time k, the calcining machine first uses the initial power Start working and set the processing time t according to the medicinal material processing information g , Processing temperature T m , start the first cycle, the process of each cycle is:

[0159] The real-time monitoring module obtains the temperature T2 of the medicinal material at this time as , the derivative of T2 for Since it cannot be directly observed The specific value and performance of , so the state observer in step 2 is used to solve and obtain , thus obtaining the state variables of the state space equation , and use the state variable x as the initial state of the system at time k in the rolling optimization method in step 4 , predict and obtain the system status at future times to , then to As Substitute it into the quadratic control law J in step 5, and obtain the system input variable sequence by taking the derivative of J and setting the derivative equal to zero , take its first value The power p of the electric heating machine at the next moment is applied to the forging machine, and the temperature control module controls the heating module of the electric heating machine to perform heating work at this power.

[0160] At the next time k+1, the calcining machine starts working with a new power p, and the second cycle begins; and the above cycle process is repeated until the medicinal material processing is completed or the set processing time t is reached at the same time. g and processing temperature T m .

[0161] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A model predictive control system for a calcining machine based on entropy weight method, characterized in that: include: The medicinal material identification module is used to obtain the smell, shape and color data of the medicinal materials after they enter the medicinal material forging machine; The processing information database includes two sub-databases; the first sub-database pre-stores information entries of medicinal materials for matching the processing time and processing temperature of the medicinal materials; the second sub-database pre-stores process entries of the medicinal materials for matching the initial power of the electric heating machine; The real-time monitoring module includes three sub-monitoring modules; the first sub-monitoring module is used to collect quality data during the medicinal material processing, the second sub-monitoring module is used to collect temperature data at different positions during the medicinal material processing, and the third sub-monitoring module is used to collect the real-time power of the electric heater; The model prediction module uses the entropy weight method-based model prediction control method of the calcining machine based on the medicinal material processing information, the real-time quality data, temperature data and the real-time power of the electric heater to predict and output the power of the electric heater in the calcining machine at the next moment; The temperature control module is used to control the heating module of the electric heating machine to perform heating work with the power output at the next moment.

2. The model predictive control system of a calcining machine based on the entropy weight method according to claim 1 is characterized in that: The medicinal material identification module includes an electronic nose for identifying the smell of medicinal materials and a high-definition camera for identifying the shape and color of medicinal materials; when the medicinal materials enter the calcining machine, the odor sampling operator and gas sensor array in the electronic nose identify the odor intensity and odor category of the medicinal materials, convert them into digital signals of the odor, and transmit them to the processing information database together with the shape and color obtained and identified by the high-definition camera.

3. The model predictive control system of a calcining machine based on the entropy weight method according to claim 2 is characterized in that: Each information entry in the first sub-database includes the type of medicinal material, the name of the medicinal material, shape, color, smell, processing time and processing temperature; using the shape, smell and color collected by the medicinal material recognition module as search conditions, the first sub-database matches the stored information entries with the search conditions to find the corresponding medicinal material type and medicinal material name, and obtains the medicinal material processing time and processing temperature from the corresponding information entry; Each process entry in the second sub-database includes the type of medicinal material and the initial power of the electric heater matching the type of medicinal material; the second sub-database selects the initial power of the electric heater in the corresponding process entry according to the type of medicinal material matched by the first sub-database, and transmits the initial power of the electric heater, processing time, and processing temperature as medicinal material processing information to the real-time monitoring module.

4. The model predictive control system of a calcining machine based on the entropy weight method according to claim 3 is characterized in that: The first sub-monitoring module includes multiple quality sensors for detecting the real-time quality of medicinal materials placed on the heating medium; the second sub-monitoring module includes multiple temperature sensors for monitoring the real-time temperature of various positions of the calcining machine during the medicinal material processing, including the initial temperature of the air in the calcining machine, the real-time temperature of the heating medium, the real-time temperature of the medicinal materials, the real-time temperature of the air in the calcining machine, and the real-time temperature of the casing; the third sub-monitoring module is used to collect the real-time power of the electric heater, and transmit these quality data, temperature data, real-time power of the electric heater and medicinal material processing information to the model prediction module.

5. A model predictive control method for a calcining machine based on entropy weight method, characterized in that: include: A nonlinear model is constructed based on the quality data, temperature data and real-time power of the electric heating machine during the medicinal material processing; Transform the nonlinear model, construct the system state space equation, and determine the coefficient matrix of the system state space equation; The backstepping method is used to linearize the established system state space equation, and a state observer is designed to observe the first-order derivative of the real-time temperature of the medicinal materials. The system state space equation obtained by linearization is discretized using a zero-order holder to construct a discretized system. For the discretized system, a rolling optimization method is used to predict the system state for a period of time in the future; A quadratic control law is designed, and the entropy weight method is used to obtain the weight coefficients of the system state and the weight coefficients of the input variables in the quadratic control law; the predicted system state for a period of time in the future is brought in, and constraints are set to solve the system input variable sequence, and the first input variable in the solved sequence is applied to the forging machine as the power of the electric heater at the next moment.

6. The model predictive control method for a calcining machine based on the entropy weight method according to claim 5 is characterized in that: The nonlinear model is as follows: ； ； in, is the real-time power of the electric heater, Working hours for electric heating machine; is the initial temperature of the air in the calciner; is the mass of the heating medium, is the specific heat capacity of the heating medium, The real-time temperature of the heating medium; To ensure the real-time quality of medicinal materials, is the specific heat capacity of the medicinal material, The real-time temperature of the medicinal materials; It is the real-time temperature of the air in the calcining machine; is the real-time temperature of the forging machine casing; , represents its first and second order derivatives; , , , All are composite functions; Set the state variable ,in , the superscript T indicates transposition; the input variable u=p, the output variable y is the real-time temperature T2 of the medicinal material, and the system state space equation is established by the above nonlinear model: ； ； in, is the first-order derivative of the system state variable, u is the input variable, y is the output variable; A is the system matrix, B is the control matrix, C is the output matrix or observation matrix, and D is the direct transfer matrix; ； 。 7. The model predictive control method for a calcining machine based on the entropy weight method according to claim 6 is characterized in that: The established system state space equation is linearized by backstepping method, and a state observer is designed to observe the value of the first-order derivative of the real-time temperature of the medicinal material, including: Using the backstepping method we can get: ； Among them, T m It is the target temperature when processing medicinal materials, i.e. the processing temperature; is an adjustable parameter, , is the first and second order derivatives of the real-time temperature T2 of the medicinal material, then: ； Therefore, the linearized system state space equation can be expressed as: ； ； Design a state observer to observe , let the state variable of the state observer be =[ , ],in ; The first-order derivative of is =[ , ], the result of the state observer is: ； ； in, , is the state variable of the state observer , The estimated value of , is the estimated value of the first-order derivative of the state variable in the state observer, is the observer gain matrix; Let the first-order derivative of the state variable x of the system state space equation be The error e between the first-order derivative of the state variable of the state observer tends to 0, and the state variable x of the system state space equation is solved.

8. The model predictive control method for a calcining machine based on the entropy weight method according to claim 6 is characterized in that: The system state space equation obtained by linearization processing is discretized by using a zero-order holder to construct a discretized system, which is expressed as: ； Among them, k is the current time, is the value of the state variable x at time k, Enter the value of variable u at time k, represents the system state at time k+1 predicted at time k, , is the discrete system coefficient matrix, we have: ； ； Among them, I is the identity matrix, A is the system matrix, B is the control matrix, and T is the sampling period.

9. The model predictive control method for a calcining machine based on the entropy weight method according to claim 8, characterized in that: The design of the quadratic control law comprises: The general form of designing the quadratic control law J is: ； Among them, Q is The weight coefficient of the system state at the moment, R is the weight coefficient of the input variable, and F is The weight coefficient of the system state at the moment; is the system state at time k+i predicted at time k, is the input variable at the k+ith moment predicted at the kth moment, i=1,2,…,N-1; The constraints of the quadratic control law J are: 。 10. The model predictive control method for a calcining machine based on the entropy weight method according to claim 9, characterized in that: The entropy weight method is used to obtain the weight coefficients of the system state and the weight coefficients of the input variables in the quadratic control law, including: Obtain the historical data of the currently processed medicinal materials, and extract the real-time temperature T2 and the derivative of the real-time temperature of the medicinal materials from time k+1 to k+N in the same prediction interval length and power p, and each data value of each historical data is taken as a sample; Establish positive indicators for each sample : ； in, For sample a ij Positive indicator, min and max are for minimum and maximum value operations, a ij Represents the data value of the j-th historical data at the k+i-th moment; Then, the positive index of each sample is calculated in turn The proportion of positive indicators at all times , the entropy value of the positive indicator corresponding to each type of historical data : ； ； in, ; Finally, we get the weight of each positive indicator of historical data. : ； in, ; The weight coefficient Q of the system state at a certain moment is a positive definite matrix, and each element on its diagonal is given by The weight coefficient R is a unit matrix, and its diagonal elements are ; but The weight coefficient F of the system state at a moment is: 。

Images