A semiconductor temperature PID control method based on ARM

Abstract

The invention discloses a semiconductor temperature PID control method based on ARM, and relates to the field of digital control technology. By collecting characteristic parameter data of gas in a pipeline, a neural network model with characteristic parameter variation as input and heater output power variation as output is constructed, a functional relationship between characteristic parameter variation and heater output power variation is output, a prediction model is constructed to output a predicted value of characteristic parameter variation, the predicted value of heater output power variation is solved according to the functional relationship, and the obtained predicted value is used as a correction value of heater output power, the temperature difference and temperature error change rate of the heater are obtained as input parameters in a fuzzy controller, and after being processed by the control rule of the fuzzy controller, a PID control parameter for directly controlling the duty cycle of a PWM module is output, and then the control value of the heater output power is obtained, and the actual value of the heater output power is calculated according to the obtained control value and correction value.

Description

Technical Field

[0001] The invention relates to the technical field of digital control, and in particular to a semiconductor temperature PID control method based on ARM. Background Art

[0002] Temperature controller refers to a series of automatic control components that produce certain special effects and conduction or disconnection actions by physically deforming the switch according to the temperature changes in the working environment, or provide temperature data to the circuit according to the different working principles of electronic components at different temperatures, so that the circuit can collect temperature data. It includes analog signal temperature controllers, integrated temperature controllers, intelligent digital temperature controllers, etc., and is widely used in various industries.

[0003] In the semiconductor production process, the gas raw materials need to be stably transported to the reaction chamber so that they can react chemically at the set process temperature. Therefore, the gas pipeline needs to be actively heated to keep the gas temperature in the pipeline stable and within the set ideal state range. However, for some gases with low boiling points under normal pressure, condensate will adhere to the inner wall of the pipeline during long-distance pipeline transportation, causing it to be blocked, making the gas flow in the pipeline worse and the volatility increased. The temperature control of the semiconductor pipeline heating process in the prior art does not take into account the impact of the change in gas flow in the pipeline, resulting in temperature overshoot and large oscillation amplitude during the control process, which in turn affects the chemical reaction in the reaction chamber and reduces the production efficiency of semiconductors. To this end, we propose a semiconductor temperature PID control method based on ARM. Summary of the invention

[0004] The main purpose of the present invention is to provide a semiconductor temperature PID control method based on ARM, which can effectively solve the problems in the background technology.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A semiconductor temperature PID control method based on ARM, comprising:

[0007] Collect characteristic parameter data of gas in pipeline and obtain the change of characteristic parameter Δλ of item i at adjacent sampling nodes i and the change ΔP of the corresponding heater output power, wherein the characteristic parameters of the gas in the pipeline include at least two of the gas pressure value, the gas density value, the gas flow value per unit time, the average flow velocity value, and the initial temperature value of the gas entering the pipeline;

[0008] To obtain the characteristic parameter change Δλ i As input, the change of heater output power ΔP is used as output to build a neural network model. According to the network model, the change of characteristic parameter Δλ is output.i The functional relationship between the change in the heater output power ΔP is expressed as: f:(Δλ 1 ,Δλ 2 ,...,Δλ i )→ΔP, wherein the neural network model is a BP neural network model with a three-layer structure of input layer-hidden layer-output layer, and the number of neurons in the input layer is equal to the number of characteristic parameter types of the gas in the pipeline, the number of neurons in the output layer is 1, and the number of neurons in the hidden layer is determined according to the following formula, specifically:

[0009]

[0010] In the formula, s represents the number of neurons in the hidden layer. is to round up the value in the brackets; p is the number of neurons in the input layer; q is the number of neurons in the output layer, q=1; α is a constant coefficient of 1-9;

[0011] According to the acquired characteristic parameter variation Δλ i The prediction model is constructed based on the time series data, and the change of the characteristic parameter Δλ is output by the model. i The predicted value of, wherein the expression of the prediction model is:

[0012] Δλ' i(t+1) =aΔλ it +a(1-a)Δλ i(t-1) +a(1-a) 2 Δλ i(t-2) +...+a(1-a) t-1 Δλ i1 +(1-a) t Δλ' i1

[0013] In the formula, Δλ' i(t+1) It is expressed as the predicted value of the change of the i-th characteristic parameter at the t+1th node time; Δλ it It is expressed as the sampling value of the change of the i-th characteristic parameter at the t-th node time, a is a constant coefficient, and a∈(0,1);

[0014] Setting the target temperature T of the pipeline heater during semiconductor processing 0 , collect the real-time temperature of the heater, calculate the temperature difference and temperature error change rate of the heater as the input parameters of the fuzzy controller, and after being processed by the control rules in the fuzzy controller, the output PID control parameters directly control the duty cycle of the PWM module, and then obtain the control value P of the heater output power out , wherein the input parameters and output parameters of the fuzzy controller satisfy the following relationship:

[0015]

[0016] In the formula, K p , K i , K d is the fuzzy PID parameter; K p ', K i ', K d ' is the PID parameter after fuzzy correction; ΔK p , ΔK i , ΔK d are the fuzzy PID parameters to be determined; in the fuzzy controller, e, ec, ΔK p , ΔK i , ΔK d The value ranges are [-3 3], [-3 3], [-0.3 0.3], [-3 3], [-0.3 0.3], where e represents the temperature difference of the heater; ec represents the temperature error change rate of the heater

[0017] Substitute the predicted value of the characteristic parameter change into the functional relationship to obtain the predicted value ΔP of the heater output power change at the t+1th node time. t+1 , and the obtained predicted value is used as the correction value of the heater output power, and the actual value of the heater output power is calculated according to the formula:

[0018]

[0019] Where P' out Expressed as the actual value of the heater output power.

[0020] The present invention has the following beneficial effects:

[0021] Compared with the prior art, the technical solution of the present invention obtains the change Δλ of the i-th characteristic parameter at adjacent sampling nodes by collecting the characteristic parameter data of the gas in the pipeline. i And the corresponding change in heater output power ΔP, to obtain the characteristic parameter change Δλ i As input, the change of heater output power ΔP is used as output to build a neural network model. According to the network model, the change of characteristic parameter Δλ is output. i The functional relationship between the change in heater output power ΔP and the change in characteristic parameter Δλ i The prediction model is constructed based on the time series data, and the change of the characteristic parameter Δλ is output by the model. i The predicted value is used to set the target temperature value T of the pipeline heater during semiconductor processing. 0, collect the real-time temperature of the heater, calculate the temperature difference and temperature error change rate of the heater as the input parameters of the fuzzy controller, and after being processed by the control rules in the fuzzy controller, the output PID control parameters directly control the duty cycle of the PWM module, and then obtain the control value P of the heater output power out , the predicted value of the characteristic parameter change is substituted into the functional relationship to obtain the predicted value ΔP of the heater output power change at the t+1th node time t+1 , and use the obtained predicted value as the correction value of the heater output power. The actual value of the heater output power is calculated according to the formula. By analyzing the impact of changes in gas fluidity in the pipeline on the temperature control of the heating process, the heating power output value is further corrected, thereby reducing the impact of temperature control caused by gas flow in the pipeline.

[0022] Compared with the prior art, the technical solution of the present invention dynamically adjusts the PWM output according to the real-time temperature data fed back during operation, so that the heating power of the heater is adjustable. The heating temperature is controlled by changing the heating power, so that the natural heat dissipation and heating of the pipeline reach a dynamic balance, and the accuracy of the heating process of the pipeline and the stability control of the insulation process are achieved;

[0023] Compared with the prior art, the technical solution of the present invention adopts a fuzzy PID control method, which can control the current heating process of the heater, take the temperature difference and the temperature error change rate of the heater as input, and convert the fuzzy processed PID controller parameters into the duty cycle value required for system control, thereby realizing precise heating control of the flowing gas in the pipeline. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of a semiconductor temperature PID control method based on ARM of the present invention;

[0025] Figure 2 This is the structural block diagram of the control process. DETAILED DESCRIPTION

[0026] The present invention will be further described below in conjunction with specific implementation methods, wherein the accompanying drawings are only used for exemplary descriptions and represent only schematic diagrams rather than actual drawings, and should not be understood as limiting the present invention. In order to better illustrate the specific implementation methods of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product.

[0027] Example 1

[0028] like Figure 1-2 As shown, the specific implementation process of the technical solution of the present invention includes the following steps:

[0029] Step 1: Collect characteristic parameter data of the gas in the pipeline and obtain the change Δλ of the i-th characteristic parameter at adjacent sampling nodes i and the corresponding change in heater output power ΔP;

[0030] It should be noted that the change in the i-th characteristic parameter Δλ i The calculation formula for the change in the heater output power ΔP is as follows:

[0031]

[0032] In the formula, Δλ it , ΔP t are respectively represented as the characteristic parameter value and the output power value at the sampling time t, wherein the characteristic parameters of the gas in the pipeline include at least two of the gas pressure value, the gas density value, the gas flow value per unit time, the average flow velocity value, and the initial temperature value of the gas entering the pipeline;

[0033] Step 2: To obtain the characteristic parameter change Δλ i As input, the change of heater output power ΔP is used as output to build a neural network model. According to the network model, the change of characteristic parameter Δλ is output. i The functional relationship between the change in the heater output power ΔP is expressed as: f:(Δλ 1 ,Δλ 2 ,...,Δλ i )→ΔP;

[0034] Among them, the neural network model is a BP neural network model with a three-layer structure of input layer-hidden layer-output layer, and the number of neurons in the input layer is equal to the number of characteristic parameter types of the gas in the pipeline, the number of neurons in the output layer is 1, and the number of neurons in the hidden layer is determined according to the following formula, specifically:

[0035]

[0036] In the formula, s represents the number of neurons in the hidden layer. is to round up the value in the brackets; p is the number of neurons in the input layer; q is the number of neurons in the output layer, q=1; α is a constant coefficient of 1-9. In this embodiment, p=5 and α=3 are taken, then In this embodiment, the constructed BP neural network model structure is an input layer with 5 neurons - a hidden layer with 6 neurons - an output layer with 1 neuron;

[0037] Among them, the steps of building the neural network model include:

[0038] Get the characteristic parameter change Δλ iThe data of the change in heater output power ΔP are normalized, and both the input variables and the output variables are normalized to the interval [0, 1]. The normalization formula is:

[0039]

[0040] In the formula, R represents the normalized data value, x represents the original data, and X max , X min Represented as the maximum and minimum values ​​in the original data respectively;

[0041] Since the learning of the neural network is mainly achieved through repeated training of samples, in this embodiment, the BP neural network toolbox of MATLAB is used to train the network using the standard BP neural network algorithm, the target error is set to 0.007, the maximum number of cycles is 1000, the transfer function of the hidden layer neurons is selected as the S-type tangent function tansig, the transfer function of the output layer neurons is selected as the S-type logarithmic function logsig, and the traingdx function is selected as the training function;

[0042] After the training is completed, the prediction accuracy of the model is verified using the validation set data. When the prediction accuracy is less than the set value, the model is built. In this embodiment, the prediction accuracy is set to be controlled within 5%;

[0043] After the neural network is constructed, the output results are subjected to regression analysis based on the predicted data output by the neural network, using the toolbox function corrcoef to obtain the linear regression analysis results, and the characteristic parameter change Δλ can be obtained. i The functional relationship between the change in heater output power ΔP.

[0044] Step 3: According to the acquired characteristic parameter change Δλ i The prediction model is constructed based on the time series data, and the change of the characteristic parameter Δλ is output by the model. i The predicted value of , where the expression of the prediction model is:

[0045] Δλ' i(t+1) =aΔλ it +a(1-a)Δλ i(t-1) +a(1-a) 2 Δλ i(t-2) +...+a(1-a) t-1 Δλ i1 +(1-a) t Δλ' i1

[0046] In the formula, Δλ' i(t+1) It is expressed as the predicted value of the change of the i-th characteristic parameter at the t+1th node time; Δλit It is expressed as the sampling value of the change of the i-th characteristic parameter at the t-th node time, a is a constant coefficient, and a∈(0,1);

[0047] It should be noted that: for the value of the constant coefficient a, the following value selection principles can be referred to, specifically:

[0048] a: During the implementation process, a fluctuation threshold can be set. If the data series does not fluctuate much and is relatively stable, that is, the data fluctuation is within the fluctuation threshold, the coefficient a should be smaller, such as a = 0.3, to reduce the correction range and enable the prediction model to include information from a longer time series;

[0049] b: If the time series has a rapid and obvious tendency to change, such as when the fluctuation between adjacent sampling nodes is greater than the set fluctuation threshold, the coefficient a should be adaptively taken to a value close to 1;

[0050] c: In practice, the test method can be used to determine the model. Several data values ​​are selected for trial calculation, and then the coefficient a that makes the prediction error smaller is taken as the parameter of the model.

[0051] Step 4: Set the target temperature T of the pipeline heater during semiconductor processing 0 , collect the real-time temperature of the heater, calculate the temperature difference and temperature error change rate of the heater as the input parameters of the fuzzy controller, and after being processed by the control rules in the fuzzy controller, the output PID control parameters directly control the duty cycle of the PWM module, and then obtain the control value P of the heater output power out , where the input parameters and output parameters of the fuzzy controller satisfy the following relationship:

[0052]

[0053] In the formula, K p , K i , K d is the fuzzy PID parameter; K p ', K i ', K d ' is the PID parameter after fuzzy correction; ΔK p , ΔK i , ΔK d are the fuzzy PID parameters to be determined; in the fuzzy controller, e, ec, ΔK p , ΔK i , ΔK d The value ranges of are [-3 3], [-3 3], [-0.3 0.3], [-3 3], [-0.3 0.3], where e represents the temperature difference of the heater; ec represents the temperature error change rate of the heater;

[0054] It should be noted that the control rules in the fuzzy controller need to be compiled according to the type of input parameters and the actual situation. In this embodiment, one form of compiling the control rules is given, as shown in the following table:

[0055] Table 1 K p Control rules table

[0056]

[0057] Table 2 K i Control rules table

[0058]

[0059]

[0060] Table 3K d Control rules table

[0061]

[0062] The fuzzy rule bases of the various output parameters established above are integrated to establish control rules based on fuzzy PID control as the theoretical basis for adaptive adjustment of the control system.

[0063] Step 5: Substitute the predicted value of the characteristic parameter change into the functional relationship to obtain the predicted value ΔP of the heater output power change at the t+1th node time t+1 , and the obtained predicted value is used as the correction value of the heater output power, and the actual value of the heater output power is calculated according to the formula:

[0064]

[0065] Where P' out Expressed as the actual value of the heater output power.

[0066] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A semiconductor temperature PID control method based on ARM, characterized in that: include: Collect characteristic parameter data of gas in pipeline and obtain the change of characteristic parameter Δλ of item i at adjacent sampling nodes i and the corresponding change in heater output power ΔP; To obtain the characteristic parameter change Δλ i As input, the change of heater output power ΔP is used as output to build a neural network model. According to the network model, the change of characteristic parameter Δλ is output. i The functional relationship between the change in the heater output power ΔP is expressed as: f:(Δλ1,Δλ2,...,Δλ i )→ΔP; According to the acquired characteristic parameter variation Δλ i The prediction model is constructed based on the time series data, and the change of the characteristic parameter Δλ is output by the model. i The predicted value of The target temperature value T0 of the pipeline heater in the semiconductor processing process is set, the real-time temperature of the heater is collected, and the temperature difference and temperature error change rate of the heater are calculated as the input parameters of the fuzzy controller. After being processed by the control rules in the fuzzy controller, the output PID control parameters directly control the duty cycle of the PWM module, and then the control value P of the heater output power is obtained. out ; Substitute the predicted value of the characteristic parameter change into the functional relationship to obtain the predicted value ΔP of the heater output power change at the t+1th node time. t+1 , and the obtained predicted value is used as the correction value of the heater output power, and the actual value of the heater output power is calculated according to the formula: Where P' out It is expressed as the actual value of the heater output power; The expression of the prediction model is: No. i(t+1) =aΔλ it +a(1-a)Dl i(t-1) +a(1-a) 2 Dl i(t-2) +...+a(1-a) t-1 Dl i1 +(1-a) t No. i1 In the formula, Δλ' i(t+1) It is expressed as the predicted value of the change of the i-th characteristic parameter at the t+1th node time; Δλ it It is expressed as the sampling value of the change of the i-th characteristic parameter at the t-th node time. a is a constant coefficient, and a∈(0,1); The characteristic parameters of the gas in the pipeline include at least two of the following: gas pressure value, gas density value, gas flow value per unit time, average flow velocity value, and initial temperature value of the gas entering the pipeline.

2. The semiconductor temperature PID control method based on ARM according to claim 1, characterized in that: The input parameters and output parameters of the fuzzy controller satisfy the following relationship: In the formula, K p , K i , K d is the fuzzy PID parameter; K p ', K i ', K d ' is the PID parameter after fuzzy correction; ΔK p , ΔK i , ΔK d are the fuzzy PID parameters to be determined.

3. The semiconductor temperature PID control method based on ARM according to claim 1, characterized in that: In the fuzzy controller, e, ec, ΔK p , ΔK i , ΔK d The value range of is [-3 3], [-3 3], [-0.3 0.3], [-3 3], [-0.3 0.3], where e represents the temperature difference of the heater; ec represents the temperature error change rate of the heater.

4. The semiconductor temperature PID control method based on ARM according to claim 1, characterized in that: The neural network model is a BP neural network model with a three-layer structure of input layer-hidden layer-output layer, and the number of neurons in the input layer is equal to the number of characteristic parameter types of the gas in the pipeline, the number of neurons in the output layer is 1, and the number of neurons in the hidden layer is determined according to the following formula, specifically: In the formula, s represents the number of neurons in the hidden layer. To round up the value in the brackets; p is the number of neurons in the input layer; q is the number of neurons in the output layer, q=1; α is a constant coefficient of 1-9.

Images