A method of manufacturing and controlling a temperature boundary controller for a power semiconductor chip

Abstract

The application discloses a power semiconductor chip temperature boundary controller manufacturing method and a control method, and specifically comprises the following steps: based on a power semiconductor chip temperature system model with a sustained residence time switching mechanism, a switching linear boundary controller and a switching nonlinear adaptive boundary controller used for eliminating the influence of boundary uncertainty are designed; the switching linear boundary controller and the switching nonlinear adaptive boundary controller are combined into a switching adaptive boundary controller; the switching adaptive boundary controller gain solved is used in the switching adaptive boundary controller; the power semiconductor chip temperature is bounded convergent; and reasonable control of the power semiconductor chip temperature is realized. The application can effectively eliminate the influence of model uncertainty on the system boundary, guarantee the stability of the power semiconductor chip temperature system, and effectively make the state of the power semiconductor chip temperature system converge to a bounded range.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power semiconductor chip, in particular to a manufacturing method and control method of power semiconductor chip temperature boundary controller. BACKGROUND

[0002] Power semiconductor chip has the ability to handle high voltage and large current, and is widely used in high-speed rail, photovoltaic power generation, electronic communication, aerospace and other fields. For such a high-integration power semiconductor chip system, the temperature of the chip is one of the important indicators to measure whether the system can work normally. If the temperature of the power semiconductor chip is not reasonably controlled, it may cause the chip temperature imbalance, and then make the integrated circuit where the chip is located damaged or unable to work normally, which will bring serious economic and safety accidents. Based on this, it is necessary to design a suitable controller for the temperature system of the chip to keep its temperature in a suitable range, while ensuring that the integrated circuit using the chip can work normally. In order to realize the control problem of the above system, domestic and foreign scholars have carried out a lot of related researches, such as sampling control, sliding mode control, boundary control, etc. Among them, boundary control has gradually become a research hotspot because of its less number of actuators, high reliability and low control cost. Power semiconductor chip is easy to overheat and cannot work normally due to its characteristics of being able to withstand high voltage and current. Therefore, the present application considers that the power semiconductor chip works by switching the heat generation module to reduce the long-term effect of each heat generation module and cause the power semiconductor chip to overheat. Therefore, it is necessary to design a suitable switching controller for the switching system for the power semiconductor chip temperature system with switching rules.

[0003] On the other hand, the boundary uncertainty caused by the effect of the external environment on the system boundary is a factor that cannot be ignored, but for the power semiconductor chip temperature system containing boundary uncertainty, the traditional boundary control usually ignores this uncertainty. Therefore, this traditional boundary control design method may not be able to smoothly stabilize the system. It is of great research significance to design a new type of switching adaptive boundary control method to ensure the stability of the system.

[0004] In addition, the convergence level of the chip temperature is also a factor that must be considered. The chip temperature converges to a fixed equilibrium point due to the influence of the external environment on the system boundary, which is not only unrealistic, but also requires more control cost. Therefore, in the actual chip temperature control process, how to establish a more relaxed convergence level is worth studying. SUMMARY

[0005] To solve the above technical problems, the present application provides a manufacturing method and control method of power semiconductor chip temperature boundary controller.

[0006] To achieve the above technical purpose, the technical scheme adopted is as follows: a power semiconductor chip temperature boundary controller manufacturing method, comprising the following steps:

[0007] Step 1, based on the power semiconductor chip temperature system model with the sustained dwell time switching mechanism, a switching linear boundary controller and a switching nonlinear adaptive boundary controller for eliminating the influence of boundary uncertainty are designed, wherein the switching linear boundary controller adopts a negative feedback boundary controller based on the switching controller gain, the switching nonlinear adaptive boundary controller adopts a nonlinear adaptive boundary controller based on the switching adaptive law and the radial basis function, the input of the switching linear boundary controller is the power semiconductor chip output temperature T(x, t), and the input of the switching nonlinear adaptive boundary controller is the temperature change rate T t (x, t) of the power semiconductor chip.

[0008] Step 2, the switching linear boundary controller and the switching nonlinear adaptive boundary controller designed in step 1 are combined into a switching adaptive boundary controller.

[0009] The power semiconductor chip temperature system model disclosed by the application is established based on the conventional power semiconductor chip temperature system model, the sustained dwell time switching operation mechanism and the boundary model uncertainty.

[0010] A power semiconductor chip temperature control method, the switching adaptive boundary controller gain solved is used in the switching adaptive boundary controller, the power semiconductor chip temperature is bounded to converge, and the power semiconductor chip temperature is reasonably controlled.

[0011] The specific method of the switching adaptive boundary controller gain solved by the application is that the sufficient condition of input-to-state exponential stability of the power semiconductor chip temperature system is determined by using the Lyapunov stability theory, the switching signal law of the power semiconductor chip temperature system model is designed by iteration and recursion based on the sufficient condition, the sufficient condition of input-to-state exponential stability is solved by using the LMI toolbox of MATLAB according to the signal law, and the switching adaptive boundary controller gain is obtained.

[0012] The application has the following beneficial effects:

[0013] 1, the switching adaptive boundary controller comprises two parts, one part is the switching linear boundary controller, which can effectively reduce the number of actuators, and the other part is the switching nonlinear adaptive boundary controller, which can effectively eliminate the influence of the system boundary model uncertainty and ensure the stability of the power semiconductor chip temperature system.

[0014] 2、The present application abandons the modeling method of power semiconductor chip temperature system using average dwell time switching mechanism with certain conservatism, and uses the more general sustained dwell time switching mechanism to model the power semiconductor chip temperature system. Further, the average dwell time switching mechanism cannot describe the fast switching phenomenon of the power semiconductor chip temperature system due to the limitation of switching frequency. However, the sustained dwell time switching mechanism can well describe this fast switching phenomenon, so that the modeling of the power semiconductor chip temperature system using the sustained dwell time switching mechanism can overcome the conservatism of the average dwell time switching mechanism modeling.

[0015] 3、The improved switching adaptive boundary controller is used for temperature control, and for a power semiconductor chip temperature system with sustained dwell time switching mechanism and boundary upper model uncertainty, the switching adaptive boundary control method designed by the present application can effectively make the state of the system converge to a bounded range.

[0016] 4、In the prior art, in order to ensure the traditional stability (asymptotic stability or exponential stability), the temperature of the system is required to converge to a fixed equilibrium point, while the sufficient condition of input-to-state exponential stability designed by the present application can ensure the system to converge to a bounded range, which is more in line with the actual situation and reduces the control cost. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a structure diagram of a power semiconductor chip temperature control method;

[0018] Figure 2 is a structure diagram of a power semiconductor chip temperature system;

[0019] Figure 3 is a production flow diagram of a power semiconductor chip temperature switching adaptive controller;

[0020] Figure 4 is a sustained dwell time switching mechanism mode change diagram;

[0021] Figure 5 is an open-loop state simulation diagram of a power semiconductor chip temperature system;

[0022] Figure 6 is a closed-loop state simulation diagram of a power semiconductor chip temperature system under the action of a switching adaptive boundary controller;

[0023] Figure 7 is a power semiconductor chip temperature system switching adaptive boundary control input simulation diagram;

[0024] Figure 8 is a closed-loop state simulation diagram of a power semiconductor chip temperature system under the action of a conventional boundary controller;

[0025] Figure 9 is a simulation diagram of a conventional boundary control input of a power semiconductor chip temperature system. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0027] Power semiconductor chips are prone to overheat due to bearing high voltage and current, which makes the power semiconductor chips unable to work normally. Therefore, the present application considers that the heat generation modules of the power semiconductor chips work alternately to reduce the long-term effect of each heat generation module and make the power semiconductor chips overheat. For this power semiconductor chip temperature model with switching rules, the present application designs a power semiconductor chip temperature boundary controller manufacturing method and a control method. For other chips, they do not need to bear large voltage and current, so it is not necessary to make their heat generation modules work alternately. Therefore, general chips do not have switching mechanism, so they are not suitable for using switching adaptive boundary control. Therefore, the switching adaptive boundary controller is only applicable to chips similar to power semiconductor chips with switching mechanism.

[0028] As shown in Figure 1 , Figure 3 , a power semiconductor chip temperature boundary controller manufacturing method comprises the following steps:

[0029] Step 1, based on the power semiconductor chip temperature system model with continuous residence time switching mechanism, a switching linear boundary controller and a switching nonlinear adaptive boundary controller for eliminating the influence of boundary uncertainty are designed, wherein the switching linear boundary controller adopts a negative feedback boundary controller based on switching controller gain, the switching nonlinear adaptive boundary controller adopts a nonlinear adaptive boundary controller based on switching adaptive law and radial basis function, the input of the switching linear boundary controller is the output temperature T(x, t) of the power semiconductor chip, and the input of the switching nonlinear adaptive boundary controller is the temperature change rate T t (x, t) of the power semiconductor chip. t The output temperature T(x, t) and the temperature change rate T

[0030] Wherein, the power semiconductor chip temperature system model is based on the traditional power semiconductor chip temperature system model, the sustained dwell time switching mechanism and the boundary model uncertainty modeling, and the specific implementation form is: the power semiconductor chip is composed of two internal structure-ignored DMOS arrays. The epitaxial layer of the power semiconductor chip is the only heat generation layer of the DMOS array. The heat of the power semiconductor chip is generated by the heat generation area H1 or H2, first, the epitaxial layer of the power semiconductor chip is considered as the only heat generation layer due to high impedance. Secondly, since the periphery of the chip is usually arranged with interfaces and the like, the impedance is mainly concentrated in a certain area of the middle of the chip. Therefore, the area where the impedance is concentrated is defined as H1 or H2 (such as Figure 2 ). Then the temperature system of the power semiconductor chip has the switching characteristics due to the change of the heat generation areas H1 and H2.

[0031] The traditional power semiconductor chip temperature system model is

[0032] T t (x,t)=A θ(t) T xx (x,t)+b θ(t) (x)T(x,t) (1)

[0033] Wherein, T(x,t) represents the temperature of the power semiconductor chip, x and t represent the space and time variables of the power semiconductor respectively. T t (x,t) and T xx (x,t) represent the first order partial derivative of the temperature T(x,t) with respect to time and the second order partial derivative with respect to space respectively. θ(t) represents the average dwell time switching signal.

[0034] The sustained dwell time switching mechanism is adopted to describe the switching characteristics. The sustained dwell time switching mechanism can not only describe the system modeled by the average dwell time switching mechanism, but also model the system (such as the system with fast switching phenomenon) which cannot be modeled by the average dwell time switching mechanism. Specifically, due to the strict restriction of the dwell time switching mechanism and the average dwell time switching mechanism, the fast switching in the power semiconductor chip temperature system cannot be described by the dwell time and average dwell time switching mechanism, so the sustained dwell time switching mechanism is adopted to describe the switching in the power semiconductor chip temperature system with fast switching characteristics. On the other hand, since the uncertainty on the boundary of the system exists objectively, the uncertainty is fully considered in the system modeling. Based on the above discussion, the power semiconductor chip temperature system model is specifically represented as:

[0035] T t (x,t)=A δ(t) T xx (x,t)+bδ(t) (x, t) (2)

[0036] The boundary conditions of the system are expressed as:

[0037] T x (0, t) = 0, T x (L, t) = u(t) + f δ(t) (T t (L, t)) (3)

[0038] where T(x, t) represents the temperature of the power semiconductor chip, T x (0, t) and T x (L, t) represent the partial derivative values of the temperature of the power semiconductor chip with respect to space at 0 and L, respectively, T t (L, t) represents the partial derivative value of the temperature of the power semiconductor chip with respect to time at L, and x and t represent the spatial and temporal variables of the temperature system of the power semiconductor chip, respectively. T t (x, t) and T xx (x, t) represent the first-order partial derivative of the temperature T(x, t) with respect to time and the second-order partial derivative with respect to space, respectively. δ(t) is a right-continuous piecewise constant function representing a continuous dwell time switching signal, and for simplicity, let i be a constant, then the subsequent δ(t) is represented by i. Therefore, A δ(t) and b δ(t) (x) are represented by A i and b i (x), respectively. Further, A i represents the diffusion coefficient of the temperature system of the power semiconductor chip and satisfies the expression: b i (x) represents the thermal resistance of the temperature system of the power semiconductor chip and satisfies the expression: Since the thermal conductivity λ h is related to the physical structure of the heat generation region, as the heat generation region changes, the thermal conductivity λ h also changes accordingly, and therefore, the change process of the thermal conductivity λ h is represented by Further, since the material and shape of the power semiconductor chip are fixed, the changes of the parameters ρ, c s , T, and s(x) can be ignored. Specifically, λ h , ρ, c s , T, and s(x) represent the meanings as shown in Table 1.

[0039] Table 1. System parameters

[0040] Symbol Meaning Unit h ]]> ​ Thermal conductivity W / (m K) p Material density Kg / m 3 ]]> c s ]]> Specific heat J / (kg K) T Chip thickness pm s(x) Chip surface area (μm) 2 ]]>

[0041] u(t) represents the control input of the system, and L represents the length of the chip. i (T t (L,t))。

[0042] To perform the subsequent steps, the following is described about the persistent dwell time switching mechanism:

[0043] Statement 1: The persistent dwell time switching mechanism is defined as:

[0044] For any switching signal δ(t), if the following two conditions are met simultaneously, δ(t) is a persistent dwell time switching signal.

[0045] 1. There are infinitely many time intervals (referred to as τ-Portion) that are not less than a positive constant τ PDT , on each τ-Portion, the value of δ(t) is constant.

[0046] 2. The interval between two consecutive τ-Portions (referred to as T-Portion) is separated by a time interval that is not greater than a positive constant T PDT , where τ PDT and T PDT are respectively referred to as persistent dwell time and persistent period, and are constants greater than zero.

[0047] Statement 2: As shown in Figure 4 , δ (q,r) represents the qth switching time of the rth stage of the system, and r and q are normal numbers. δ (q,r) represents the number of switching times of the system in the T-Portion of the rth stage.

[0048] For such unknown nonlinear function f i (T t (L,t)), the present application uses radial basis function neural network technology to approximate f i (T t (L,t)), which can be specifically represented as:

[0049]

[0050] In the formula: represents an ideal weight vector, the estimated value of which is represented by , where, and χ i are constants greater than zero. represents the radial basis function of the input boundary temperature rate T t (L,t), and ε i (T t(L, t) represents the approximation error and satisfies is a constant greater than zero, represents the transpose of

[0051] The specific implementation method of step 1 is:

[0052] The switching linear boundary controller adopts a traditional negative feedback boundary controller based on switching controller gain, and the state input signal of the controller is the temperature T(x, t) of the power semiconductor chip. The switching nonlinear adaptive boundary controller adopts a nonlinear adaptive boundary controller based on switching adaptive law and radial basis function, which is used to compensate for the influence of boundary uncertainty. At the same time, the state input of the switching nonlinear adaptive boundary controller is the measured temperature rate T t (x, t). Further, the specific expressions of the switching linear boundary controller and the switching nonlinear adaptive boundary controller are:

[0053] The switching linear boundary controller u L (t):

[0054] u L (t) = -K i T(L, t) (5)

[0055] In the formula: K i represents the controller gain, and T(L, t) represents the boundary temperature of the system.

[0056] The switching nonlinear adaptive boundary controller u N (t):

[0057]

[0058] In the formula: represents the estimated value of the ideal weight vector . represents the radial basis function of the input boundary temperature rate T t (L, t).

[0059] Step 2, combine the switching linear boundary controller and the switching nonlinear adaptive boundary controller designed in step 1 into a switching adaptive boundary controller.

[0060] The specific implementation method is that the switching adaptive boundary controller designed in step 2 satisfies the expression:

[0061] Due to the introduction of the switching adaptive boundary controller u(t), it can be ensured that the system reaches stability and tends to be within a bounded range under the action of the designed control input.

[0062] The method for controlling temperature of power semiconductor chip is characterized in that the solved switching adaptive boundary controller gain is used in the switching adaptive boundary controller (7) to realize the reasonable control of the temperature of the power semiconductor chip.

[0063] The specific method for solving the switching adaptive boundary controller gain is that the sufficient condition of input-to-state exponential stability of the power semiconductor chip temperature system is determined by using Lyapunov stability theory, and the switching signal law of the power semiconductor chip temperature system model is designed by iteration and recursion method on the basis of the sufficient condition, and the switching adaptive boundary controller gain is solved by using the LMI toolbox of MATLAB according to the signal law.

[0064] The solving process of the switching adaptive boundary controller gain is as follows:

[0065] Step S1, the stability of the power semiconductor chip temperature system is analyzed according to the Lyapunov stability theory, and the sufficient condition of input-to-state exponential stability of the power semiconductor chip temperature system is determined.

[0066] Firstly, the Lyapunov function is constructed as follows:

[0067]

[0068] In the formula, P represents the given adaptive gain, i represents a positive definite matrix, represents the estimation error, wherein, T T (x, t) is the transpose of T(x, t), χ i , are all normal numbers.

[0069] Then, in order to obtain the solvable sufficient condition, the derivative of formula (8) is obtained, and according to the boundary condition (3), Wirtinger inequality and Young inequality, the following formula (9) can be obtained:

[0070]

[0071] In the formula, θ1, θ2 and are greater than zero.

[0072] Further, the adaptive law of the switching nonlinear adaptive boundary controller is designed as follows:

[0073]

[0074] wherein: i i is a constant greater than zero.

[0075] The adaptive law of the switching nonlinear adaptive boundary controller is combined with (8), (9) and Young inequality, to obtain a Lyapunov function satisfying the condition, i.e.

[0076]

[0077] wherein:

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] wherein, col[T(L,t),T(x,t)] represents a column vector composed of T(L,t) and T(x,t), represents and 2i i the minimum value in. represents the maximum value of thermal resistance, ∈ and i i is a constant greater than zero, thus obtaining the condition satisfying is: Φ i < 0.

[0084] Step S2, designing a suitable continuous dwell time switching signal law: on the basis of sufficient conditions, the switching signal law of the power semiconductor chip temperature system model is designed by iteration and recursion method.

[0085] Let and wherein, and μ are a constant greater than zero and a constant greater than 1 respectively. Further, it can be obtained that:

[0086]

[0087] wherein: according to instruction 2, and are two new variables defined by us, and β is an auxiliary variable.

[0088] Further, by iteration and recursion method, it can be obtained that:

[0089]

[0090] wherein: f represents the maximum switching frequency on the rth stage T-Portion, and is a constant given greater than zero.

[0091] Combining (8) and (18), we get:

[0092]

[0093] where: σ = (T PDT f+1)lnμ-α(f -1 +τ PDT )<0,

[0094] λ i,min (P i ) and λ i,max (P i ) represent the minimum and maximum eigenvalues of P i , respectively,

[0095] Therefore, according to the above analysis, the invention obtains the sufficient condition of system stability: in the case of Φ i <0, and σ = (T PDT f+1)lnμ-α(f -1 +τ PDT )<0, the closed-loop system (2)-(3) satisfies the exponential input-to-state stability, that is, the state of the system converges to a bounded range with a convergence rate of α. Therefore, under the constraints of Φ i <0, and σ = (T PDT f+1)lnμ-α(f -1 +τ PDT )<0, the temperature of the power semiconductor chip temperature system converges to a bounded range with a convergence rate of α, and the dwell time switching signal used to describe the switching rule of the power semiconductor chip temperature system satisfies σ = (T PDT f+1)lnμ-α(f -1 +τ PDT )<0.

[0096] Finally, by decoupling the matrix of formula (14), substituting the known system parameters and the parameters given in the switching adaptive boundary controller design process into formula (12) and (14)-(16), and then using the LMI toolbox to obtain the switching adaptive boundary controller gain.

[0097] To explain the control effect of the scheme of the present application in detail, simulation will be carried out in MATLAB next, and the system parameters are set as follows in the simulation: the system modal number i=2. When i=1, the diffusion coefficient of the system A1=1, the thermal resistance Model uncertainty f1(T t (L,t))=10sin(T t (L,t)); when i=2, the diffusion coefficient of the system A2=2, the thermal resistance Model uncertainty f2(T t (L,t)=11sin(T t (L,t). Through MATLAB simulation, the open-loop evolution of the power semiconductor chip temperature system Figure 5 indicates that the system is divergent, and therefore it is necessary to exert control on the system. Further, the controller designed in the present application is exerted on the boundary of the system, and the closed-loop evolution of the system state Figure 6 and the switching adaptive boundary control input Figure 7 indicates that the switching adaptive boundary controller designed is effective. Further, in the case where unknown uncertainty exists in the boundary of the system, if a conventional boundary controller is designed directly by ignoring the uncertainty, then under the action of the controller, the state evolution of the system Figure 8 and the control input Figure 9 indicates that the controller designed by ignoring the uncertainty cannot stabilize the power semiconductor chip temperature system. Therefore, by comparing the control effects of the switching adaptive boundary control and the conventional boundary control, the switching adaptive boundary controller designed in the present application is obviously superior to the conventional boundary controller.

[0098] The above description of the disclosed embodiments enables one skilled in the art to make or use the present application. Numerous modifications to these embodiments will be apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Therefore, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of manufacturing a power semiconductor chip temperature boundary controller, the method comprising: providing a power semiconductor chip; providing a temperature sensor; providing a temperature boundary controller; and coupling the temperature sensor to the temperature boundary controller. The method comprises the following steps: Step 1, based on the power semiconductor chip temperature system model with the mechanism of switching with constant residence time, design switching linear boundary controller and switching nonlinear adaptive boundary controller to eliminate the influence of boundary uncertainty, wherein the switching linear boundary controller adopts negative feedback boundary controller based on switching controller gain, and the switching nonlinear adaptive boundary controller adopts nonlinear adaptive boundary controller based on switching adaptive law and radial basis function; the input of the switching linear boundary controller is the output temperature T(x, t) of the power semiconductor chip, and the input of the switching nonlinear adaptive boundary controller is the temperature change rate T t (x, t) of the power semiconductor chip; The power semiconductor chip temperature system model is specifically expressed as: T t (x,t) = A δ(t) T xx (x,t) + b δ(t) (x)T(x,t) The boundary condition of the system is expressed as: T x (0,t) = 0, T x (L,t) = u(t) + f δ(t) (T t (L,t)) where T(x, t) represents the temperature of the power semiconductor chip, T x (0, t) and T x (L, t) represent the partial derivative values of the temperature of the power semiconductor chip with respect to space at 0 and L, respectively, T t (L, t) represents the partial derivative value of the temperature of the power semiconductor chip with respect to time at L, x and t represent the spatial and temporal variables of the temperature system of the power semiconductor chip, respectively; T t (x, t) and T xx (x, t) represent the first-order partial derivative of the temperature T(x, t) with respect to time and the second-order partial derivative with respect to space, respectively; δ(t) is a right-continuous piecewise constant function representing the persistent dwell time switching signal, and δ(t) is defined as i is a constant, then the subsequent δ(t) is represented by i; A δ(t) and b δ(t) (x) are represented by A i and b i (x), respectively; A i represents the diffusion coefficient of the temperature system of the power semiconductor chip and satisfies the expression: b i (x) represents the thermal resistance of the temperature system of the power semiconductor chip and satisfies the expression: Since the thermal conductivity λ h is related to the physical structure of the heat generation region, as the heat generation region changes, the thermal conductivity λ h also changes accordingly, so the change process of the thermal conductivity λ h is represented by λ ; Since the material and shape of the power semiconductor chip are fixed, the changes of the parameters ρ, c s , T and s(x) are ignored; λ h is the thermal conductivity, ρ is the material density, c s is the specific heat, T is the chip thickness, and s(x) is the chip surface area; The persistent dwell time switching mechanism is described as follows: Explanation 1: The persistent dwell time switching mechanism is defined as: For any switching signal δ(t), if the following two conditions are met simultaneously, δ(t) is a persistent dwell time switching signal; 1. There are infinitely many time intervals not smaller than a positive constant τ PDT called τ-Portions, on each of which the value of δ(t) is constant; 2. The interval between two consecutive τ-Portions is called a T-Portion separated by a time interval not greater than a positive constant T PDT , where τ PDT and T PDT are respectively called the duration of the dwell time and the duration of the cycle, and are positive constants; Description 2: denotes the qth switching time of the rth stage of the system, r and q are normal numbers, denotes the number of switching times of T-Portion of the rth stage of the system; The specific expressions of the switching linear boundary controller and the switching nonlinear adaptive boundary controller are as follows: Switching linear boundary controller u L (t): u L (t) = -K i T(L, t) where: K i represents the controller gain, T(L, t) represents the boundary temperature of the system; Switching nonlinear adaptive boundary controller u N (t): where: represents an estimate of the ideal weight vector represents that the input is the boundary temperature rate of change T t a radial basis function of (L, t);​ Step 2: The switching linear boundary controller and the switching nonlinear adaptive boundary controller designed in step 1 are combined into a switching adaptive boundary controller.

2. A method of manufacturing a power semiconductor chip temperature boundary controller as defined in claim 1, characterized by: The power semiconductor chip temperature system model is based on a traditional power semiconductor chip temperature system model, a persistent dwell time switching mechanism and boundary model uncertainty.

3. A method of controlling the temperature of a power semiconductor chip, characterized by: The specific implementation method is that the switching adaptive boundary controller gain solved is used in the switching adaptive boundary controller of claim 1 or 2, so that the power semiconductor chip temperature is bounded and convergent, and reasonable control of the power semiconductor chip temperature is realized.

4. The method of claim 3, wherein: the temperature of the power semiconductor chip is controlled by: determining a temperature of the power semiconductor chip; and adjusting the temperature of the power semiconductor chip by adjusting the temperature of the heat sink. The specific method for solving the switching adaptive boundary controller gain is that a sufficient condition for input-to-state exponential stability of the power semiconductor chip temperature system is determined by using Lyapunov stability theory, on the basis of the sufficient condition, a switching signal law of the power semiconductor chip temperature system model is designed through iteration and recursion, and the switching adaptive boundary controller gain is obtained by solving the sufficient condition for input-to-state exponential stability by using an LMI toolbox of MATLAB according to the signal law.

Images