An infrared radiation intensity control method based on constraint optimization and inverse system decoupling

Abstract

The application discloses an infrared radiation intensity control method based on constraint optimization and inverse system decoupling. First, the application establishes a mapping relationship between spectral radiation intensity and molybdenum-silicon rod temperature based on blackbody radiation theory, and converts spectral radiation intensity control into temperature control. Then, aiming at the coupling relationship between the molybdenum-silicon rod temperature and the cavity temperature, a BP neural network inverse system model is constructed to realize decoupling of the double-temperature control channel. Next, a comprehensive objective optimization function is established, which includes a temperature tracking error, a spectral radiation intensity error, a control quantity change smoothing term, a channel coupling suppression term and a constraint penalty term, and an improved whale optimization algorithm is used to optimize the PID controller parameters. Finally, the optimized PID parameters are applied to the closed-loop control system to realize stable control of the output spectral radiation intensity of the infrared radiation simulation source. The application improves the spectral radiation intensity control precision and the system operation stability, and has good engineering application value.

Description

Technical Field

[0001] This invention belongs to the field of infrared radiation control technology, specifically relating to an infrared radiation intensity control method based on constraint optimization and inverse system decoupling. Background Technology

[0002] Infrared radiation simulation devices are used to generate controllable infrared radiation and can be used for performance verification and calibration of infrared detectors, infrared imaging systems, seekers, and thermal control systems. Most existing infrared radiation simulation devices are based on blackbody radiation theory, utilizing high-temperature blackbody sources, electric heating rods, or other heating elements as radiation sources. According to Planck's radiation law and the Stefan-Boltzmann law, radiation intensity is closely related to temperature; therefore, existing devices typically control radiation intensity by adjusting the temperature of the radiation source.

[0003] For infrared radiation simulation devices using silicon molybdenum rods as radiation sources, there is a coupling relationship between the operating temperature of the silicon molybdenum rod and the cavity temperature. Furthermore, the system is also affected by changes in ambient temperature, cooling conditions, and control variables, resulting in a multivariable, nonlinear, and strongly coupled system. When using traditional PID control methods, the two temperature control channels typically act directly on the temperature-controlled coupled object, easily leading to inter-channel interference and problems such as response lag, regulation oscillation, increased overshoot, and decreased control accuracy. Although inverse system control methods or intelligent optimization methods exist for temperature control and parameter tuning, most methods still primarily rely on temperature error as the optimization basis, lacking unified consideration of output spectral radiation intensity error, smoothness of control variable changes, coordination between the two temperature channels, and system operating boundary constraints. Therefore, existing methods struggle to meet the control requirements of infrared radiation simulation sources for stable and accurate output of spectral radiation intensity in a specified band. Summary of the Invention

[0004] To address the problems in existing infrared radiation simulation devices, such as the coupling between the temperature of the silicon molybdenum rod and the cavity temperature, the difficulty in achieving stable and precise control of the output spectral radiation intensity, and the insufficient consideration of radiation target and system constraints during control parameter tuning, this invention proposes an infrared radiation intensity control method based on constraint optimization and inverse system decoupling.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: First, the method establishes a mapping relationship between the output spectral radiation intensity and temperature, transforming radiation intensity control into temperature control; addressing the problem of coupling between the silicon molybdenum rod temperature and the cavity temperature, making it difficult to independently adjust the two temperature variables, a BP neural network inverse system decoupling control structure is constructed, and incremental PID controllers are set in the two temperature control channels respectively to achieve coordinated control of the two temperature channels; addressing the problem of insufficient consideration of radiation target, channel coupling influence, control quantity smoothness, and system operation constraints in the existing parameter tuning process, a comprehensive objective optimization function is established, and a constraint optimization method is used to solve the PID parameters of the two temperature control channels, thereby improving the control accuracy of the output spectral radiation intensity of the infrared radiation simulation source and the system operation stability.

[0006] The present invention provides an infrared radiation intensity control method based on constraint optimization and inverse system decoupling, comprising the following steps:

[0007] S1: Based on the blackbody radiation theory and introducing the radiation source spectral emissivity correction, a mapping relationship is established between the output spectral radiation intensity of the infrared radiation simulation source and the temperature of the silicon molybdenum rod, transforming radiation intensity control into temperature control.

[0008] S2: Based on the mapping relationship established in S1, and combined with the coupling relationship between the temperature of the silicon molybdenum rod and the temperature of the cavity in the infrared radiation simulation source temperature control system, construct the inverse system model of the temperature control coupling system; learn the inverse system characteristics of the infrared radiation simulation source temperature control coupling system to obtain the BP neural network inverse system model;

[0009] S3: Connect the BP neural network inverse system model obtained in S2 in series with the temperature control coupling system to construct an equivalent decoupled dual temperature control channel, and set an incremental PID controller in each control channel to form an inverse system decoupled control structure.

[0010] S4: Based on the inverse system decoupling control structure described in S3, a comprehensive objective optimization function is established with the output spectral radiance intensity error as the core evaluation term. The comprehensive objective optimization function includes a temperature tracking error term, a spectral radiance intensity error term calculated from the temperature-spectral radiance intensity mapping relationship, a control quantity change smoothing term, a channel coupling suppression term, and a constraint penalty term.

[0011] S5: Based on the comprehensive objective optimization function and combined with the operating constraints of silicon molybdenum rod temperature, cavity temperature, driving current, and cooling system operating power, an improved whale optimization algorithm is used to solve for the parameters of each incremental PID controller; the improved whale optimization algorithm generates an initial population through a Logistic-Tent hybrid chaotic mapping and introduces adaptive inertia weights during the iterative search process. - Greedy perturbation strategy to obtain the optimal PID parameters that satisfy the system constraints;

[0012] S6: Update the optimal PID parameters obtained in S5 online to the corresponding incremental PID controller; during system operation, collect the temperature of the silicon molybdenum rod and the cavity temperature in real time, and calculate the estimated value of the current output spectral radiation intensity based on the temperature-spectral radiation intensity mapping relationship; the incremental PID controller outputs the control adjustment amount according to the temperature deviation signal, and outputs the silicon molybdenum rod drive current and cooling system operating power after decoupling through the BP neural network inverse system, realizing closed-loop control of the output spectral radiation intensity of the infrared radiation simulation source based on temperature feedback and radiation intensity mapping.

[0013] The advantages of this invention compared to the prior art are:

[0014] 1. This invention proposes a mapping method between the output spectral radiation intensity of an infrared radiation simulation source and its operating temperature, transforming the radiation intensity control problem into a temperature control problem, and realizing adjustable control of the output spectral radiation intensity within a specified band range.

[0015] 2. This invention addresses the coupling issue between the operating temperature of the silicon molybdenum rod and the cavity temperature by constructing a dual-temperature inverse system decoupling control structure. By introducing a BP neural network inverse system between the two temperature control channels and the temperature control coupling system, the cross-influence between the two temperature channels can be reduced, improving the stability and control accuracy of the dual-temperature control process.

[0016] 3. To address the issue of insufficient consideration of radiation targets, system constraints, and coupling effects in existing parameter tuning processes, this invention introduces a comprehensive objective optimization function in the PID parameter optimization process. This function comprises temperature tracking error, spectral radiation intensity error, control quantity change smoothing term, channel coupling suppression term, and constraint penalty term. The spectral radiation intensity error term and the channel coupling suppression term are used to adjust the search weights of the improved whale optimization algorithm, enabling the parameter optimization process to adaptively search based on the radiation output deviation of the infrared radiation simulation source and the coupling state of the dual temperature channels. Therefore, the optimized PID parameters not only meet temperature tracking requirements but also balance radiation intensity control accuracy, channel coordination, and system operational safety. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the closed-loop control structure for infrared radiation intensity.

[0018] Figure 2 This is a raw data distribution diagram of temperature and spectral radiance intensity within the corresponding wavelength range;

[0019] Figure 3 This is a distribution chart of the normalized sample data;

[0020] Figure 4 This is a schematic diagram of the inverse system structure of a BP neural network;

[0021] Figure 5 This is a block diagram of a dual-temperature decoupling control structure based on a BP neural network inverse system.

[0022] Figure 6 Flowchart for improving whale algorithm parameter optimization;

[0023] Figure 7 This is a comparison chart of the temperature response curves of a silicon molybdenum rod under the traditional PID control method and the method of this invention. Figure 7 (a) in the figure is the temperature response curve of the silicon molybdenum rod under the traditional PID control method; Figure 7 (b) in the figure is the temperature response curve of the silicon molybdenum rod under the method of the present invention;

[0024] Figure 8 This is a comparison chart of the cavity temperature response curves under the traditional PID control method and the method of this invention. Figure 8 (a) in the figure is the cavity temperature response curve under the traditional PID control method; Figure 8 (b) in the figure is the cavity temperature response curve under the method of the present invention. Detailed Implementation

[0025] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0026] like Figure 1 The diagram shown is the overall flowchart of the infrared radiation intensity control method of the present invention. During the operation of the infrared radiation simulation source device, there is a coupling relationship between the temperature of the silicon molybdenum rod and the cavity temperature, which can easily affect the control accuracy and stability of the output spectral radiation intensity. Therefore, the present invention provides an infrared radiation intensity control method based on constraint optimization and inverse system decoupling, used to achieve closed-loop control of the output spectral radiation intensity of the infrared radiation simulation source. The specific implementation steps are as follows:

[0027] Step 1: Establish the mapping relationship between the output spectral radiance intensity of the infrared radiation simulation source and temperature.

[0028] The infrared radiation simulation source device uses a single silicon molybdenum rod as the infrared radiation simulation source. Its radiation spectrum range is 1.5μm to 5.5μm, and its normal operating temperature range is 1500K to 2000K. The output spectral radiation intensity can be adjusted by adjusting the temperature of the silicon molybdenum rod.

[0029] This embodiment is based on the blackbody radiation theory, which includes Planck's radiation law to describe the spectral radiation distribution characteristics within a selected wavelength range, and Stefan-Boltzmann's law to characterize the overall trend of the total radiation energy changing with temperature. The spectral emissivity of the silicon molybdenum rod surface and the energy utilization coefficient of the infrared radiation simulation source device are introduced for correction, and a mapping relationship between the temperature of the silicon molybdenum rod and the spectral radiation intensity within a specified wavelength range is established, as shown in Equation (1-1).

[0030] in, For wavelength variables, For a infinitesimal element of wavelength, Let be the first radiation constant, and take . , Let be the second radiation constant, and take . , The operating temperature of the silicon molybdenum rod. , The radiation source temperature is respectively Spectral emissivity and energy utilization of infrared radiation simulation source device at that time These are the lower and upper limits of the selected band, respectively; in this embodiment, they are both set to 3.5. and 4.0 , For band range spectral radiance S is the radiation area of ​​a single silicon molybdenum rod radiation source, and θ is the radiation field of view.

[0031] Equation (1-1) provides the relationship between the temperature of the silicon molybdenum rod and the spectral radiation intensity within the selected band range. This transforms the output spectral radiation intensity control into silicon molybdenum rod temperature control, providing a foundation for subsequent temperature control coupling system modeling, inverse system decoupling control structure construction, and control parameter optimization.

[0032] Step 2: Establish a BP neural network inverse system model for the temperature control coupling system.

[0033] During the operation of the infrared radiation simulation source device, there is a coupling relationship between the temperature of the silicon molybdenum rod and the temperature of the cavity. The system is simultaneously affected by changes in ambient temperature and control variables, making it a multi-input, multi-output nonlinear coupled system. To achieve the inverse solution of the control input variables, this embodiment establishes a BP neural network inverse system model for the temperature control coupled system.

[0034] This BP neural network consists of an input layer, hidden layers, and an output layer, where the input layer variable is the temperature control channel adjustment amount of the silicon molybdenum rod. Cavity temperature control channel adjustment amount and ambient temperature The output layer variable is the silicon molybdenum rod drive current. and cooling system operating power .

[0035] in, It is the temperature control channel adjustment amount of the silicon molybdenum rod at the nth sampling time. It is the adjustment amount of the cavity temperature control channel at the nth sampling time. It is the ambient temperature at the nth sampling time. It is the silicon molybdenum rod driving current at the nth sampling time. The system operating power at the nth sampling time.

[0036] It should be noted that the actual input to the temperature control coupling system is still the silicon molybdenum rod drive current. and cooling system operating power The control channel adjustment quantity is the intermediate adjustment signal output by the incremental PID controller, which serves as the input to the BP neural network inverse system model. After decoupling and transformation by the BP neural network inverse system model, the actual driving current and cooling system operating power acting on the temperature control coupling system are obtained.

[0037] Based on the above variable relationships, the input-output relationship of the temperature control system can be expressed as: in This represents the nonlinear dynamic relationship of the temperature control system. To solve for the control input, its inverse mapping relationship needs to be established: in, This represents the inverse system function of the temperature control system. Since this inverse mapping relationship is difficult to obtain directly through analytical methods, a backpropagation neural network is used to learn and model it, thereby obtaining the inverse system model of the temperature control system.

[0038] S2-1. Data Acquisition.

[0039] The infrared radiation simulation source device is used to collect operating data under different working conditions according to a set sampling period. The collected variables include the temperature of the silicon molybdenum rod. Cavity temperature Ambient temperature Drive current and cooling system operating power Based on the temperature of the silicon molybdenum rod Its set value Deviation between them, and cavity temperature Its set value The deviation between them is used to obtain the temperature control channel adjustment amount of the silicon molybdenum rod. Adjustment amount of cavity temperature control channel .in, and As the input variable of the inverse system of the BP neural network, the driving current and cooling system operating power As the output variable of the inverse system of a BP neural network, ambient temperature Introduced as a disturbance variable, this forms the training samples for the BP neural network inverse system, used to learn the nonlinear inverse mapping relationship between the actual control input and different temperature regulation requirements and environmental disturbances.

[0040] Based on the temperature-spectral radiance mapping relationship established in step one, the spectral radiance data within the corresponding wavelength range is calculated from the temperature of the silicon molybdenum rod, forming a temperature-spectral radiance sample relationship for subsequent calculation of the spectral radiance error term. For example... Figure 2 As shown, Figure 2 This is the original data distribution map, which reflects the sample distribution between temperature and spectral radiance within the corresponding wavelength range.

[0041] S2-2. Data preprocessing.

[0042] To reduce the impact of differences in the dimensions of different physical quantities on neural network training, the running data collected in S2-1 is normalized. This embodiment uses the minimum-maximum normalization method, the expression of which is: in: This is the original data. For the normalized data, and These represent the minimum and maximum values ​​of the variable in the sample data, respectively. After normalization, all input and output variables are mapped to a uniform numerical range. For example... Figure 3 As shown, Figure 3 This is a distribution chart of the normalized sample data.

[0043] S2-3. Neural Network Structure Design.

[0044] like Figure 4 As shown, a BP neural network model is constructed to learn the inverse system characteristics of a temperature-controlled coupled system. This BP neural network includes an input layer, hidden layers, and an output layer, where the input layer variable is the temperature control channel adjustment amount of the silicon molybdenum rod. Cavity temperature control channel adjustment amount and ambient temperature The output layer variable is the silicon molybdenum rod drive current. and cooling system operating power .

[0045] In this embodiment, the BP neural network adopts a four-layer structure, including an input layer, a first hidden layer, a second hidden layer, and an output layer; the input layer has 3 nodes, the first hidden layer has 10 nodes, the second hidden layer has 6 nodes, and the output layer has 2 nodes. Both the first and second hidden layers use the ReLU activation function, the expression of which is: The output layer uses the Linear activation function, whose expression is: The ReLU function is used for nonlinear mapping of the hidden layer, and the Linear function is used for continuous control variable output of the output layer.

[0046] S2-4. Neural Network Training.

[0047] The BP neural network was trained offline using the preprocessed S2-2 data to learn the nonlinear inverse mapping relationship between the input and output of the temperature control coupling system. The mean square error function was used as the performance evaluation metric during training, and its expression is: Where M is the number of training samples. For the actual output value of the sample, The neural network predicts the output value. By continuously adjusting the network weights and thresholds through backpropagation, the BP neural network inverse system model of the temperature control coupling system is obtained when the training error meets the set accuracy requirements.

[0048] Because the infrared radiation simulation source temperature control system is simultaneously affected by ambient temperature disturbances and changes in control quantities, and because there is a significant coupling relationship between the silicon molybdenum rod temperature and the cavity temperature, it is difficult to directly establish the inverse mapping relationship of the control input using analytical methods. Therefore, this embodiment introduces ambient temperature as a disturbance variable into the inverse system modeling process and employs a BP neural network to learn the characteristics of the inverse system. This enables the system to determine the corresponding drive current and cooling system operating power based on the adjustment requirements of the two temperature control channels and the ambient temperature disturbances.

[0049] The inverse system model established in this way can not only reflect the nonlinear relationship of the two-temperature coupled objects, but also improve the model's adaptability to changes in actual operating conditions, providing a model basis for the subsequent construction of a decoupling control structure for the two-temperature inverse system.

[0050] Step 3: Construct a decoupled control structure for a dual-temperature inverse system.

[0051] To reduce the cross-coupling effect between the silicon molybdenum rod temperature and the cavity temperature, the BP neural network inverse system model obtained in step two is connected in series with the temperature control coupling system of the infrared radiation simulation source device, forming a dual-temperature inverse system decoupled control structure. For example... Figure 5 As shown, the coupling relationship between the temperature of the silicon molybdenum rod and the cavity temperature is compensated by the backpropagation (BP) neural network inverse system obtained through training. This structurally decouples the originally coupled temperature control system, thus forming two relatively independent temperature control channels: the temperature of the silicon molybdenum rod and the temperature of the cavity. Control channel and cavity temperature Control channels. Based on this, incremental PID controllers are set in two temperature control channels to perform closed-loop regulation of the silicon molybdenum rod temperature and the cavity temperature.

[0052] During system operation, the setpoint for the silicon molybdenum rod temperature is first determined based on the target spectral radiance, and the setpoint for the cavity temperature is simultaneously determined based on system operating requirements. The system collects the silicon molybdenum rod temperature and cavity temperature in real time and compares them with the corresponding setpoints to obtain a temperature control deviation signal. This deviation signal is then input to the respective incremental PID controllers for adjustment calculations, yielding the control adjustment quantity. This control adjustment quantity serves as the input signal to the inverse BP neural network system. After decoupling calculation by the inverse BP neural network system, the actual system control input is obtained, thereby adjusting the heating power of the silicon molybdenum rod and the operating power of the cooling system, respectively.

[0053] In the closed-loop control process of the temperature control system, the deviation between the controlled variable and its set value is defined as follows:

[0054] The temperature control deviation of the silicon molybdenum rod is: in, This represents the measured temperature of the silicon molybdenum rod at the nth sampling time. The temperature setpoint is calculated from the target spectral radiation intensity using equation (1-1);

[0055] The cavity temperature control deviation is: in, This represents the measured temperature of the cavity at the nth sampling time. The temperature setpoint for the cavity is used; the temperature deviation signal is used as the input to the PID controller to calculate the control adjustment.

[0056] An incremental PID controller is set up in each of the two temperature control channels, and its output is used as the input signal of the inverse system of the BP neural network. The calculation formulas for the output of the two incremental PID controllers are as follows.

[0057] Temperature control channel for silicon molybdenum rod: Cavity temperature control channel: in, This represents the input of the BP neural network inverse system module in the temperature control channel of the silicon molybdenum rod at time n. This is the input corresponding to time n-1; , , These are the error signals of the silicon molybdenum rod temperature at times n, n-1, and n-2, respectively. , , These are the proportional, integral, and derivative coefficients of the temperature control channel for the silicon molybdenum rod, respectively.

[0058] This represents the input of the BP neural network inverse system module in the cavity temperature control channel at time n. This is the input corresponding to time n-1; , , These represent the error signals of the cavity temperature at times n, n-1, and n-2, respectively. , , These are the proportional, integral, and derivative coefficients of the cavity temperature control channel, respectively.

[0059] The BP neural network inverse system obtained in step two is used to calculate the system control input, and its output consists of two control signals. and .in, Mainly used for regulating the temperature of silicon molybdenum rods. It is mainly used to regulate the cavity temperature. Through the above control structure, the cross-coupling effect between the two temperature control channels can be reduced, providing a basis for subsequent optimization of control parameters and stable control of the output spectral radiance intensity of the infrared radiation simulation source.

[0060] In traditional dual-channel temperature control, the two control channels typically act directly on the temperature control coupling system. Adjusting the temperature of the silicon molybdenum rod affects the cavity temperature, and vice versa, easily leading to overshoot, oscillation, and prolonged settling time.

[0061] To address the aforementioned issues, this embodiment introduces a BP neural network inverse system between the two incremental PID controllers and the temperature control coupling system. This system converts the temperature control channel adjustment output from the PID controllers into drive current and cooling system operating power, ensuring that the control quantity is decoupled and distributed before entering the temperature control coupling system. With this structure, the silicon molybdenum rod temperature control channel is primarily used for radiation source temperature regulation, while the cavity temperature control channel is mainly used for cavity thermal state regulation, thus reducing the cross-influence between the two temperature channels.

[0062] Since step one has transformed the output spectral radiance control into silicon molybdenum rod temperature control, the dual-temperature decoupled control structure not only improves the temperature response process but also directly affects the stability of the radiance output. Through this structure, the silicon molybdenum rod temperature can more accurately correspond to the target spectral radiance, and the cavity temperature can maintain a relatively stable thermal environment, providing a foundation for subsequent PID parameter optimization and closed-loop radiance control.

[0063] Step 4: Construct a comprehensive objective optimization function for spectral radiance intensity control.

[0064] Based on the decoupled control structure of the dual-temperature inverse system constructed in step three, in order to match the incremental PID parameters in the two temperature control channels with the dual-temperature coupling characteristics of the infrared radiation simulation source device, the output spectral radiation intensity control target, and the system operation constraints, a comprehensive objective optimization function is established to evaluate the overall performance of the control system, and based on this, the incremental PID parameters in the two temperature control channels are constrained and optimized.

[0065] The comprehensive objective optimization function consists of a temperature tracking error term. Spectral radiation intensity error term Smoothing term of control quantity change Channel coupling suppression term and constraints and penalties Its composition, its expression is: in, These are weighting coefficients, used to adjust the relative weights of each evaluation item in the overall objective optimization function. , .

[0066] The temperature tracking error term The expression is: in, This represents the number of sampling steps. and These are the measured values ​​of the silicon molybdenum rod temperature and the cavity temperature at the nth sampling time, respectively; The temperature setpoint for the silicon molybdenum rod is obtained by inversely calculating the target spectral radiance intensity according to equation (1-1). Set the cavity temperature value; and This is the temperature error weighting coefficient.

[0067] The spectral radiation intensity error term The expression is: in, The band spectral radiance is calculated from the temperature of the silicon molybdenum rod at the nth sampling time using equation (1-1). The target is the spectral radiance intensity. Compared to parameter tuning methods that only use temperature error as the optimization target, the spectral radiance intensity error term directly incorporates the output radiation control target of the infrared radiation simulation source into the PID parameter optimization process, enabling the parameter search results to simultaneously serve both temperature tracking accuracy and spectral radiance intensity control accuracy.

[0068] The control quantity change smoothing term The expression is: in, and These represent the drive current and the cooling system operating power at the nth sampling time, respectively. and These are the control quantities corresponding to the previous sampling time; and The weighting coefficient for the change in the control quantity.

[0069] The channel coupling suppression term The expression is: in: in, This is due to temperature control deviation of the silicon molybdenum rod; This refers to the cavity temperature control deviation. The channel coupling suppression term characterizes the impact on the coordinated control performance of the system when deviations in two temperature control channels coexist, and suppresses the situation where both temperature channels deviate from the set state simultaneously during parameter optimization.

[0070] The constraint penalty item The expression is: in: in, and These are the lower and upper limits of the temperature for silicon molybdenum rods, respectively. and These are the lower and upper limits of the cavity temperature, respectively. and These are the lower and upper limits of the drive current, respectively. and These are the lower and upper limits of the cooling system's operating power, respectively. For the first The change in driving current at the sampling time is the first sampling time. Changes in cooling system operating power at each sampling time and These are the allowable upper limits for changes in drive current and changes in cooling system operating power, respectively. , , This refers to the corresponding constraint penalty weight coefficient.

[0071] Step 5: Solve the PID parameter constraint optimization based on the improved whale optimization algorithm.

[0072] like Figure 6 As shown, this embodiment uses the improved whale algorithm with constraint optimization to solve the incremental PID parameters of the two temperature control channels in step three. The specific process is as follows.

[0073] S5-1. Parameter range determination and population initialization.

[0074] First, the parameter sets and search ranges of the proportional, integral, and derivative coefficients of the incremental PID controllers in the two temperature control channels are determined. Based on this, an initial whale population is generated using a Logistic-Tent mixed chaotic mapping to initialize the individual whale parameters. Each individual whale in the population corresponds to a set of PID parameter combinations.

[0075] S5-2. Fitness calculation and initial population screening.

[0076] The PID parameters corresponding to each individual whale are substituted into the decoupled control structure of the dual-temperature inverse system constructed in step three for simulation calculation. The comprehensive objective optimization function value is obtained based on the response results of the temperature-controlled coupled system, and this value is used as the individual fitness value. Subsequently, individuals with fitness values ​​lower than a set threshold are selected to form the final initial whale population for subsequent iterative searches.

[0077] S5-3. Location Update.

[0078] During the iteration process, individual whales update their parameters based on their current optimal position. During updates, the system switches between prey encirclement mode, bubble net predation mode, and random search mode based on random numbers and a non-linear convergence factor to search the PID parameter space, while simultaneously introducing an adaptive inertia weighting factor. ;

[0079] The adaptive inertia weight factor According to the spectral radiation intensity error term and channel coupling suppression term The normalized result is adjusted, and its expression is: in, Let be the adaptive inertia weight for the t-th iteration; and These are the minimum and maximum values ​​of the inertia weight, respectively. and These are the normalized values ​​of the spectral radiance error term and the channel coupling suppression term, respectively, at the t-th iteration.

[0080] When the spectral radiance error term or the channel coupling suppression term is large Increase the value to enhance the whale's global search capability in the PID parameter space; when the spectral radiance error term and the channel coupling suppression term are decreased, This is reduced to enhance local search capabilities. Consequently, the PID parameter search process can be adjusted based on the radiation output deviation of the infrared radiation simulation source and the coupling state of the dual temperature channels.

[0081] S5-4. Local perturbation, boundary checking and optimal individual update.

[0082] After each iteration, the current best individual is evaluated. - The greedy perturbation operation generates candidate new solutions near the current optimal solution, and its update formula is: in, It is a candidate new solution generated in the t-th iteration. This represents the current optimal position for an individual whale. The location of a randomly selected individual whale. A random number within the interval [0, 1] The disturbance amplitude coefficient, The perturbation probability is denoted as . After generating candidate solutions, boundary checks are performed on the PID parameters represented by the individual whales; for parameters exceeding the search range, they are adjusted to the corresponding boundary range. Subsequently, whales satisfying the boundary conditions are substituted into the system for simulation, their fitness values ​​are recalculated within a set simulation time, and the current optimal individual position is updated using a greedy strategy.

[0083] S5-5. Iteration Termination

[0084] The process of position update, local perturbation, boundary check, and fitness calculation is repeated until the set maximum number of iterations is reached. After the iteration is completed, the PID parameter combination corresponding to the whale individual with the best fitness value is output as the optimal PID parameter set.

[0085] By using the above-mentioned comprehensive objective optimization function and constraint optimization solution process, temperature tracking accuracy, output spectral radiation intensity control accuracy, control quantity smoothness, channel coupling effect, and system operation constraints can be uniformly incorporated into the PID parameter optimization process.

[0086] Traditional PID parameter tuning typically relies primarily on temperature error. However, for a dual-temperature coupled system using an infrared radiation simulation source, tuning parameters solely based on temperature error fails to simultaneously reflect output spectral radiation intensity deviation, the coupling effect of the two temperature channels, abrupt changes in control input, and system operational boundary constraints. To address this, this embodiment incorporates a spectral radiation intensity error term, a control input change smoothing term, a channel coupling suppression term, and a constraint penalty term into the comprehensive objective optimization function. This allows the PID parameter solution to move beyond simply focusing on temperature tracking, simultaneously optimizing for radiation output accuracy, channel coordination, control input stability, and operational safety.

[0087] Among them, the spectral radiation intensity error term is used to directly introduce the output radiation target into the parameter solution process, so that the temperature adjustment of the silicon molybdenum rod corresponds to the radiation output of the specified band; the channel coupling suppression term is used to suppress the situation where the two temperature channels deviate from the set state at the same time; the control quantity change smoothing term is used to limit the sudden changes in drive current and cooling system operating power; and the constraint penalty term is used to limit the temperature, control quantity, and control quantity change amplitude from exceeding the limit.

[0088] During the parameter search process, the adaptive inertial weights adjust the search capability based on the spectral radiation intensity error term and the channel coupling suppression term. When the radiation error or coupling effect is large, the global search is enhanced; when the error decreases, the local search is enhanced. Simultaneously, by combining ε-greedy perturbation and constraint judgment, the candidate parameters not only possess superior fitness but also satisfy constraints such as the temperature of the silicon molybdenum rod, the cavity temperature, the drive current, the operating power of the cooling system, and the amplitude of control quantity changes. The resulting PID parameters are more suitable for the dual-temperature coupled system of the infrared radiation simulation source, balancing radiation intensity control accuracy, dual-temperature channel coordination, control quantity stability, and system operational safety.

[0089] Step 6: Update the optimal PID parameters and implement closed-loop control of spectral radiance.

[0090] The optimal PID parameters obtained in step five are then updated to the two incremental PID controllers in the silicon molybdenum rod temperature control channel and the cavity temperature control channel, respectively, so that the parameter solution results directly affect the decoupled control structure of the dual-temperature inverse system constructed in step three. During system operation, the temperatures of the silicon molybdenum rod and the cavity are acquired in real time and compared with their corresponding setpoints to obtain two temperature control deviation signals. The two incremental PID controllers calculate the control adjustment amount based on their respective deviation signals and use it as the input signal to the BP neural network inverse system.

[0091] The BP neural network inverse system performs decoupled calculations on the input signal, outputting the drive current and cooling system operating power. The drive current is used to regulate the temperature of the silicon molybdenum rod, and the cooling system operating power is used to regulate the cavity temperature. After these two control quantities are applied to the temperature control coupling system, the system outputs the silicon molybdenum rod temperature and cavity temperature, which are then fed back to the corresponding control channels, forming a dual-temperature closed-loop control.

[0092] Since step one has established the mapping relationship between the temperature of the silicon molybdenum rod and the output spectral radiant intensity, the system can estimate the current output spectral radiant intensity based on temperature feedback. After updating the optimal PID parameters obtained in step five to the two incremental PID controllers, the parameter optimization results can be directly applied to the decoupled control structure of the dual-temperature inverse system. The system ensures radiation source temperature tracking through silicon molybdenum rod temperature feedback, maintains thermal environment stability through cavity temperature feedback, and outputs drive current and cooling system operating power through the BP neural network inverse system, thereby realizing closed-loop control of the output spectral radiant intensity of the infrared radiation simulation source based on temperature feedback and radiant intensity mapping estimation.

[0093] Furthermore, to verify the effectiveness of the method of the present invention, the control method is implemented and tested through a simulation platform in this embodiment.

[0094] A simulation model of an infrared radiation simulation source temperature control system was built in the Simulink environment. The simulation model includes a temperature control coupled system module, a BP neural network inverse system module, a dual temperature control channel module, and a constraint optimization parameter solving module.

[0095] During the simulation, the temperature setpoint of the silicon molybdenum rod is first calculated based on the target spectral radiation intensity using the mapping relationship established in step one, and a dual-temperature control target is constructed by combining the cavity temperature setpoint. The system collects the temperature of the silicon molybdenum rod and the cavity temperature in real time, and uses them as feedback signals to input into the incremental PID controller. After decoupling calculation by the BP neural network inverse system, the driving current and the operating power of the cooling system are obtained, thereby realizing closed-loop control.

[0096] To illustrate the control effect of the method of the present invention, the method of the present invention is compared with the traditional PID control method, and the temperature changes and output spectral radiation intensity changes during the system response process are recorded and analyzed.

[0097] (1) Temperature control effect of silicon molybdenum rod.

[0098] like Figure 7 As shown in (a), this is the temperature response curve of the silicon molybdenum rod under the traditional PID control method; Figure 7 As shown in (b) of the figure, this is the temperature control effect curve of the silicon molybdenum rod under the method of the present invention.

[0099] As can be seen from the figure, the traditional PID control method has a certain overshoot during the adjustment process, accompanied by oscillation, and the system takes a long time to stabilize; while the method of the present invention can achieve a fast response, and the temperature change process is smooth without obvious oscillation.

[0100] To further quantify the control performance of the two methods, statistical analysis was performed on key dynamic indicators, and the results are shown in Table 1.

[0101] Table 1. Comparison of Temperature Control Performance of Silicon Molybdenum Rods: rise time / s Approximately 2.4 seconds Approximately 0.5s Overshoot / % Approximately 9% Approximately 0% oscillation exist No obvious oscillation Settling time / s Approximately 15 seconds Approximately 1~2 seconds

[0102] As can be seen from Table 1, under the simulation conditions of this embodiment, the method of the present invention exhibits better control performance in terms of response speed, overshoot, and stability.

[0103] (2) The effect of cavity temperature control.

[0104] like Figure 8 As shown in (a), this is the control response curve of the cavity temperature under the traditional control method; Figure 8 As shown in (b) of the figure, this is the temperature control effect curve of the cavity under the method of the present invention.

[0105] As can be seen from the figure, the traditional method has certain fluctuations during the adjustment process and the response process is relatively slow; while the method of the present invention has a more stable system response and significantly reduced temperature fluctuations.

[0106] To further compare control performance, a statistical analysis was conducted on key indicators of cavity temperature, and the results are shown in Table 2.

[0107] Table 2 Comparison of Cavity Temperature Control Performance: rise time / s Approximately 4~5 seconds Approximately 0.8s Overshoot / % Approximately 17% Approximately 0% oscillation exist No obvious oscillation Settling time / s Approximately 20 seconds Approximately 1~2 seconds

[0108] As can be seen from Table 2, the method of the present invention effectively shortens the response time and reduces the temperature fluctuation range while ensuring system stability.

[0109] (3) Overall effect description

[0110] In summary, the method of the present invention can effectively reduce the coupling effect between the two temperature channels and improve the control accuracy and stability of the output spectral radiation intensity of the infrared radiation simulation source.

Claims

1. A method for controlling infrared radiation intensity based on constraint optimization and inverse system decoupling, characterized in that, Includes the following steps: S1: Based on the blackbody radiation theory and introducing the radiation source spectral emissivity correction, a mapping relationship is established between the output spectral radiation intensity of the infrared radiation simulation source and the temperature of the silicon molybdenum rod, transforming radiation intensity control into temperature control. S2: Based on the mapping relationship established in S1, and combined with the coupling relationship between the temperature of the silicon molybdenum rod and the temperature of the cavity in the infrared radiation simulation source temperature control system, construct the inverse system model of the temperature control coupling system; learn the inverse system characteristics of the infrared radiation simulation source temperature control coupling system to obtain the BP neural network inverse system model; S3: Connect the BP neural network inverse system model obtained in S2 in series with the temperature control coupling system to construct an equivalent decoupled dual temperature control channel, and set an incremental PID controller in each control channel to form an inverse system decoupled control structure. S4: Based on the inverse system decoupling control structure described in S3, a comprehensive objective optimization function is established with the output spectral radiance intensity error as the core evaluation term. The comprehensive objective optimization function includes a temperature tracking error term, a spectral radiance intensity error term calculated from the temperature-spectral radiance intensity mapping relationship, a control quantity change smoothing term, a channel coupling suppression term, and a constraint penalty term. S5: Based on the comprehensive objective optimization function and combined with the operating constraints of silicon molybdenum rod temperature, cavity temperature, driving current, and cooling system operating power, an improved whale optimization algorithm is used to solve for the parameters of each incremental PID controller; the improved whale optimization algorithm generates an initial population through a Logistic-Tent hybrid chaotic mapping and introduces adaptive inertia weights during the iterative search process. - Greedy perturbation strategy to obtain the optimal PID parameters that satisfy the system constraints; S6: Update the optimal PID parameters obtained in S5 to the corresponding incremental PID controller online; during system operation, collect the temperature of the silicon molybdenum rod and the cavity temperature in real time, and calculate the estimated value of the current output spectral radiance based on the temperature-spectral radiance mapping relationship. The incremental PID controller outputs the control adjustment amount based on the temperature deviation signal. After decoupling through the inverse system of the BP neural network, it outputs the silicon molybdenum rod drive current and the cooling system operating power, realizing closed-loop control of the output spectral radiation intensity of the infrared radiation simulation source based on temperature feedback and radiation intensity mapping.

2. The infrared radiation intensity control method based on constraint optimization and inverse system decoupling according to claim 1, characterized in that, The infrared radiation simulation source described in S1 uses a single silicon molybdenum rod as the infrared radiation simulation source, with a radiation spectrum range of 1.5μm to 5.5μm and a normal operating temperature range of 1500K to 2000K. The mapping relationship between the output spectral radiance of the infrared radiation simulation source and the temperature of the silicon molybdenum rod, as described in S1, is specifically established using a standard blackbody radiation source as the calculation benchmark. In a three-dimensional coordinate system, the surface area, spatial angle, radiant power, radiant intensity, and spectral radiant intensity are derived. Based on Planck's radiation law, a mapping relationship between spectral radiant intensity and temperature within a specified wavelength range is established. Furthermore, the Stefan-Boltzmann law is used to explain the overall trend of radiant output variation with temperature, establishing a mapping relationship for a temperature of T and a radiation wavelength range of [missing information]. The spectral radiance response relationship is as follows: in, For wavelength variables, For a infinitesimal element of wavelength, The first radiation constant, The second radiation constant, The operating temperature of the silicon molybdenum rod. The temperature of the radiation source is Spectral emissivity at that time The energy utilization coefficient of the infrared radiation simulation source device. These are the lower and upper limits of the selected band, respectively. For band range The spectral radiance, in units of S is the radiation area of ​​a single silicon molybdenum rod radiation source, and θ is the radiation field of view. The spectral radiation intensity output by the infrared radiation simulation source device is controlled by adjusting the temperature of the silicon molybdenum rod.

3. The infrared radiation intensity control method based on constraint optimization and inverse system decoupling according to claim 1, characterized in that, S2 includes the following steps: S2-1. Construct a temperature control coupling system for the infrared radiation simulation source device. The system output of the temperature control coupling system is the temperature of the silicon molybdenum rod. and cavity temperature The system input is the drive current of the silicon molybdenum rod controlled by the host computer. and cooling system operating power Ambient temperature Introduced as a confounding variable; S2-2. Collect operating data of the infrared radiation simulation source device under different working conditions according to the sampling period. The operating data includes silicon molybdenum rod temperature, cavity temperature, ambient temperature, silicon molybdenum rod drive current, and cooling system operating power. Obtain the silicon molybdenum rod temperature control channel adjustment amount based on the temperature deviation signal. Adjustment amount of cavity temperature control channel ; S2-3. Establish a BP neural network model, which includes an input layer, a hidden layer, and an output layer. The input layer variables include the temperature control channel adjustment amount of the silicon molybdenum rod obtained in step S2-2. Cavity temperature control channel adjustment amount and ambient temperature The output layer variable is the silicon molybdenum rod drive current. and cooling system operating power ; S2-4. Construct a training sample set using the operational data collected in step S2-2, and adjust the temperature control channel of the silicon molybdenum rod according to the data obtained in step S2-2. Cavity temperature control channel adjustment amount Adjust the temperature control channel of the silicon molybdenum rod. Cavity temperature control channel adjustment amount and ambient temperature As input, the silicon molybdenum rod drive current and cooling system operating power As output, the BP neural network model is trained to learn the adjustment amount of the temperature control channel for the silicon molybdenum rod. Cavity temperature control channel adjustment amount and ambient temperature The nonlinear inverse mapping relationship between the temperature control coupling system input described in step S2-1 is used to obtain the BP neural network inverse system model.

4. The infrared radiation intensity control method based on constraint optimization and inverse system decoupling according to claim 1, characterized in that, S3 includes the following steps: S3-1. Connect the BP neural network inverse system model trained in S2 in series with the temperature control coupling system to construct an equivalent decoupled dual temperature control structure. The output of the BP neural network inverse system model corresponds to the driving current of the silicon molybdenum rod controlled by the host computer. and cooling system operating power And serve as the input to the temperature control coupling system; S3-2. The temperature of the silicon molybdenum rod is used as a reference. Deviation from its set value and cavity temperature The deviation between the set value and the set value is used as the input of two incremental PID controllers to construct a dual temperature control channel, which includes a silicon molybdenum rod temperature control channel and a cavity temperature control channel. S3-3. The outputs of the two incremental PID controllers are used as the control channel adjustment inputs to the BP neural network inverse system model, where the temperature of the silicon molybdenum rod is... The incremental PID controller in the control channel outputs the temperature control channel adjustment amount for the silicon molybdenum rod, which is the cavity temperature. The incremental PID controller in the control channel outputs the adjustment amount of the cavity temperature control channel; the BP neural network inverse system model is based on the adjustment amount of the silicon molybdenum rod temperature control channel. Cavity temperature control channel adjustment amount and ambient temperature After decoupling calculations, the output silicon molybdenum rod drive current is determined. and cooling system operating power .

5. The infrared radiation intensity control method based on constraint optimization and inverse system decoupling according to claim 1, characterized in that, S4 includes the following steps: S4-1. Based on the dual temperature control channels constructed in S3, establish a comprehensive objective optimization function. ; S4-2. The comprehensive objective optimization function Temperature tracking error term Spectral radiation intensity error term Smoothing term of control quantity change Channel coupling suppression term and constraints and penalties Its composition, its expression is: in, These are weighting coefficients, all of which are non-negative, and The temperature tracking error term Used to characterize the temperature of silicon molybdenum rods and cavity temperature Tracking accuracy for each set value; the spectral radiance error term Used to characterize the temperature of the silicon molybdenum rod The deviation between the actual spectral radiance intensity and the target spectral radiance intensity calculated through the mapping relationship between spectral radiance intensity and temperature; the smoothing term of the control quantity change. Used to limit drive current and cooling system operating power The magnitude of the change; the channel coupling suppression term Used to characterize the cross-coupling effect between the two temperature control channels; the constraint penalty term Used for temperature control of silicon molybdenum rods Cavity temperature Drive current Cooling system operating power In addition, constraints are imposed on the range of change of control quantities. When any variable exceeds the preset safety range or the range of change exceeds the preset threshold, a corresponding penalty is introduced. Before calculating the comprehensive objective optimization function, the temperature tracking error term is... Spectral radiation intensity error term Smoothing term of control quantity change Channel coupling suppression term and constraints and penalties Each evaluation item is normalized to convert it into a dimensionless quantity before being weighted and summed.

6. The infrared radiation intensity control method based on constraint optimization and inverse system decoupling according to claim 1, characterized in that, S5 includes the following steps: S5-1. Construct the parameter vector to be optimized based on the proportional coefficient, integral coefficient, and derivative coefficient of the two incremental PID controllers in the silicon molybdenum rod temperature control channel and the cavity temperature control channel. : in, These are the proportional, integral, and derivative coefficients of the temperature control channel for the silicon molybdenum rod; These are the proportional, integral, and derivative coefficients of the cavity temperature control channel, respectively. The search space is determined based on the allowable range of each parameter, and the feasible region of the parameters is determined by combining the constraints of silicon molybdenum rod temperature, cavity temperature, driving current and cooling system operating power; within the feasible region of the parameters, the initial whale population is generated by using Logistic-Tent hybrid chaotic mapping. S5-2. Substitute the PID parameters corresponding to each individual whale in the initial population into the comprehensive objective optimization function. In this process, the individual fitness value is calculated by combining the response results of the temperature control coupling system, and the individual with the best fitness is selected as the current optimal solution; S5-3. During the iteration process, the position of the individual whale is updated based on the current optimal solution to search for a better combination of PID parameters; different search methods are selected based on random numbers and nonlinear convergence factors during the position update process, while an adaptive inertia weight factor is introduced. ; The adaptive inertia weight factor According to the spectral radiation intensity error term and channel coupling suppression term The normalized result is adjusted, and its expression is: in, Let be the adaptive inertia weight for the t-th iteration; and These are the minimum and maximum values ​​of the inertia weight, respectively. and These are the normalized values ​​of the spectral radiance error term and the channel coupling suppression term, respectively, at the t-th iteration. S5-4. After each iteration, introduce the current best individual... - The greedy perturbation operation generates candidate new solutions near the current optimal solution, and its perturbation update formula is: in, It is a candidate new solution generated in the t-th iteration. This represents the current optimal position for an individual whale. The location of a randomly selected individual whale. A random number within the interval [0, 1] The disturbance amplitude coefficient, Let be the probability of perturbation, and By using the above perturbation method, new candidate solutions are generated near the current optimal solution, while retaining the ability to search in the direction of random individuals, so as to reduce the probability of the algorithm getting stuck in local optima; The system response is calculated based on the PID parameters corresponding to the candidate solution, and constraints are determined. When the silicon molybdenum rod temperature, cavity temperature, drive current, cooling system operating power, and control quantity change amplitude corresponding to the candidate solution all meet the preset constraints, the fitness of the candidate solution is compared with the current optimal solution. When the comprehensive objective function value corresponding to the candidate solution is less than the comprehensive objective function value corresponding to the current optimal solution, the candidate solution replaces the current optimal solution; otherwise, the current optimal solution remains unchanged. S5-5. Repeat the whale individual position update and fitness calculation. - The greedy disturbance and system constraint determination process continues until the maximum number of iterations is reached or the fitness function meets the preset accuracy requirements. The PID parameter combination corresponding to the whale individual with the best fitness is then output as the optimal control parameter.

7. The infrared radiation intensity control method based on constraint optimization and inverse system decoupling according to claim 1, characterized in that, In step S6, the optimal PID parameters obtained in step S5 are updated to the two incremental PID controllers in the dual temperature control channel; during system operation, the temperature of the silicon molybdenum rod is collected in real time. and cavity temperature The temperature deviation signal is obtained by comparing it with the corresponding set value, and then input into the BP neural network inverse system model after being processed by the corresponding incremental PID controller; the BP neural network inverse system model outputs the driving current. and cooling system operating power To adjust the temperature of the silicon molybdenum rod and cavity temperature This enables closed-loop control of the output spectral radiance intensity of an infrared radiation simulation source based on temperature feedback and radiance intensity mapping estimation.

Images