Integrated control platform of programmable multi-mode environmental test box

Abstract

The invention relates to the technical field of programmable multi-mode integrated control, in particular to an integrated control platform of a programmable multi-mode environmental test chamber, which comprises a multi-mode sensing unit, a dynamic logic reconstruction unit, a virtual verification sandbox unit and a feedback optimization unit, the multi-modal sensing unit synchronously collects parameters such as temperature and humidity to generate environment data streams, the dynamic logic reconstruction unit analyzes parameter interaction by using a long-short-term memory neural network in combination with a double-attention mechanism to generate an optimal control logic chain, the virtual verification sandbox unit constructs digital twin pre-execution control logic, and the virtual verification sandbox unit constructs digital twin pre-execution control logic. The physical actuator is activated after the error rate is ensured to be lower than a preset threshold value, the feedback optimization unit dynamically adjusts the function block selection weight according to the actual response time delay and the energy consumption deviation, and a high-precision and low-energy-consumption intelligent control scheme is provided for the reliability test of high-end products in the fields of aerospace, electronic components and the like.

Description

Technical Field

[0001] This invention relates to the field of programmable multi-mode integrated control technology, and more specifically, to an integrated control platform for a programmable multi-mode environmental test chamber. Background Technology

[0002] Programmable multi-mode integrated control is an important technology. In high-end manufacturing fields such as aerospace and electronic components, environmental test chambers are core equipment for verifying the environmental adaptability and reliability of products by simulating complex environments such as temperature, humidity and vibration. With the increasing demand for intelligent testing, multi-mode dynamic environment simulation has put forward higher requirements for the parameter coordination accuracy, response speed and energy consumption management of the control platform.

[0003] However, traditional integrated control technology suffers from core problems such as lag in multi-parameter coupling control and low efficiency in logic reconfiguration. Existing solutions use pre-set PID control module combinations, lacking dynamic analysis of the interaction between parameters such as temperature, humidity, and vibration. When the temperature change slope exceeds 5℃ / min, the humidity compensation logic is not adjusted synchronously, leading to increased humidity overshoot and distorted test data. Fixed function block combinations do not label power consumption and response accuracy metadata. Under high-frequency conditions with vibration main frequency exceeding 100Hz, low-precision control modules are still used, increasing the error caused by vibration amplitude and exceeding the test standard requirements. In addition, traditional systems do not establish a virtual verification mechanism. When physical actuators are directly loaded with control logic, the risk of actuator damage due to parameter conflicts increases. This rigid control logic and lack of verification mechanism ultimately result in low success rate of complex environment simulation on traditional platforms, higher energy consumption than advanced levels, and difficulty in meeting the reliability testing requirements of high-end products. To solve this technical problem, we provide an integrated control platform for a programmable multi-mode environmental test chamber. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated control platform for a programmable multi-mode environmental test chamber to solve the problems mentioned in the background art.

[0005] 1. Due to the lag in multi-parameter coupling control in traditional solutions, test data is prone to distortion. Therefore, this case uses the dual attention mechanism of the dynamic logic reconstruction unit to analyze the interaction of parameters, generate the optimal control logic chain, and improve data accuracy.

[0006] 2. Due to the low efficiency of traditional system logic refactoring and the high risk of actuator damage, this case uses a virtual verification sandbox unit to pre-execute the control logic and dynamically replace the function block that causes overshoot, thereby improving system reliability.

[0007] To achieve the above objectives, an integrated control platform for a programmable multi-mode environmental test chamber is provided, comprising the following units:

[0008] The multimodal sensing unit synchronously collects temperature, humidity, vibration, and light intensity parameters through a sensor array, generating a timestamped environmental data stream;

[0009] The dynamic logic reconfiguration unit analyzes the environmental data flow based on the long short-term memory neural network, outputs the multi-parameter composite fluctuation pattern for future time periods, and stores the pre-compiled function block set of the programmable logic controller through the logic unit library. Each function block is labeled with power consumption level and response accuracy metadata. According to the fluctuation pattern, the function block combination is dynamically selected from the logic unit library to generate the optimal control logic chain that meets the preset constraints.

[0010] The virtual verification sandbox unit loads the optimal control logic chain in an independent container for pre-execution. When the error rate is less than 0.1%, the physical actuator is activated, and the actual response delay and energy consumption deviation of the physical actuator are fed back to the dynamic logic reconstruction unit through the feedback optimization unit to dynamically adjust the function block selection weight.

[0011] As a further improvement to this technical solution, the dynamic logic reconfiguration unit performs the following steps:

[0012] The environmental data stream is spatiotemporally aligned to eliminate differences in sampling frequencies among multiple sensors;

[0013] The aligned data is input into a long short-term memory neural network, which includes a dual attention mechanism, wherein a time attention layer captures long-period trends and a parameter-coupled attention layer resolves the interaction weights of temperature, humidity and vibration.

[0014] Output a multi-parameter composite fluctuation pattern for future time periods. The multi-parameter composite fluctuation pattern is represented in the form of a three-dimensional tensor, with dimensions including a time axis, a parameter type axis, and a fluctuation intensity axis.

[0015] As a further improvement to this technical solution, the method for constructing the logic unit library specifically includes:

[0016] The pre-compiled function block set is classified according to the control scenario into basic control set, security protection set and composite response set. The metadata labeled for each function block includes power consumption level, response accuracy and conflict label. The power consumption level is the power consumption range under actual full load operation, the response accuracy is the steady-state error range under standard step input, and the conflict label is a list of other function block IDs that are prohibited from being activated at the same time as this function block.

[0017] As a further improvement to this technical solution, when the dynamic logic reconfiguration unit generates the optimal control logic chain:

[0018] Based on the three-dimensional tensor fluctuation pattern, key control constraints are extracted. The key control constraints are: if the fluctuation intensity axis shows a temperature change slope greater than or equal to 5℃ / min, then a PID function block with a response accuracy less than or equal to 0.5% is forcibly selected; if the output vibration-temperature interaction weight of the parameter coupling attention layer is greater than 0.7, then the coupling compensator in the composite response set is activated.

[0019] A Pareto optimal solution is selected from non-conflicting functional block combinations using a multi-objective optimization algorithm, provided that power consumption ≤ threshold and accuracy ≥ threshold.

[0020] As a further improvement to this technical solution, the conflict resolution mechanism of the functional block combination includes:

[0021] When the conflict labels carried by the selected function blocks overlap, the dynamic priority arbitrator is activated. The working principle of the dynamic priority arbitrator is as follows:

[0022] If there is a conflict between security protection function blocks, the function block with the higher security level will be retained, and the security level will be classified according to the preset security integrity level standard; otherwise, the function block with the largest response time delay will be discarded.

[0023] As a further improvement to this technical solution, the fitness function construction method of the multi-objective optimization algorithm is as follows:

[0024] Real-time analysis of the high-frequency component energy ratio and the number of gradient abrupt change points in the three-dimensional tensor fluctuation mode, dynamically allocating power consumption weight coefficients, accuracy weight coefficients, and delay weight coefficients, specifically satisfying:

[0025] When a temperature fluctuation slope greater than or equal to 8℃ / min and a vibration frequency greater than 100Hz are detected, the accuracy weighting coefficient is increased to 0.7 and the delay weighting coefficient is decreased to 0.1.

[0026] When in a constant temperature steady state, the dynamic power consumption weighting coefficient is set to 0.8 and the low power function block combination strategy is activated.

[0027] As a further improvement to this technical solution, the dynamically allocated power consumption weighting coefficient specifically includes:

[0028] The random forest regression model is trained based on historical experimental data. The model takes the tensor feature value of the fluctuation mode as input and outputs the optimized ratio of power consumption weight coefficient, accuracy weight coefficient and delay weight coefficient. The model parameters are corrected online by the energy consumption deviation data of the feedback optimization unit.

[0029] As a further improvement to this technical solution, the method for pre-executing the virtual verification sandbox unit in an independent container is as follows:

[0030] A digital twin with the same electromechanical characteristics as the physical actuator is constructed. The initial values ​​of the current environmental parameters and the optimal control logic chain are synchronously injected into the twin. The simulation is run at 10 times the speed, and the logic conflict flag, actuator torque overshoot and instantaneous power consumption curve are monitored in real time. The error rate is determined to be up to standard and the physical actuator is activated only when the conflict flag is 0, the overshoot is less than or equal to 4%, and the peak power consumption is less than 115% of the rated power for three consecutive simulation cycles.

[0031] As a further improvement to this technical solution, the exception handling mechanism of the virtual verification sandbox unit includes:

[0032] If an overshoot of more than 4% is detected, the control logic chain is automatically reconstructed within the container. Based on the real-time status data of the digital twin, the function block that caused the overshoot is forcibly replaced and re-verified. The replacement rule is to call a function block of the same type with a response accuracy improved by at least two levels from the logic unit library.

[0033] As a further improvement to this technical solution, the operation method of the feedback optimization unit is as follows:

[0034] The actual response delay and energy consumption deviation of the physical actuator are captured by a high-speed data bus, and mapped to the metadata tags of each functional block in the logic unit library to establish a formula for updating the functional block selection weight:

[0035] Among them, T actual The actual response time is given by ΔE, where ΔE is the energy consumption deviation, k is the learning rate factor, and T is the actual response time. indeal To predict the target time delay required for the fluctuation pattern, Select weights for the updated function blocks. Weights are selected for the original function blocks, where e is a constant. After each trial cycle, the weights of all function blocks are recalculated and the priority index of the logic unit library is updated.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] In an integrated control platform for a programmable multi-mode environmental test chamber, a dynamic logic reconfiguration unit captures the long-term trends of environmental parameters and the interaction weights of temperature, humidity, and vibration parameters through a long short-term memory neural network combined with a dual attention mechanism. This outputs the multi-parameter composite fluctuation mode in the form of a three-dimensional tensor, improving the accuracy of parameter coupling analysis and effectively solving the problem of insufficient analysis of parameter interaction in traditional solutions. Based on a logic unit library, it dynamically selects functional block combinations, labels power consumption levels and response accuracy metadata, and combines multi-objective optimization algorithms and conflict resolution mechanisms. When the temperature change slope is greater than or equal to 5℃ / min, it forces the selection of a high-precision PID functional block, satisfying the constraints of power consumption ≤ threshold and accuracy ≥ threshold, generating the optimal control logic chain. This improves the efficiency of control logic reconfiguration, solves the problems of rigidity and high energy consumption in traditional fixed functional block combination logic, and enhances control accuracy and adaptability under complex operating conditions. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating the overall workflow of the present invention;

[0039] Figure 2 This is a system architecture diagram of the present invention.

[0040] The meanings of the labels in the diagram are as follows:

[0041] 1. Multimodal perception unit; 2. Dynamic logic reconstruction unit; 3. Virtual verification sandbox unit; 4. Feedback optimization unit. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] This invention provides an integrated control platform for a programmable multi-mode environmental test chamber. Please refer to [link / reference]. Figure 1 As shown, the following units:

[0044] The multimodal sensing unit 1 synchronously collects temperature, humidity, vibration and light intensity parameters through a sensor array to generate a timestamped environmental data stream;

[0045] The dynamic logic reconfiguration unit 2 analyzes the environmental data flow based on the long short-term memory neural network, outputs the multi-parameter composite fluctuation pattern for future periods, and stores the pre-compiled function block set of the programmable logic controller through the logic unit library. Each function block is labeled with power consumption level and response accuracy metadata. According to the fluctuation pattern, the function block combination is dynamically selected from the logic unit library to generate the optimal control logic chain that meets the preset constraints.

[0046] After the multimodal sensing unit 1 generates the environmental data stream, in order to accurately analyze the dynamic change trend of environmental parameters, the dynamic logic reconstruction unit 2 needs to perform spatiotemporal alignment on the data and analyze it through a long short-term memory neural network. The specific implementation method is as follows:

[0047] Dynamic logic reconfiguration unit 2 performs the following steps:

[0048] Spatiotemporal alignment processing is performed on the environmental data stream to eliminate the sampling frequency differences among multiple sensors. Since the sampling frequencies of temperature, humidity, and vibration sensors are different, linear interpolation is first used to upsample low-frequency data, and then all sensor data are unified to the highest sampling frequency to ensure consistency in the time dimension. For example, the temperature sensor collects data once every 100ms, and in the 10ms sampling interval of the vibration sensor, nine intermediate temperature values ​​are generated by linear interpolation to align the data of different sensors on the time axis. At the same time, hardware timestamps are used to correct the clock deviation of each sensor. When the timestamp difference exceeds 50μs, the Kalman filter algorithm is used to predict and compensate for missing data, eliminating the analysis error caused by spatiotemporal asynchrony and laying the foundation for subsequent analysis.

[0049] The aligned data is input into a Long Short-Term Memory (LSTM) neural network, which contains 128 memory units and can effectively capture long-term dependencies of environmental parameters. The temporal attention layer uses a self-attention mechanism to calculate weights for data from the past 1000 time steps, highlighting long-term trends, such as hourly temperature changes. For example, in a scenario of gradual temperature increase, the temporal attention layer assigns a weight of 0.7 to data from the last 3 seconds, weakening the influence of earlier data. The parameter coupling attention layer analyzes the interaction weights of temperature, humidity, and vibration, and constructs a parameter correlation matrix by calculating mutual information coefficients. For example, when the temperature rises sharply by 5°C / min, the mutual information coefficient between humidity and temperature reaches 0.8, and the mutual information coefficient between vibration and temperature reaches 0.6. This layer outputs a three-dimensional weight tensor, clarifying the coupling strength between parameters and improving analytical accuracy.

[0050] The neural network outputs a multi-parameter composite fluctuation pattern for the next 5 minutes, represented by a three-dimensional tensor: time axis × parameter type axis × fluctuation intensity axis. The time axis has a resolution of 1 second and a total of 300 time steps. The parameter type axis includes four dimensions: temperature, humidity, vibration, and illumination. The fluctuation intensity axis maps the parameter change amplitude to the [0, 1] interval through normalization. For example, the temperature fluctuation intensity is 0.9 in the 100th second, corresponding to a sudden change slope of 8℃ / min; the humidity fluctuation intensity is 0.5, corresponding to a change rate of 2%RH / min; and the vibration fluctuation intensity is 0.7, corresponding to a dominant frequency of 120Hz. This tensor comprehensively describes the future fluctuation characteristics of multiple parameters, providing a precise basis for the generation of control logic. Through the above spatiotemporal alignment and dual attention mechanism analysis, the dynamic logic reconstruction unit 2 achieves high-precision processing of environmental data streams and prediction of future fluctuation patterns, providing scientific data support for the subsequent selection of functional block combinations and effectively solving the problem of test data distortion caused by traditional control lag.

[0051] To provide a structured control module for the dynamic logic reconfiguration unit 2, the logic unit library needs to pre-compile functional blocks according to scenarios and annotate metadata. The specific implementation method is as follows:

[0052] The specific methods for constructing the logic unit library include:

[0053] Based on the typical operating conditions of the environmental test chamber, the pre-compiled function block set is classified into a basic control set, a safety protection set, and a composite response set. The basic control set includes, but is not limited to, temperature PID control and humidity proportional control. For example, the temperature PID control function block uses an incremental algorithm with a sampling period of 100ms, suitable for conventional temperature and humidity control scenarios. The safety protection set includes, but is not limited to, over-temperature emergency stop and vibration overload protection. The over-temperature emergency stop function block sets the temperature threshold to 110% of the upper limit and cuts off the heating power supply within 0.5 seconds upon triggering. The composite response set integrates temperature and humidity coupling compensation and vibration-temperature coordinated control. For example, the temperature and humidity coupling compensation function block is based on a linear model of humidity changing with temperature, with a humidity change rate of 0.5%RH / ℃. It synchronously adjusts the humidification amount when the temperature fluctuates, improving the function block retrieval efficiency and facilitating rapid matching of operating conditions by the dynamic logic reconfiguration unit 2. Each function block is labeled with three types of metadata: power consumption level, response accuracy, and conflict tag. The power consumption level is determined through full-load operation testing. For instance, the temperature PID control function block, with a measured power consumption range of 9.5-10.3kW at a heating power of 10kW, is labeled as a high power consumption level. The response accuracy is tested under a standard step input, such as humidity ratio. When the control function block has a step input of 20%RH, the steady-state error range is less than or equal to 1%RH, and the marked response accuracy is 1%. Conflict labels are set based on the control logic compatibility of the function blocks. For example, if there is a conflict between the over-temperature emergency stop function block and the heating power adjustment function block, the conflict label records the other party's ID to avoid simultaneous activation and control logic contradictions. Metadata labeling makes the compatibility and performance indicators of the function blocks clear and verifiable, providing data support for subsequent conflict resolution and multi-objective optimization. A relational database is used to store the function block set, and a three-level index structure of "scenario-function block-metadata" is established. The scenario table records the scenario IDs and descriptions of basic control, safety protection, etc. The function block table stores the code, parameters, and scenario ID of each module. The metadata table associates the function block ID with metadata such as power consumption and accuracy. Multi-table queries are realized through database views. For example, when querying the function block with a response accuracy of less than or equal to 0.5% in the "sudden temperature change" scenario, the system can filter out the PID control modules that meet the conditions, supporting the real-time calling requirements of the dynamic logic reconfiguration unit 2, improving the retrieval and loading efficiency of function blocks, and providing efficient and reliable module support for the dynamic logic reconfiguration unit 2 to generate the optimal control logic chain.

[0054] After outputting the multi-parameter composite fluctuation mode in the form of a three-dimensional tensor through a long short-term memory neural network, the dynamic logic reconstruction unit 2 needs to extract key control constraints from it and combine them with a multi-objective optimization algorithm to generate the optimal control logic chain. The specific implementation method is as follows:

[0055] When dynamic logic reconfiguration unit 2 generates the optimal control logic chain:

[0056] To ensure control accuracy under complex operating conditions, key control constraints are analyzed from the three-dimensional tensor fluctuation mode. When the fluctuation intensity axis shows a temperature change slope greater than or equal to 5℃ / min, such as when the temperature rises from 25℃ to 50℃ in 3 minutes with a slope of 8.3℃ / min, the system automatically identifies this feature through the gradient calculation module and forces the selection of a PID function block with a response accuracy less than or equal to 0.5% from the logic unit library to ensure that the temperature control accuracy meets the requirements of harsh operating conditions. If the vibration-temperature interaction weight output by the parameter coupling attention layer is greater than 0.7, it indicates that the two are strongly correlated. In this case, the vibration-temperature coupling compensator in the composite response set is activated. This compensator is based on the mutual information model of the two and adjusts the temperature control parameters in real time, effectively avoiding control lag caused by parameter coupling.

[0057] Under the constraints of power consumption ≤ threshold and accuracy ≥ threshold, a non-dominated sorting genetic algorithm with an elitist strategy is used to select the optimal solution from non-conflicting functional block combinations. First, the weight coefficients of power consumption, accuracy, and delay are dynamically adjusted based on the energy proportion of high-frequency components and the number of gradient mutation points in the three-dimensional tensor. Then, 100 functional block combinations are generated as the initial population. Through crossover and mutation operations, iterative optimization is performed to calculate the power consumption, accuracy index, and delay performance of each combination. Non-dominated solutions are retained to form the Pareto front. Finally, the comprehensive optimal solution is selected from the Pareto front based on the fuzzy membership function. Before optimization, functional block combinations are filtered by conflict labels. For example, when selecting a temperature PID control functional block, heating power adjustment functional blocks recorded in its conflict labels are automatically excluded to ensure no conflict within the combination. If functional block conflicts occur during the optimization process, such as conflicts between the safety protection set and the basic control set functional blocks, a dynamic priority arbitrator is activated.

[0058] If safety protection function blocks are involved, modules with higher safety integrity levels are retained according to the safety integrity level standard, such as over-temperature emergency stop level which is higher than conventional temperature control. Otherwise, the function block with the largest response time delay is discarded. This avoids control logic failure caused by function block conflicts, ensures the safety and stability of system operation, improves the environmental test chamber's adaptability to complex dynamic environments, and provides precise control assurance for high-end product reliability testing.

[0059] The fitness function of a multi-objective optimization algorithm is constructed as follows:

[0060] Real-time analysis of the energy proportion of high-frequency components and the number of gradient abrupt changes in the three-dimensional tensor fluctuation mode; dynamic allocation of power consumption weighting coefficients, accuracy weighting coefficients, and delay weighting coefficients, specifically satisfying:

[0061] When a temperature fluctuation slope greater than or equal to 8℃ / min and a vibration frequency greater than 100Hz are detected, the accuracy weighting coefficient is increased to 0.7 and the delay weighting coefficient is decreased to 0.1.

[0062] When in a constant temperature steady state, the dynamic power consumption weighting coefficient is set to 0.8 and the low power function block combination strategy is activated.

[0063] When the dynamic logic reconstruction unit 2 generates the optimal control logic chain, in order for the multi-objective optimization algorithm to dynamically adjust the optimization direction according to different working conditions, its fitness function needs to be constructed in real time based on the three-dimensional tensor fluctuation mode. The specific implementation method is as follows:

[0064] Dynamically allocate power consumption weighting coefficients, specifically including:

[0065] To adapt the optimization algorithm to different environmental fluctuations, the high-frequency component energy ratio in the 3D tensor fluctuation mode is first analyzed using Fast Fourier Transform (FFT). The high-frequency component energy ratio is the proportion of energy with a frequency greater than 10Hz to the total energy. The number of gradient abrupt change points, i.e., points where the parameter change rate exceeds a threshold per unit time, is also counted. When a temperature fluctuation slope greater than or equal to 8℃ / min and the dominant vibration frequency is greater than 100Hz (e.g., the temperature rises from 20℃ to 28℃ within 1 minute with a slope of 8℃ / min), and the vibration sensor detects a high-frequency vibration with a dominant frequency of 120Hz, this is determined to be a high-frequency complex operating condition. In this case, the accuracy weighting coefficient is increased to 0.7 to emphasize the importance of control accuracy, while the delay weighting coefficient is reduced to 0.1 to lower the requirement for response speed. This ensures that control accuracy is prioritized under harsh conditions, avoiding data distortion due to insufficient accuracy. Actual testing shows that improving parameter control accuracy, when the fluctuation intensity of each parameter displayed in the 3D tensor is less than 0.2 (e.g., temperature fluctuation within ±0.5℃, humidity fluctuation within ±1%RH), indicates a steady-state temperature condition. In this phase, to reduce energy consumption, the power consumption weighting coefficient is set to 0.8, and the low-power function block combination strategy in the logic unit library is activated. For example, the energy-saving PID control module and the sensor sampling strategy in sleep mode are called to solve the problem of energy waste in the steady-state stage of traditional solutions. In order to ensure that the fitness function can accurately reflect the current operating conditions, its weighting coefficient is updated every 100ms. During the update, historical operating conditions data and current three-dimensional tensor features are combined, and the weight changes are smoothed by the exponential moving average method to avoid optimization oscillations caused by sudden weight changes. For example, when transitioning from high-frequency operating conditions to the steady-state stage, the weighting coefficient is gradually adjusted by a smoothing factor of 0.9, so that the optimization algorithm can smoothly switch optimization objectives, improving the adaptability and effectiveness of the control logic chain. Through the above fitness function construction method, the multi-objective optimization algorithm can dynamically adjust the optimization focus according to the fluctuation characteristics of environmental parameters. Under high-frequency complex operating conditions, priority is given to ensuring control accuracy, and energy consumption optimization is emphasized in the isothermal steady-state stage. This enables the control logic chain to achieve optimal performance under different operating conditions, improving the testing accuracy and energy utilization efficiency of the environmental test chamber.

[0066] The virtual verification sandbox unit 3 loads the optimal control logic chain in an independent container for pre-execution. When the error rate is less than 0.1%, the physical actuator is activated. The actual response delay and energy consumption deviation of the physical actuator are fed back to the dynamic logic reconfiguration unit 2 through the feedback optimization unit 4, and the function block selection weight is dynamically adjusted.

[0067] After the dynamic logic refactoring unit 2 generates the optimal control logic chain, in order to avoid directly loading it to the physical actuator and causing failure, the virtual verification sandbox unit 3 needs to perform pre-execution verification of the control logic in an independent container. The specific implementation method is as follows:

[0068] The method for pre-executing Virtual Verification Sandbox Unit 3 in an independent container is as follows:

[0069] To ensure the accuracy of virtual verification, a digital twin is constructed based on the electromechanical characteristics of the physical actuator. Finite element analysis is used to calibrate the model parameters, ensuring that the error between the twin's dynamic response and the physical actuator is less than or equal to 3%. For example, for a heating actuator, the twin needs to accurately simulate its heat capacity and heat dissipation coefficient to ensure the accuracy of temperature control simulation. After generating the optimal control logic chain, the initial values ​​of the current environmental parameters and the control commands in the logic chain are synchronously injected. Real-time data interaction between the twin and the control logic is achieved through a data interface, guaranteeing the real-time performance of the virtual verification. To improve verification efficiency, the simulation runs at 10 times the actual operating speed in an independent container; for example, a 5-minute actual operation can be simulated in 30 seconds. The simulation is completed, and three key indicators are monitored in real time during the simulation: logic conflict flag, generated by the function block conflict detection module in the control logic chain (e.g., flag is set to 1 when heating and cooling function blocks are activated simultaneously); actuator torque overshoot, calculated using a twin motor model: overshoot = (measured torque - rated torque) / rated torque × 100%; and instantaneous power consumption curve, simulating the real-time power consumption of each actuator at a sampling frequency of 100Hz. For example, when the control logic chain includes both heating power adjustment and over-temperature emergency stop function blocks, if the temperature setpoint exceeds the over-temperature threshold, the conflict flag will be set to 1, indicating a logical contradiction. This allows for comprehensive detection of potential problems in the control logic. Verification pass conditions are set as follows:

[0070] If, for three consecutive simulation cycles, the conflict flag is 0, the overshoot is less than or equal to 4%, and the peak power consumption is less than 115% of the rated power within one minute of the actual running time of each cycle, the error rate is determined to be less than 0.1% and the physical actuator is activated. If any indicator fails to meet the standard in a certain cycle, the simulation is rerun and the control logic chain is adjusted until the conditions are met, reducing the risk of actuator damage. Through the above-mentioned digital twin pre-execution verification mechanism, the virtual verification sandbox unit 3 realizes the comprehensive reliability verification of the control logic chain, effectively avoiding equipment damage and test interruption caused by control logic conflicts, and providing a solid verification guarantee for the safe and stable operation of the environmental test chamber.

[0071] When the control logic chain is pre-executed and verified in virtual verification sandbox unit 3, an exception handling mechanism based on digital twin is set up to handle abnormal situations that occur during the simulation in a timely manner. The specific implementation method is as follows:

[0072] The exception handling mechanism of Virtual Verification Sandbox Unit 3 includes:

[0073] During the simulation operation of the digital twin, the overshoot of parameters such as actuator torque and temperature is calculated in real time. When the overshoot exceeds 4%, the system automatically identifies it as a control logic anomaly and immediately triggers a control logic chain reconfiguration process within the container. This prevents damage to the physical actuator or distortion of test data due to excessive overshoot, ensuring timely problem detection. Based on the real-time status data of the digital twin, such as the output signals of each functional block and the feedback data of the actuator, the fault tracing algorithm locates the functional block causing the overshoot. For example, in a temperature overshoot scenario, the proportional coefficient of the PID control functional block is found to be set too high, causing the system to over-respond. After identifying the problematic functional block, a similar functional block with a response accuracy improved by at least two levels is called from the logic unit library according to the replacement rules. For example, the original PID control with a response accuracy of 1% is replaced with a PID control block with a response accuracy improved by 1%. The function block is replaced with a high-precision PID control function block with a response accuracy of 0.3% to ensure that the replaced function block has stronger control capabilities. After the function block is forcibly replaced, the control logic chain is regenerated and simulated in the digital twin at 10 times the speed. The logic conflict flag, overshoot, and power consumption peak are monitored again. If the overshoot is still greater than 4%, the response accuracy of the function block is further improved by 2 levels each time and re-verified until the overshoot is less than or equal to 4%. Through this iterative optimization mechanism, the overshoot problem of the control logic chain is completely solved, effectively improving the stability of the control logic. Through the above-mentioned anomaly handling mechanism, the virtual verification sandbox unit 3 can automatically identify and solve the overshoot anomaly problem during the simulation operation, avoiding loading defective control logic chains onto the physical actuator.

[0074] After the physical actuator is activated in the virtual verification sandbox unit 3, in order to continuously optimize the control logic, the feedback optimization unit 4 needs to update the function block selection weights based on the actual operating data of the actuator. The specific implementation method is as follows:

[0075] The operation method of feedback optimization unit 4 is as follows:

[0076] The actual response delay and energy consumption deviation of the physical actuator are captured by a high-speed data bus, and mapped to the metadata tags of each functional block in the logic unit library to establish a formula for updating the functional block selection weight:

[0077] Among them, T actualThe actual response time is given by ΔE, where ΔE is the energy consumption deviation, k is the learning rate factor, and T is the actual response time. ideal To predict the target time delay required for the fluctuation pattern, Select weights for the updated function blocks. Choose weights for the original function blocks, where e is a constant, and if the actual response delay T actual Greater than the target delay T ideal Then (T) ideal -T actual A negative value of ΔE results in a lower weight, while a positive value results in a higher weight. The larger the energy consumption deviation ΔE, the greater the weight of the exponential term e. -ΔE The smaller the value, the more significant the correction effect on the weights. After each test cycle, the weights of all functional blocks are recalculated and the priority index of the logic unit library is updated according to the updated weight values. The priority index is arranged in descending order, with functional blocks of higher weights at the top. This makes it easier for the dynamic logic reconfiguration unit 2 to prioritize the selection of functional blocks with better performance, ensuring that the priority of the logic unit library always reflects the actual performance of the functional blocks. Through the above feedback optimization mechanism, the feedback optimization unit 4 realizes the dynamic updating of the functional block selection weights and the continuous optimization of the logic unit library, providing data-driven optimization support for the long-term efficient operation of the environmental test chamber control platform.

[0078] This invention generates an environmental data stream by synchronously collecting parameters such as temperature and humidity through a multimodal sensing unit 1. A dynamic logic reconstruction unit 2 uses a long short-term memory neural network combined with a dual attention mechanism to analyze the interaction of parameters and generate an optimal control logic chain. A virtual verification sandbox unit 3 constructs a digital twin to pre-execute the control logic and activates the physical actuator after ensuring that the error rate is below a preset threshold. A feedback optimization unit 4 dynamically adjusts the selection weight of function blocks according to the actual response delay and energy consumption deviation. This invention provides a high-precision, low-energy-consumption intelligent control solution for the reliability testing of high-end products in aerospace, electronic components and other fields.

[0079] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An integrated control platform for a programmable multi-mode environmental test chamber, characterized in that, Includes the following units: The multimodal sensing unit (1) synchronously collects temperature, humidity, vibration and light intensity parameters through a sensor array to generate an environmental data stream with timestamps; The dynamic logic reconfiguration unit (2) analyzes the environmental data flow based on the long short-term memory neural network, outputs the multi-parameter composite fluctuation mode for future periods, and stores the pre-compiled function block set of the programmable logic controller through the logic unit library. Each function block is labeled with power consumption level and response accuracy metadata. According to the fluctuation mode, the function block combination is dynamically selected from the logic unit library to generate the optimal control logic chain that meets the preset constraints. The virtual verification sandbox unit (3) loads the optimal control logic chain in an independent container for pre-execution. When the error rate is less than 0.1%, the physical actuator is activated, and the actual response delay and energy consumption deviation of the physical actuator are fed back to the dynamic logic reconstruction unit (2) through the feedback optimization unit (4) to dynamically adjust the function block selection weight.

2. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 1, characterized in that, The dynamic logic reconfiguration unit (2) performs the following steps: The environmental data stream is spatiotemporally aligned to eliminate differences in sampling frequencies among multiple sensors; The aligned data is input into a long short-term memory neural network, which includes a dual attention mechanism, wherein a time attention layer captures long-period trends and a parameter-coupled attention layer resolves the interaction weights of temperature, humidity and vibration. Output a multi-parameter composite fluctuation pattern for future time periods. The multi-parameter composite fluctuation pattern is represented in the form of a three-dimensional tensor, with dimensions including a time axis, a parameter type axis, and a fluctuation intensity axis.

3. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 2, characterized in that, The method for constructing the logic unit library specifically includes: The pre-compiled function block set is classified according to the control scenario into basic control set, security protection set and composite response set. The metadata labeled for each function block includes power consumption level, response accuracy and conflict label. The power consumption level is the power consumption range under actual full load operation, the response accuracy is the steady-state error range under standard step input, and the conflict label is a list of other function block IDs that are prohibited from being activated at the same time as this function block.

4. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 3, characterized in that, When the dynamic logic reconfiguration unit (2) generates the optimal control logic chain: Based on the three-dimensional tensor fluctuation pattern, key control constraints are extracted. The key control constraints are: if the fluctuation intensity axis shows a temperature change slope greater than or equal to 5℃ / min, then a PID function block with a response accuracy less than or equal to 0.5% is forcibly selected; if the output vibration-temperature interaction weight of the parameter coupling attention layer is greater than 0.7, then the coupling compensator in the composite response set is activated. A Pareto optimal solution is selected from non-conflicting functional block combinations using a multi-objective optimization algorithm, provided that power consumption ≤ threshold and accuracy ≥ threshold.

5. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 4, characterized in that, The conflict resolution mechanism for the functional block combination includes: When the conflict labels carried by the selected function blocks overlap, the dynamic priority arbitrator is activated. The working principle of the dynamic priority arbitrator is as follows: If there is a conflict between security protection function blocks, the function block with the higher security level will be retained, and the security level will be classified according to the preset security integrity level standard; otherwise, the function block with the largest response time delay will be discarded.

6. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 4, characterized in that, The fitness function of the multi-objective optimization algorithm is constructed as follows: Real-time analysis of the high-frequency component energy ratio and the number of gradient abrupt change points in the three-dimensional tensor fluctuation mode, dynamically allocating power consumption weight coefficients, accuracy weight coefficients, and delay weight coefficients, specifically satisfying: When a temperature fluctuation slope greater than or equal to 8℃ / min and a vibration frequency greater than 100Hz are detected, the accuracy weighting coefficient is increased to 0.7 and the delay weighting coefficient is decreased to 0.

1. When in a constant temperature steady state, the dynamic power consumption weighting coefficient is set to 0.8 and the low power function block combination strategy is activated.

7. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 6, characterized in that, The dynamically allocated power consumption weighting coefficients specifically include: The random forest regression model is trained based on historical experimental data. The model takes the tensor feature value of the fluctuation mode as input and outputs the optimized ratio of power consumption weight coefficient, accuracy weight coefficient and delay weight coefficient. The model parameters are corrected online through the energy consumption deviation data of the feedback optimization unit (4).

8. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 1, characterized in that, The method for pre-executing the virtual verification sandbox unit (3) in an independent container is as follows: A digital twin with the same electromechanical characteristics as the physical actuator is constructed. The initial values ​​of the current environmental parameters and the optimal control logic chain are synchronously injected into the twin. The simulation is run at 10 times the speed, and the logic conflict flag, actuator torque overshoot and instantaneous power consumption curve are monitored in real time. The error rate is determined to be up to standard and the physical actuator is activated only when the conflict flag is 0, the overshoot is less than or equal to 4%, and the peak power consumption is less than 115% of the rated power for three consecutive simulation cycles.

9. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 8, characterized in that, The exception handling mechanism of the virtual verification sandbox unit (3) includes: If an overshoot of more than 4% is detected, the control logic chain is automatically reconstructed within the container. Based on the real-time status data of the digital twin, the function block that caused the overshoot is forcibly replaced and re-verified. The replacement rule is to call a function block of the same type with a response accuracy improved by at least two levels from the logic unit library.

10. The integrated control platform for a programmable multi-mode environmental test chamber according to claim 1, characterized in that, The operation method of the feedback optimization unit (4) is as follows: The actual response delay and energy consumption deviation of the physical actuator are captured by a high-speed data bus, and mapped to the metadata tags of each functional block in the logic unit library to establish a formula for updating the functional block selection weight: Among them, T actual The actual response time is given by ΔE, where ΔE is the energy consumption deviation, k is the learning rate factor, and T is the actual response time. ideal To predict the target time delay required for the fluctuation pattern, Select weights for the updated function blocks. Weights are selected for the original function blocks, where e is a constant. After each trial cycle, the weights of all function blocks are recalculated and the priority index of the logic unit library is updated.

Images