Method for laser thermal conductance flow leveling for ultra-high throughput genetic sequencing

Abstract

The application discloses a method for laser thermal conductance flow balance for ultra-high-throughput gene sequencing, and relates to the field of sequencing thermal conductance flow control. The method quantifies and models the heat transfer effect between lasers in advance by establishing a thermal crosstalk coupling coefficient matrix. When the set power to be applied to each laser is known, the model can be used to calculate the predicted temperature rise caused by the thermal crosstalk generated by adjacent lasers on the target laser in real time. Based on the prediction result, a prospective feedforward control signal is generated to actively drive the thermoelectric cooler (TEC) to compensate in advance, so that most of the thermal disturbance is offset before it actually affects the temperature sensor. At the same time, the parallel PID feedback control loop continues to be responsible for eliminating any model prediction deviation and unmodeled disturbance, ensuring the final steady-state accuracy, and realizing rapid and prospective balance of the laser temperature.

Description

Technical Field

[0001] This application relates to the field of sequencing thermal conductivity control, and more specifically, to a method for balancing the thermal conductivity of a laser used in ultra-high-throughput gene sequencing. Background Technology

[0002] Ultra-high-throughput gene sequencing technology is a core tool in modern life science research and clinical diagnosis, relying on multicolor laser systems to precisely excite fluorescent labels bound to different bases. To ensure the accuracy and high signal-to-noise ratio of sequencing results, the output wavelength and power of the laser must be maintained at extremely high stability. These critical performance parameters are extremely sensitive to operating temperature; any minute temperature fluctuation can lead to signal acquisition errors, thus severely affecting sequencing quality. Therefore, precise, stable, and rapid temperature control of the multi-laser modules in the sequencer is a fundamental prerequisite for ensuring its high-performance operation.

[0003] Driven by the design trends of pursuing high throughput and instrument miniaturization, sequencers typically integrate multiple lasers compactly onto a shared, highly thermally conductive heat dissipation substrate to ensure precise alignment of the optical path. Current technologies generally employ independent single-input single-output (SISO) PID feedback control strategies for the temperature control of these lasers. A significant drawback of this strategy is that each laser's temperature control loop is an isolated information island, adjusting its corresponding thermoelectric cooler (TEC) power solely based on its own temperature sensor readings, completely unaware of the operating status (such as power changes) of adjacent lasers. When a laser operates, the waste heat it generates inevitably conducts through the shared substrate to neighboring lasers; for the latter, this heat flow represents an unpredictable and sudden external thermal disturbance.

[0004] This phenomenon, known as thermal crosstalk, is the root of the problem. Due to the physical delay in heat conduction and the lag in the response of the control system, traditional PID controllers can only begin compensating after a temperature deviation has been detected. During sequencing, the laser needs to frequently and rapidly cycle through different operating states. This delayed, reactive control method leads to significant overshoot and fluctuations in the target laser's temperature when disturbances occur, failing to meet the stringent temperature stability requirements (e.g., ±0.01°C) of ultra-high-throughput sequencing. Therefore, thermal crosstalk, caused by shared physical thermal bridges and amplified by the information barriers and inherent delays of traditional control strategies, has become a key technological bottleneck restricting the performance improvement of multi-laser modules and the further development of sequencing throughput and accuracy. Summary of the Invention

[0005] To address the problems in the prior art, according to one aspect of this application, a method for balancing the thermal conductivity of a laser used in ultra-high-throughput gene sequencing is provided, comprising: Obtain the set power vector, actual temperature vector, and set temperature vector; Calculate the temperature error vector between the actual temperature vector and the set temperature vector; Parallel independent feedback control is applied to the temperature error vector to obtain the PID control output vector; Based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector to obtain the crosstalk predicted temperature rise vector. The feedforward compensation control signal is generated based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector to obtain the feedforward control output vector. The feedforward control output vector and the PID control output vector are combined to obtain the final TEC drive vector.

[0006] In the above-mentioned method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing, calculating the temperature error vector between the actual temperature vector and the set temperature vector includes: calculating the temperature error vector between the actual temperature vector and the set temperature vector using the following formula, wherein the formula is: ;in, To set the target stable temperature for each laser in the temperature vector, This represents the real-time temperature values ​​of each laser in the actual temperature vector. This refers to the temperature error of each laser.

[0007] In the above-mentioned method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing, parallel independent feedback control of the temperature error vector is performed to obtain the PID control output vector, including: performing parallel independent feedback control of the temperature error vector using the following formula, wherein the formula is: ;in, and Here are the proportional gain coefficient, integral gain coefficient, and differential gain coefficient for each laser. To account for the temperature error of each laser, These are the various feedback control quantities in the PID control output vector.

[0008] In the above-mentioned method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing, based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector to obtain the crosstalk predicted temperature rise vector. This includes: based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector using the following formula: ;in, Indicates laser For every 1 watt of heat power generated, the laser... The resulting steady-state temperature effect, To set the power vector of the laser Target output power setting value For lasers Crosstalk is used to predict temperature rise.

[0009] In the aforementioned method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing, the generation of a feedforward compensation control signal based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector to obtain the feedforward control output vector includes: generating the feedforward compensation control signal based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector using the following formula: The formula is: ;in, Indicates laser The TEC thermoelectric conversion gain coefficient, For lasers The feedforward control quantity.

[0010] In the above-mentioned method for balancing the thermal conductivity of lasers used in ultra-high throughput gene sequencing, the control signal synthesis of the feedforward control output vector and the PID control output vector to obtain the final TEC driving vector includes: adding the feedforward control output vector and the PID control output vector to obtain a composite control signal vector; and performing amplitude limiting processing on the composite control signal vector to obtain the final TEC driving vector.

[0011] Compared to existing technologies, this application provides a method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing. Instead of passively waiting for temperature deviations before adjustment, it pre-quantifies and models the heat transfer effect between lasers by establishing a thermal crosstalk coupling coefficient matrix. When the system knows the set power to be applied to each laser, it can use this model to calculate in real time the predicted temperature rise of the target laser caused by thermal crosstalk from neighboring lasers. Based on this prediction, the system generates a forward-looking feedforward control signal to actively drive the thermoelectric cooler (TEC) to compensate in advance, thus largely offsetting the thermal disturbance before it actually affects the temperature sensor. Simultaneously, a parallel PID feedback control loop continues to eliminate any model prediction biases and unmodeled disturbances, ensuring the final steady-state accuracy. This proactive elimination of predictable disturbances through a feedforward mechanism fundamentally solves the technical problem of traditional independent PID control's inability to effectively suppress thermal crosstalk due to information silos and response lag, achieving rapid and forward-looking temperature balancing of the laser. Attached Figure Description

[0012] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings.

[0013] Figure 1This is a flowchart of a method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application.

[0014] Figure 2 This is a schematic diagram of the data flow in a laser thermal conductivity balancing method for ultra-high throughput gene sequencing according to an embodiment of this application.

[0015] Figure 3 This is a flowchart of step 6 in the method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application.

[0016] Figure 4 This is a data flow diagram of step 6 in the method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application. Detailed Implementation

[0017] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. It should be understood that the drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0018] To address the technical challenges mentioned above, this application proposes a method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing. Figure 1 This is a flowchart of a method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application. Figure 2 This is a schematic diagram of the data flow in a laser thermal conductivity balancing method for ultra-high throughput gene sequencing according to an embodiment of this application. Figure 1 and Figure 2 As shown, the method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application includes: Step 1, obtaining a set power vector, an actual temperature vector, and a set temperature vector; Step 2, calculating the temperature error vector between the actual temperature vector and the set temperature vector; Step 3, performing parallel independent feedback control on the temperature error vector to obtain a PID control output vector; Step 4, performing thermal crosstalk feedforward perturbation quantization on the set power vector based on the thermal crosstalk coupling coefficient matrix to obtain a crosstalk prediction temperature rise vector; Step 5, generating a feedforward compensation control signal based on the crosstalk prediction temperature rise vector and the TEC thermoelectric conversion gain vector to obtain a feedforward control output vector; Step 6, synthesizing a control signal from the feedforward control output vector and the PID control output vector to obtain the final TEC drive vector.

[0019] In step 1, the setpoint power vector, actual temperature vector, and setpoint temperature vector are obtained. It should be understood that achieving extreme temperature stability is the cornerstone of ensuring sequencing data quality in multi-laser integrated modules for ultra-high-throughput gene sequencing. However, dynamic thermal crosstalk caused by shared heat sinks among the lasers is an insurmountable obstacle for traditional independent feedback control methods. To solve this problem, a composite control architecture is needed that can accurately correct existing temperature deviations and proactively predict and compensate for impending thermal disturbances. Therefore, this application first obtains the setpoint power vector, actual temperature vector, and setpoint temperature vector, providing the necessary, real-time, and fundamental input data for the subsequent two core stages: feedback control and feedforward control. The setpoint power vector serves as the basis for feedforward prediction, revealing the source and intensity of thermal disturbances; while the actual temperature and setpoint temperature vectors are the cornerstone of feedback correction, quantifying the gap between the actual effect of the current control and the target.

[0020] To further clarify the specific embodiments of the present invention, a detailed description will be provided below in conjunction with a specific application scenario. This embodiment aims to describe a complete process for performing the initial steps of a thermal conductivity balancing method in a sequencing module composed of four lasers (labeled as laser 1, laser 2, laser 3, and laser 4, respectively).

[0021] In an exemplary operation, the specific process of step 1 is as follows: The power vector is the set of target output power values ​​issued to each laser by the upper-level sequencing process controller at any given time; it directly reflects the amount of heat to be generated. The actual temperature vector is the set of temperature readings monitored and reported in real time by high-precision temperature sensors, such as negative temperature coefficient thermistors, closely mounted near each laser; it characterizes the current thermal state of the module. The set temperature vector is the set of ideal operating temperature targets pre-set and fixed in the controller based on the optimal operating performance of each laser (such as wavelength and power stability).

[0022] At a certain point in a sequencing cycle, the control unit begins data acquisition. A setpoint temperature vector is established, either before the device leaves the factory or during each calibration. This determination is achieved through experimental calibration, which involves testing the key performance parameters of the lasers at different temperature points and selecting the temperature point with the optimal overall performance as the setpoint. For example, if calibration determines that the optimal operating temperature for all four lasers is 25.00 degrees Celsius, then the setpoint temperature vector is determined to be [25.00, 25.00, 25.00, 25.00]. This vector remains constant throughout the sequencing process.

[0023] Simultaneously, the control device acquires two other dynamic vectors in real time through its hardware interface. For obtaining the actual temperature vector, high-precision thermistors deployed on the four laser modules convert the sensed temperature into a resistance signal. This signal is then sampled, converted, and digitally filtered at kilohertz frequencies by a high-resolution analog-to-digital converter circuit, such as a 24-bit ADC, to obtain an accurate real-time temperature value. For example, at the current sampling moment, the acquired temperatures of the four lasers might be [25.02, 24.99, 25.00, 24.98], which constitutes the actual temperature vector at that moment.

[0024] To obtain the set power vector, the sequencing flow controller needs to excite the fluorescent dye corresponding to base 'A' based on the current biochemical reaction step. For example, laser 1 corresponds to base 'A' and requires an excitation power of 500 milliwatts, while the other three lasers are not needed at this time. The sequencing flow controller then generates an instruction, which is parsed by the lower-level control device to form the set power vector. This vector is dynamically changing and may change to excite the dye corresponding to base 'T' in the next sequencing cycle.

[0025] The final output consists of three distinct vectors: a constant set temperature vector [25.00, 25.00, 25.00, 25.00], a real-time acquired actual temperature vector [25.02, 24.99, 25.00, 24.98], and a set power vector determined by upper-level instructions.

[0026] In step 2, the temperature error vector between the actual temperature vector and the set temperature vector is calculated. Accordingly, the essence of the composite control architecture proposed in this application lies in the synergistic effect of feedforward control and feedback control. The feedforward part aims to actively counteract predictable thermal crosstalk, while the feedback control plays an indispensable role, responsible for correcting all prediction biases of the feedforward model, responding to unmodeled external disturbances, and eliminating the inherent steady-state error of the system. Any effective feedback control's starting point and basis stems from the precise quantification of the deviation between the current system state and the desired target. Therefore, before executing any feedback adjustment action, it is necessary to clearly know the magnitude and direction of the deviation. To this end, this application further calculates the temperature error vector between the actual temperature vector and the set temperature vector to accurately generate this crucial deviation signal for each laser, thereby providing accurate and real-time drive input for subsequent parallel independent PID feedback control.

[0027] In one exemplary operation, step 2, calculating the temperature error vector between the actual temperature vector and the set temperature vector, includes: calculating the temperature error vector between the actual temperature vector and the set temperature vector using the following formula: ;in, To set the target stable temperature for each laser in the temperature vector, This represents the real-time temperature values ​​of each laser in the actual temperature vector. This represents the temperature error of each laser. The formula is a basic subtraction operation performed by the microprocessor or digital signal processor within the control device. The subscript 'i' in the formula represents the index of the laser in the vector, ranging from 1 to 4 in this application. In the specific numerical calculation, the control device performs subtraction on corresponding elements of the two input vectors one by one. For the laser with index 1, its temperature error E1 is calculated as: 25.00 - 25.02 = -0.02 degrees Celsius. This negative result clearly indicates that the current actual temperature of laser 1 is 0.02 degrees Celsius higher than its target temperature. For the laser with index 2, its temperature error E2 is calculated as: 25.00 - 24.99 = +0.01 degrees Celsius. This positive result indicates that the current actual temperature of laser 2 is 0.01 degrees Celsius lower than its target temperature. Similarly, for the laser with index 3, its temperature error E3 is 25.00 - 25.00 = 0.00 degrees Celsius, indicating that the laser's temperature is within the ideal range. For the laser with index 4, its temperature error E4 is 25.00 - 24.98 = +0.02 degrees Celsius, indicating that its temperature is 0.02 degrees Celsius lower than the target.

[0028] After calculating the errors for all four lasers, the control unit combines these individual error values ​​in their original order to form a temperature error vector. Based on the above calculations, the generated temperature error vector is [-0.02, +0.01, 0.00, +0.02]. This vector is a complete and quantified snapshot of the current thermal imbalance state of the entire laser module. It not only contains information on whether each laser has deviated from the target, but also precisely indicates the direction and degree of deviation through the sign and magnitude of the value.

[0029] In step 3, parallel independent feedback control is applied to the temperature error vector to obtain the PID control output vector. It is understandable that although the introduced feedforward control mechanism can preemptively offset most predictable thermal crosstalk, no physical model is perfect. The effectiveness of feedforward control highly depends on the accuracy of the thermal crosstalk model, and in actual operation, there will always be uncertainties such as dynamic changes not covered by the model, small fluctuations in ambient temperature, and parameter drift caused by component aging. These residual, unpredictable disturbances, along with the system's own steady-state error, will accumulate and cause the temperature to deviate from its precisely set target if left untreated. Therefore, this application needs to include a closed-loop correction mechanism. Thus, by applying parallel independent feedback control to the temperature error vector to obtain the PID control output vector, this crucial line of defense can be constructed. It utilizes the accurate temperature error calculated in the previous step to independently generate a real-time, dynamic correction control signal for each laser, thereby ensuring that the final temperature can be accurately and stably maintained at the set point.

[0030] In one exemplary operation, step 3, performing parallel independent feedback control on the temperature error vector to obtain a PID control output vector, includes: performing parallel independent feedback control on the temperature error vector using the following formula, wherein the formula is: ;in, and Here are the proportional gain coefficient, integral gain coefficient, and differential gain coefficient for each laser. To account for the temperature error of each laser, These are the feedback control quantities in the PID control output vector. In other words, this vector representing the current deviation state is transformed into a command vector, i.e., the PID control output vector, through a set of parallel, independent PID controllers. This means that the module equips the four lasers with four independent digital PID control algorithm instances, each instance processing a corresponding element in the temperature error vector and calculating an independent feedback control quantity.

[0031] This formula consists of three core parts. The first part is the proportional term ( It affects the current temperature error. The system provides an immediate response; the larger the error, the stronger the control action, aiming to quickly reduce the deviation. Part Two: Integral Term ( This term accumulates all errors over a past period. Its main function is to eliminate the system's steady-state error; even small, persistent deviations are amplified by the integral action until they are completely eliminated, ensuring long-term control accuracy. In digital implementation, this term is represented as the sum of historical errors multiplied by a time step. The third part is the differential term (…). This term (gain coefficient) responds to the rate of change of the error. It is predictive; when the error changes too rapidly, it has a suppressive effect, preventing excessive overshoot and oscillations in the system response, thus improving the system's dynamic stability and response speed. In digital implementation, this term is approximated by calculating the difference between the current error and the previous error. The three gain coefficients in the formula... and These are key parameters that determine the performance of a PID controller. These parameters need to be precisely tuned for the specific thermodynamic characteristics (such as heat capacity and thermal resistance) of each laser and its corresponding thermal control loop. The tuning process is completed during the equipment commissioning phase and can employ methods such as empirical trial and error, the Ziegler-Nichols tuning method, or more advanced system identification and simulation optimization methods. The goal is to enable the closed-loop system to recover to a stable state as quickly as possible and with minimal overshoot after being disturbed. For example, after repeated experiments and optimizations of the four laser modules, the PID parameters for each laser might be: (Kp1, Ki1, Kd1) = (50, 10, 5) for laser 1, (Kp2, Ki2, Kd2) = (52, 11, 4.8) for laser 2, and so on; each laser has its own set of optimized parameters.

[0032] In the specific calculation execution, the processor within the control device performs PID calculations in parallel for each element of the temperature error vector. Taking laser 1 as an example, its current error E1 is -0.02. The output of the proportional term is 50*(-0.02)=-1.0. For the integral term, if the value of its internal accumulator at the previous moment was -0.15, the accumulated value at the current moment will be updated. For the derivative term, if the error value of the previous control cycle is -0.018, the error change rate is (-0.02-(-0.018)) / Δt, where Δt is the control cycle. This change rate is multiplied by the derivative gain 5 to obtain the derivative control quantity. The three results are added together to obtain the total feedback control quantity UPID_1 for laser 1. The same process also occurs in parallel on the controllers of lasers 2, 3, and 4, which calculate using their respective PID parameters and error values ​​(+0.01, 0.00, +0.02).

[0033] Ultimately, all the independently calculated feedback control variables UPID_1, UPID_2, UPID_3, and UPID_4 are combined to form the PID control output vector. For example, this vector might be calculated as [-1.21, +0.65, +0.05, +1.24]. Each element of this vector represents a specific, quantified control force that needs to be applied to the corresponding laser TEC to correct the current temperature deviation.

[0034] In step 4, based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector to obtain the predicted temperature rise vector of crosstalk. It should be understood that relying solely on PID feedback control is essentially a reactive, reactive adjustment strategy; it can only respond after the temperature has already deviated. However, in a compact multi-laser layout, the source (which laser is operating) and intensity (operating power) of thermal crosstalk caused by the operation of neighboring lasers are completely known, and therefore its thermal impact is largely predictable. If this prior information can be utilized to shift from passive correction to active prevention, the response delay problem of feedback control can be fundamentally overcome. Therefore, this application introduces a thermal crosstalk coupling coefficient matrix to perform thermal crosstalk feedforward perturbation quantization on the set power vector to quantify the temperature impact that the impending thermal crosstalk will cause on each laser, thus providing a quantitative and forward-looking basis for subsequent active compensation.

[0035] In one exemplary operation, step 4, which involves performing thermal crosstalk feedforward perturbation quantization on a set power vector based on the thermal crosstalk coupling coefficient matrix to obtain a crosstalk predicted temperature rise vector, includes: performing thermal crosstalk feedforward perturbation quantization on the set power vector based on the thermal crosstalk coupling coefficient matrix using the following formula: ;in, Indicates laser For every 1 watt of heat power generated, the laser... The resulting steady-state temperature effect, To set the power vector of the laser Target output power setting value For lasers The crosstalk is used to predict the temperature rise. The thermal crosstalk coupling coefficient matrix is ​​defined as an N×N square matrix, which is 4×4 in this four-laser example, and its elements are... The effect of the thermal power generated by laser j on the steady-state temperature of laser i was precisely quantified. Specifically, The physical meaning of this matrix is ​​how many degrees Celsius the temperature of laser i will steadily rise for every 1 watt of heat power generated by laser j. This matrix is ​​the core model characterizing the internal heat transfer properties of this specific laser module, and it is determined through a one-time experimental calibration during the equipment manufacturing or calibration phase. The calibration process is as follows: First, all lasers are placed at a stable initial temperature. Then, each laser is individually subjected to a known constant power. Drive laser j to operate while keeping the other lasers off. Once the entire module reaches a new thermal equilibrium, accurately measure and record the steady-state temperature rise of each of the other lasers i. Finally, through calculation The coupling coefficients are obtained. By repeating the above process with each laser as a heat source, the entire coupling coefficient matrix can be constructed. Since the heat of the laser itself is managed by its own TEC, and crosstalk is calculated, the diagonal elements of the matrix... Zero. An example of a calibrated matrix might look like this (unit: degrees Celsius / watt): Where K12 = 0.10 indicates that for every 1 watt of thermal power generated by laser 2, the temperature of laser 1 will rise by 0.10 degrees Celsius. Then, thermal crosstalk is calculated based on the above formula. This formula means that for the target laser i, the total crosstalk temperature rise it experiences is equal to the set power of each of all other lasers (j from 1 to N, and j ≠ i). Its corresponding coupling coefficient The sum of products. Continuing the previous example, the control unit obtains a setpoint power vector of [0.5, 0.0, 0.0, 0.0] (in watts), indicating that laser 1 operates at 0.5 watts, and the other lasers are off. The control unit will use this vector and a preset coupling coefficient matrix to calculate each element of the crosstalk predicted temperature rise vector. For laser 1 (i=1), its predicted temperature rise... =K12·Pset_2+K13·Pset_3+K14·Pset_4=0.10·0.0+0.04·0.0+0.02·0.0=0.00 degrees Celsius. This indicates that laser 1 will not experience thermal crosstalk from the outside since no other lasers are operating. For laser 2 (i=2), its predicted temperature rise is... =K21·Pset_1+K23·Pset_3+K24·Pset_4=0.11·0.5+0.09·0.0+0.05·0.0=+0.055 degrees Celsius. This predicts that the operation of laser 1 will cause the temperature of laser 2 to rise by 0.055 degrees Celsius. For laser 3 (i=3), its predicted temperature rise is... =K31·Pset_1+K32·Pset_2+K34·Pset_4=0.04·0.5+0.09·0.0+0.10·0.0=+0.020 degrees Celsius. For laser 4 (i=4), its predicted temperature rise is... =K41·Pset_1+K42·Pset_2+K43·Pset_3=0.02·0.5+0.05·0.0+0.10·0.0=+0.010 degrees Celsius. After completing the calculations for all four lasers, combining these results yields the crosstalk prediction temperature rise vector: [0.00,+0.055,+0.020,+0.010]. This vector precisely quantifies the magnitude of the thermal disturbance that will act on each laser due to the current power setting.

[0036] In particular, in the demanding applications of ultra-high-throughput gene sequencing, lasers require frequent and rapid switching, resulting in a constantly changing thermal environment within the laser module. The prediction method based on the static thermal crosstalk coupling coefficient matrix previously described in this application assumes that the thermal effects occur instantaneously and can be simply linearly superimposed. However, this static linear model neglects the inherent time process of heat conduction within the physical structure; that is, it takes time for heat to diffuse from the source laser to the target laser, and the temperature response is not a step change but a gradual process. In system identification and control theory, this process can be precisely characterized as: time delay (θ), the shortest time required for heat to propagate from the source (laser j) to the target (laser i), during which the temperature of laser i remains completely unaffected; and time constant (τ), which describes the time required for the temperature of laser i to change from its initial state to near its new steady-state temperature after the arrival of heat, reflecting the thermal inertia of the system. In high-frequency switching scenarios, thermal disturbances may not reach their steady-state impact before the heat source laser is shut down. In this case, the static model, focusing only on the endpoint, will produce significant prediction errors, leading to inaccurate amplitude and timing of feedforward compensation and weakened compensation effect. To fundamentally solve this problem, the time dimension information of heat conduction needs to be accurately modeled. Therefore, this application proposes a preferred implementation method: based on the thermal crosstalk coupling coefficient matrix, a dynamic thermal response model is introduced to replace the original static gain model by performing parallel prediction and feedforward compensation of dynamic thermal crosstalk between lasers on a set power vector. This model comprehensively depicts the dynamic characteristics of the entire process from power input from one laser to its temperature impact on another, thereby achieving accurate prediction of the entire thermal disturbance process and providing a high-quality data foundation for generating feedforward compensation signals with precise timing matching and dynamic amplitude adjustment.

[0037] Based on this, in a preferred exemplary operation, step 4, based on the thermal crosstalk coupling coefficient matrix, performs thermal crosstalk feedforward perturbation quantization on the set power vector to obtain the crosstalk predicted temperature rise vector, including: The thermal crosstalk effect between each pair of lasers is systematically identified to obtain the continuous transfer function vector of its dynamic thermal response model. First, a mathematical model that accurately describes the physical process of thermal crosstalk needs to be established as the theoretical basis for all subsequent calculations, resulting in a set of continuous-time transfer functions that can fully characterize the dynamic characteristics of heat transfer between any two lasers within the module. In practice, the laser module first needs to undergo a one-time offline experimental calibration; this process is called system identification. For each pair of lasers (source laser j to target laser i, where j is not equal to i), a known power excitation signal (such as a step signal) is applied to laser j, while the temperature response curve of laser i is acquired at high frequency. Then, using the system identification algorithm, the acquired input-output data is fitted to a pre-defined dynamic model structure. In engineering practice, the first-order time-delayed (FOPDT) model is widely used because it can describe such thermal processes with sufficient accuracy. Its continuous transfer function is as follows: In this formula, To describe the power from laser j Temperature change to laser i The transfer function. It is a Laplace transform complex variable, and the model contains three key parameters: The steady-state gain represents the final temperature effect that the thermal disturbance can achieve. The time delay represents the shortest time required for heat to travel from laser j to laser i. The time constant reflects the system's thermal inertia and describes the speed of its temperature response. This key parameter can be obtained through offline experimental calibration. For example, the thermal response parameter between a pair of lasers can be obtained through calibration as: time delay. The time constant is 0.5 seconds. The time is 2.0 seconds. By repeating this identification process for all laser pairs within the module, a complete set of continuous transfer functions can be obtained, forming the basis of the dynamic thermal response model.

[0038] Based on the discretization coefficients, the continuous transfer function is converted into a discrete-time difference equation to obtain the temperature influence vector. Next, since the digital control device operates in the discrete time domain, it cannot directly process the continuous-time transfer function. Therefore, it needs to be converted into a discrete algorithm that can be iteratively executed within each control cycle, establishing a real-time, step-by-step prediction equation for the thermal crosstalk effect of each pair of lasers. In specific implementation, each continuous transfer function obtained in the previous step is discretized according to the fixed control cycle T of the control device. The determination of this cycle aims to balance the sampling rate required to capture the system's thermal dynamics with the computation time required to execute the control algorithm, such as T = 0.1 seconds. For example, using methods such as a zero-order hold, it can be converted into a discrete-time difference equation of the following form: This equation describes the temperature effect component caused by laser j on laser i during the current k-th control cycle. It depends on the influence component of the previous cycle. and power input at a certain point in the past The discretization coefficients among them and Calculated from continuous model parameters: Meanwhile, continuous time delay Converted into discrete delay steps Its value is Divide by control period The integer part, i.e. , This is to round down. Thus, the dynamic thermal response of each laser pair is expressed as a recursive formula that can be efficiently executed in the controller.

[0039] Based on the temperature influence vector, a crosstalk prediction temperature rise vector is obtained. That is, the thermal crosstalk experienced by any laser at a given moment is the sum of the thermal effects from all other operating lasers. Therefore, it is necessary to merge all independent influence components to obtain a total, comprehensive prediction result, thereby generating a single, accurate crosstalk prediction temperature rise value for each laser that reflects the intensity of the total thermal disturbance experienced at the current moment. In specific implementation, during each control cycle k, the control device performs the following calculations in parallel for each target laser i (from 1 to N): it iterates through all other source lasers j (from 1 to N, and j is not equal to i), using the difference equation obtained in the previous step, and reads data from the historical power data buffer. And combined with the stored state of the previous cycle Calculate the temperature influence component for the current cycle. After calculating the effects of all j on i, they are summed as shown in the following formula: Calculated This is the predicted temperature rise of laser i due to total crosstalk during the current period k. This applies to all lasers... Combined, these elements form the final crosstalk prediction temperature rise vector. This output vector integrates historical power information and a precise dynamic model, enabling high-fidelity prediction of the dynamic evolution of thermal disturbances. This provides high-quality data input for subsequent feedforward compensation, achieving excellent suppression of thermal crosstalk. Furthermore, based on this high-fidelity dynamic prediction, the generated feedforward compensation signal will also be a dynamic signal profile. It can begin pre-cooling before the actual arrival of the thermal disturbance and continuously adjust the cooling capacity during the disturbance process, achieving precise matching with the disturbance waveform. This fundamentally suppresses temperature fluctuations, rather than merely reducing their peak values, achieving both foresight and smoothness in the compensation. Additionally, since the feedforward compensation has already handled most of the predictable disturbances, the PID feedback loop is left to process small residuals such as model errors and random disturbances. This allows the PID controller to operate in a more linear region, avoiding integral saturation and other problems, making the entire closed-loop system more stable and robust.

[0040] In step 5, a feedforward compensation control signal is generated based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector to obtain the feedforward control output vector. Accordingly, the previous steps have successfully quantified the abstract thermal crosstalk effect into a specific, predictable crosstalk predicted temperature rise vector. This vector accurately depicts the temperature impact of an upcoming thermal disturbance on each laser. However, this prediction itself remains at the thermodynamic level; it is merely a quantitative description of the problem, not a direct instruction for a solution. The final actuator of the control device is the thermoelectric cooler (TEC), whose behavior is driven by electrical signals (such as voltage or current). Therefore, between prediction and action, the predicted temperature change needs to be accurately translated into an electrical signal capable of driving the TEC to produce an equivalent reverse thermal effect. To this end, this application generates a feedforward compensation control signal based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector precisely to complete this crucial conversion, generating a quantified, executable, and forward-looking compensation control instruction for each laser.

[0041] In one exemplary operation, step 5, generating a feedforward compensation control signal based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector to obtain a feedforward control output vector, includes: generating the feedforward compensation control signal based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector using the following formula: ;in, Indicates laser The TEC thermoelectric conversion gain coefficient, For lasers The feedforward control quantity. Each element of the TEC thermoelectric conversion gain vector. The gain coefficient represents the cooling or heating efficiency of the thermoelectric cooler (TEC) of the corresponding laser i. Its physical meaning is: how much control signal (e.g., a driving voltage in volts) needs to be applied to the TEC to achieve a stable 1-degree Celsius temperature change in laser i. This gain coefficient is a key conversion factor connecting the control command (electrical signal) and the control effect (temperature change). This vector is also obtained through experimental calibration during the equipment commissioning phase. The calibration process is as follows: For each laser i, under a stable environment without external thermal disturbance, a known, constant control signal U_test (e.g., 1 volt) is applied to its TEC. After its temperature reaches a new stable point, the temperature change ΔT_eff caused by this signal is accurately measured. Then, the thermoelectric conversion gain coefficient of the TEC is Gi = U_test / ΔT_eff. This process is repeated for all TECs in the module to obtain the complete TEC thermoelectric conversion gain vector. For example, after calibration, the resulting gain vector might be [20.0, 21.0, 20.5, 20.8], in volts per degree Celsius.

[0042] For each laser i, the required feedforward compensation control signal This is equivalent to the predicted crosstalk temperature rise it will experience. Multiply by its own TEC thermoelectric conversion gain coefficient This calculation is performed by the processor within the control unit using element-wise multiplication of the input vector. Specifically, the control unit has obtained the crosstalk prediction temperature rise vector [0.00, +0.055, +0.020, +0.010] from the previous step, in degrees Celsius. Now, the control unit will use this vector and the preset TEC thermoelectric conversion gain vector [20.0, 21.0, 20.5, 20.8] to calculate the feedforward control output vector. For laser 1 (i=1), its feedforward control quantity... =20.0·0.00=0.00 volts. Since crosstalk temperature rise was not predicted, no feedforward compensation is needed. For laser 2 (i=2), its feedforward control quantity... =21.0·(+0.055)=+1.155 volts. This calculation result indicates that, in order to actively counteract the upcoming temperature rise of +0.055 degrees Celsius, a cooling signal of 1.155 volts needs to be pre-applied to the TEC of laser 2, such that the positive voltage here is for cooling. For laser 3 (i=3), its feedforward control quantity... =20.5·(+0.020)=+0.410 volts. For laser 4 (i=4), its feedforward control quantity is... =20.8·(+0.010)=+0.208 volts.

[0043] After completing the calculations for all four lasers, these independent control variables are combined in their original order to form the final output feedforward control vector: [0.00, +1.155, +0.410, +0.208]. This vector is the final product of the feedforward control path, containing a precise and forward-looking set of electrical signal commands. Unlike the parallel-calculated PID control output vector, which is designed for ex-post correction, this vector is designed for ex-ante prevention. It will drive the corresponding TEC to generate a just-right reverse thermal effect before thermal crosstalk actually causes temperature fluctuations, thus nipping the disturbance in the bud.

[0044] In step 6, the feedforward control output vector and the PID control output vector are synthesized to obtain the final TEC drive vector. That is, two complementary control signals are generated through two parallel paths: one is the PID feedback control output, based on real-time error and designed for ex-post correction, which ensures the final control accuracy and stability; the other is the feedforward compensation control output, based on power preset and designed for ex-ante prevention, which overcomes the inherent delay of feedback control and actively suppresses major thermal disturbances. However, these two independent sets of digital instructions, calculated based on different principles, ultimately need to act on the same physical actuator, the thermoelectric cooler (TEC). Therefore, this application synthesizes the feedforward control output vector and the PID control output vector to seamlessly integrate the two control concepts of forward-looking predictive compensation and high-precision feedback correction, forming a unified, optimal final execution instruction that conforms to physical hardware constraints.

[0045] In one exemplary operation Figure 3 This is a flowchart of step 6 in the method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application. Figure 4 This is a data flow diagram of step 6 in the method for balancing the thermal conductivity of a laser for ultra-high throughput gene sequencing according to an embodiment of this application. Figure 3 and Figure 4 As shown, step 6, which synthesizes the feedforward control output vector and the PID control output vector to obtain the final TEC drive vector, includes: step 61, adding the feedforward control output vector and the PID control output vector to obtain the composite control signal vector; step 62, performing amplitude limiting processing on the composite control signal vector to obtain the final TEC drive vector.

[0046] In the above operation, the specific process of step 6 is as follows: First, step 61 is executed, which adds the feedforward control output vector and the PID control output vector to obtain the composite control signal vector. The addition operation is performed by the arithmetic logic unit in the control device, which sums the two input vectors of the same dimension element by element. The specific calculation process is as follows: For the laser with index 1, its composite control signal is the sum of the PID component and the feedforward component, i.e., -1.21 + 0.00 = -1.21 volts. For the laser with index 2, its composite control signal is +0.65 + 1.155 = +1.805 volts. For the laser with index 3, its composite control signal is +0.05 + 0.410 = +0.460 volts. For the laser with index 4, its composite control signal is +1.24 + 0.208 = +1.448 volts. By performing this operation on all elements, an intermediate result is obtained: a composite control signal vector [-1.21, +1.805, +0.460, +1.448]. This vector theoretically combines the advantages of two control strategies, including both active compensation for known disturbances and precise correction for unknown deviations.

[0047] Next, step 62 is performed, which limits the composite control signal vector to obtain the final TEC drive vector. This is crucial to ensure that the control commands can be executed safely and accurately on the physical hardware. Any actual hardware circuit has physical upper and lower limits for its output capability. For example, the digital-to-analog converter (DAC) used to generate the TEC drive voltage and the subsequent power amplifier circuit have a limited output voltage range. If the theoretically calculated composite control signal exceeds this range, it will not only fail to be executed correctly by the hardware (resulting in saturation or clipping), but may also damage the circuit. Therefore, each element of the composite control signal vector needs to be limited. An upper limit and a lower limit need to be preset, and these values ​​are strictly determined by the hardware specifications of the TEC drive circuit. For example, if the drive circuit is designed to output from -5.0 volts to +5.0 volts, then the upper limit is +5.0 and the lower limit is -5.0. These values ​​are stored in the non-volatile memory of the control device. The limiting process involves comparing each element of the composite control signal vector with the preset upper and lower limits. If the value of an element is between the upper and lower limits, its value remains unchanged; if the value of an element is greater than the upper limit, its value is forcibly set to the upper limit; if the value of an element is less than the lower limit, its value is forcibly set to the lower limit. In this example, the resulting composite control signal vector is [-1.21, +1.805, +0.460, +1.448]. Comparing this to the range [-5.0, +5.0]: -1.21 is within this range and remains unchanged; +1.805 is within this range and remains unchanged; +0.460 is within this range and remains unchanged; +1.448 is within this range and remains unchanged. Therefore, at this specific moment, the limiting process does not change any value. However, in some cases of severe dynamic response or extremely large disturbances, the composite signal may well exceed the range. For example, if the calculated value of the composite control signal for a laser is +6.2 volts, after this step, its value will be forcibly constrained to +5.0 volts.

[0048] After limiting, the final output, the final TEC drive vector, is obtained. In this example, this vector is [-1.21, +1.805, +0.460, +1.448]. This vector is a set of directly executable, safe, and optimized digital instructions. It will be sent by the control unit to four independent digital-to-analog converters via a digital bus. These converters will then convert these digital quantities into precise analog voltages, ultimately driving the four TECs to perform cooling or heating at the corresponding power, thereby achieving ultra-high precision dynamic balance of the temperature of the entire laser module.

[0049] In summary, the method for balancing the thermal conductivity of lasers used in ultra-high-throughput gene sequencing, based on the embodiments of this application, has been clarified. Instead of passively waiting for temperature deviations before adjustment, it pre-quantifies and models the heat transfer effect between lasers by establishing a thermal crosstalk coupling coefficient matrix. When the system knows the set power to be applied to each laser, it can use this model to calculate in real time the predicted temperature rise of the target laser caused by thermal crosstalk from neighboring lasers. Based on this prediction, the system generates a forward-looking feedforward control signal to actively drive the thermoelectric cooler (TEC) to compensate in advance, thereby largely offsetting the thermal disturbance before it actually affects the temperature sensor. Simultaneously, the parallel PID feedback control loop continues to eliminate any model prediction biases and unmodeled disturbances, ensuring the final steady-state accuracy. This method of actively eliminating predictable disturbances through a feedforward mechanism fundamentally solves the technical problem of traditional independent PID control's inability to effectively suppress thermal crosstalk due to information silos and response lag, achieving rapid and forward-looking balancing of laser temperature.

Claims

1. A method for balancing the thermal conductivity of a laser used in ultra-high-throughput gene sequencing, characterized in that, include: Obtain the set power vector, actual temperature vector, and set temperature vector; Calculate the temperature error vector between the actual temperature vector and the set temperature vector; Parallel independent feedback control is applied to the temperature error vector to obtain the PID control output vector; Based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector to obtain the crosstalk predicted temperature rise vector. The feedforward compensation control signal is generated based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector to obtain the feedforward control output vector. This includes: generating the feedforward compensation control signal based on the crosstalk predicted temperature rise vector and the TEC thermoelectric conversion gain vector using the following formula: [Formula omitted for brevity] in, Indicates laser The TEC thermoelectric conversion gain coefficient, For lasers Feedforward control quantity For lasers Crosstalk predicts temperature rise; The control signal is synthesized from the feedforward control output vector and the PID control output vector to obtain the final TEC drive vector, including: adding the feedforward control output vector and the PID control output vector to obtain a composite control signal vector; and performing amplitude limiting processing on the composite control signal vector to obtain the final TEC drive vector.

2. The method for balancing the thermal conductivity of a laser for ultra-high-throughput gene sequencing according to claim 1, characterized in that, Calculating the temperature error vector between the actual temperature vector and the set temperature vector includes: calculating the temperature error vector between the actual temperature vector and the set temperature vector using the following formula, wherein the formula is: in, To set the target stable temperature for each laser in the temperature vector, This represents the real-time temperature values ​​of each laser in the actual temperature vector. This refers to the temperature error of each laser.

3. The method for balancing the thermal conductivity of a laser for ultra-high-throughput gene sequencing according to claim 2, characterized in that, To obtain the PID control output vector by performing parallel independent feedback control on the temperature error vector, the following method is used: Parallel independent feedback control of the temperature error vector is performed using the following formula: in, and Here are the proportional gain coefficient, integral gain coefficient, and differential gain coefficient for each laser. To account for the temperature error of each laser, These are the various feedback control quantities in the PID control output vector.

4. The method for balancing the thermal conductivity of a laser for ultra-high-throughput gene sequencing according to claim 1, characterized in that, Based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector to obtain the crosstalk predicted temperature rise vector, including: based on the thermal crosstalk coupling coefficient matrix, thermal crosstalk feedforward perturbation quantization is performed on the set power vector using the following formula, wherein the formula is: in, Indicates laser For every 1 watt of heat power generated, the laser... The resulting steady-state temperature effect, To set the power vector of the laser The target output power setting value.

Images