Physical design method of microfluidic biochip considering volume management and channel storage

Abstract

The application discloses a micro-fluidic biochip physical design method considering volume management and channel storage. Before starting the iteration process of the simulated annealing, a feasible initial solution is designed in advance, and the initial solution includes two stages of initial layout and initial wiring. For the adjustment of the current solution, three adjustment strategies are provided: 1) translating components or storage channels; 2) rotating components or storage channels; and 3) exchanging the positions of two components. After selecting a certain adjustment strategy, the wiring of the related components needs to be removed, and the wiring is rewound after the adjustment strategy is executed. Finally, in order to avoid the conflict problem caused by the adjustment strategy, the execution time of the flow processing task is adjusted through the re-scheduling technology, so that each adjusted solution is conflict-free. The application considers the layout and wiring together, effectively reduces the biochemical reaction completion time, the flow channel length and the number of cross points, and greatly improves the quality of the layout and wiring.

Description

Technical Field

[0001] This invention belongs to the field of computer-aided design technology, specifically relating to a physical design method for microfluidic biochips. Background Technology

[0002] In recent years, researchers have proposed several microfluidic chip architecture design methods to continuously improve chip performance and reduce manufacturing costs. Wajid Hassan Minhass proposed a top-down design flow to construct the fluid layer architecture of CFMBs while minimizing the execution time of biochemical reactions. Qin Wang proposed a sequence pair-based physical design method, using simulated annealing to iteratively adjust the placement results in the placement stage and a negotiation-based method in the routing stage, comprehensively considering both stages of physical design. Chun-Xun Lin introduced a barrier-around right-angle Steiner tree model in integrated circuits to reduce the channel length within the chip. Kailin Yang used a routing strategy with arbitrary bending angles to replace the traditional Manhattan channel with a 90° corner, significantly improving routing flexibility. Huang X and Xing Huang considered actual fluid transport within the design flow to automate the design of CFMBs. Xing Huang proposed a design flow called MiniControl to systematically consider minimizing the number of control ports. However, the above works are all implemented under traditional chip architectures with dedicated memory and do not consider the huge potential of channels in fluid storage.

[0003] Chunfeng Liu first proposed the Distributed Channel Storage Architecture (DCSA), and further improved the functionality of the channel storage by introducing fluid ports. Xing Huang integrated the cleaning task into the high-order synthesis and physical design process of DCSA, ultimately generating a physical design scheme based on distributed channel storage. Zhisheng Chen first proposed the fluid layer physical design problem based on DCSA and proposed a top-down synthesis algorithm to generate an efficient solution. However, none of the above works considered the volume management problem caused by actual fluid operations.

[0004] Continuous microfluidic biochips (CFMBs) have revolutionized biochemical assays due to their high efficiency, high precision, and high throughput. Currently, CFMBs are widely used in the fields of biochemistry and biomedicine. For example... Figure 1As shown, CFMBs consist of two physical layers: a flow layer and a control layer. The channels in the flow layer, called flow channels, transport biochemical samples and reagents under external pressure. The channels in the control layer, called control channels, are connected to control ports that connect to the external pressure source, controlling the opening and closing of microvalves placed between the flow layer and the control layer. In CFMBs, a dedicated storage device can be created using combinations of microvalves to store intermediate fluids generated during the biochemical reaction. However, this storage method has several limitations: 1) limited storage unit capacity; 2) the storage location cannot be moved once determined; 3) the storage device requires a large chip area; and 4) limited storage bandwidth. Therefore, to address these limitations, DCSA allows the intermediate fluids generated by each component during the biochemical reaction to be stored in the flow channel near that component. This allows the flow channel to switch arbitrarily between storage and transport functions according to actual needs. In DCSA (Distributed Volume Assisted Amplification), since the volume of each flow channel is limited, if the volume of fluid to be stored exceeds the capacity of the selected channel, some fluid will occupy other flow channels. Removing this fluid may result in insufficient fluid input volume for subsequent fluid operations, preventing the final experiment from completing normally. Conversely, it may delay the execution of other flow processing tasks, ultimately significantly delaying the overall completion time of the biochemical reaction. Currently, no physical design algorithm has been proposed for actual volume management under DCSA. Therefore, during the physical design phase, detailed consideration of flow path planning under DCSA, avoiding resource conflicts between different flow processing tasks, and reasonably considering the volume constraints of flow channels are crucial for ensuring the normal execution of CFMBs (Flow Flow Management Blocks). Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, this invention provides a microfluidic biochip physical design method that considers volume management and channel storage. Before the iterative process of simulated annealing begins, a feasible initial solution is pre-designed, which includes two stages: initial layout and initial routing. For adjusting the current solution, three adjustment strategies are proposed: 1) translating components or storage channels; 2) rotating components or storage channels; 3) swapping the positions of two components. After selecting a certain adjustment strategy, the routing of the relevant components needs to be removed, and routing rerouting is performed after the adjustment strategy is executed. Finally, to avoid conflicts caused by adjustment strategies, this invention uses rescheduling technology to adjust the execution time of the stream processing task, thereby ensuring that each adjusted solution is conflict-free. This invention considers both layout and routing pass / fail, effectively reducing the completion time of biochemical reactions, the length of flow channels, and the number of intersections, greatly improving the quality of layout and routing.

[0006] The technical solution adopted by this invention to solve its technical problem includes the following steps: Step 1: Initial layout stage; Set a component-intensive expansion area Used in the middle of the chip to constrain unplaced components. Inside; Then calculate each component. The priority is calculated using the following formula:

[0007] in, Representation Component With components The number of connections between them Representation Component With components The number of connections between them Indicates the allocated component library; Insert all components into the priority queue, and then dequeue them in sequence. inside, if If there is not enough space to place the current component, it needs to be expanded. Until the placement conditions are met; Step 2: Initial unrouting stage; Step 2-1: Set up the flow processing tasks, including fluid input task, excess fluid removal task, waste fluid removal task, source input task, fluid storage task, and fluid extraction task; Step 2-2: Construct storage channels. First, sort the stream processing tasks in ascending order based on their completion time. Then, group the stream processing tasks using the scan-line algorithm, with tasks within the same group executed in parallel. After grouping, route the stream processing tasks within each group sequentially. Before routing, reduce the cost of previously routed grid cells. Next, determine if the stream ports or waste ports bound to the stream processing task have been placed on the port layer. If not, select an available location on the port layer such that the stream port or waste port is closest to the component of its bound stream processing task. Finally, construct the corresponding stream channel using the A* algorithm based on the type of stream processing task. During the stream channel construction process, the A* algorithm continuously selects the grid cell with the lowest routing cost for routing. The current stream processing task... The flow path corresponding to the flow channel to the cell mesh Cost function The calculation is as follows:

[0008] In formula (2), for Manhattan distance to the target location and distance from the starting location to the target location The sum of Manhattan distances; in formula (3), for When wiring The current cost, for The initial cost, For one variable; For all currently wired stream processing tasks, the flow path passes through... The number of tasks; For the process The number of storage channels; It is an infinite quantity; The size will depend on The system will dynamically adjust based on the wiring conditions. Case 1: If If there is no flow path routed here, then for Scenario 2: In the same group of stream processing tasks, to avoid resource conflicts between parallel stream processing tasks within the same group, after a stream processing task completes its routing, the cost of the mesh cells it has routed needs to be set to... Right now ,otherwise, According to The number of streaming channels and storage channels deployed on the device can be dynamically adjusted. ; Step 3: Generate nearest neighbor solutions; After the initial solution is obtained, the initial solution is defined as follows: This allows them to participate in the generation of subsequent nearest-neighbor solutions; The current layout and routing solution can be changed using the following three adjustment strategies: Strategy 1: Randomly select a component and change its position; Strategy 2: Randomly select a component and rotate it clockwise. Strategy 3: Randomly select two components and swap their positions; Abstract the storage channel into a length of... A component with a width of 1 unit participates in both Strategy 1 and Strategy 2; the generation of nearest neighbor solutions is defined as... Each adjustment strategy is selected with equal probability and adjusted accordingly. Before implementing the adjustment strategy, the flow paths of the selected components and other flow paths that intersect with these flow paths are removed; after the adjustment strategy is implemented, the removed flow paths are rerouted according to the flow path planning algorithm of the storage driver. Step 4: Rescheduling; Based on the timing diagram, determine the prerequisite tasks for each stream processing task; that is, the current stream processing task can only be executed after all prerequisite tasks are completed. After calculating the prerequisite tasks, the actual completion time of each stream processing task needs to be recalculated. If the current stream processing task... start time Earlier than the maximum completion time of all prerequisite tasks Therefore, the start time of the task needs to be postponed to [date to be filled in]. After that; and then regarding and If there are overlapping tasks in the flow paths, their start times need to be postponed to avoid resource conflicts. After; for End time The calculation formula is as follows:

[0009] in, for Execution time, for fluid volume, For the unit volume of the flow channel, express The length of the flow channel, This indicates the velocity of the fluid in the flow channel; Step 5: Cost function; After the rescheduling is completed, a comparison is required. and The quality of the solution is evaluated using the following formula:

[0010] in, This represents the completion time of the biochemical reaction at the current nearest solution. The total length of the flow channel. The number of intersections, For three given constants; like Having more Lower cost value, i.e. If the solution is correct, then accept the solution; however, during the generation of nearest neighbor solutions, it may cause... It has higher cost value, that is To enhance global search capabilities and avoid getting trapped in local optima, simulated annealing algorithms must use probabilistic... To accept the worse solution, among which .

[0011] Preferably, the initial area of ​​the component dense expansion region is , and For the length and width of the chip, and It is an integer greater than 1.

[0012] The beneficial effects of this invention are as follows: This invention takes layout and wiring into account, effectively reducing the completion time of biochemical reactions, the length of flow channels, and the number of intersections, thus greatly improving the quality of layout and wiring. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the CFMB structure.

[0014] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] The purpose of this invention is to design a physical design method for microfluidic biochips that considers volume management and channel storage, thereby obtaining a physical design solution that optimizes the actual scheduling time, flow channel length, and number of intersections.

[0017] This invention is implemented using the following approach: a microfluidic biochip physical design method considering volume management and channel storage, the flowchart of which is shown below. Figure 2 As shown, the strategies involved in each stage are listed on the left, and the main steps of the proposed method are on the right. This invention takes into account placement and routing pass / fail judgment through the Simulated Annealing (SA) algorithm, effectively reducing the completion time of the biochemical reaction, the length of the flow channel, and the number of intersections, thus greatly improving the quality of placement and routing.

[0018] The present invention specifically includes the following steps: 1. Construction of the Initial Solution: Before the iterative process of simulated annealing begins, a feasible initial solution is pre-designed. The initial solution includes two stages: initial placement and initial routing. Regarding placement, this invention proposes an algorithm based on region expansion and component connection priority to generate the initial placement. Furthermore, routing construction is a crucial means of adjusting the solution in subsequent iterations. Therefore, this invention proposes channel-store-driven flow path planning to generate routing results, thereby obtaining a conflict-free and feasible physical design solution.

[0019] 2. Simulated Annealing Iteration: First, for adjusting the current solution, this invention proposes three adjustment strategies: 1) translating components or storage channels; 2) rotating components or storage channels; 3) swapping the positions of two components. Second, after selecting a certain adjustment strategy, the wiring of the relevant components needs to be removed, and the wiring is rerouted after the adjustment strategy is executed. Finally, to avoid conflicts caused by the adjustment strategy, this invention uses rescheduling technology to adjust the execution time of the stream processing task, thereby ensuring that each adjusted solution is conflict-free.

[0020] 1. Initial Solution Layout Stage In the layout stage, this invention sets an initial area of... dense expansion area of ​​components Used in the middle of the chip to constrain unplaced components. Inside, among them, and For the length and width of the chip, and The integers are greater than 1, and then each component is calculated. Its priority is calculated using the following formula:

[0021] Then all components are inserted into a priority queue and dequeued in sequence. inside, if If there is not enough space to place the current component, it needs to be expanded. (i.e., shrink) and Continue until the placement conditions are met.

[0022] 2. Initial Dewiring Stage First, to avoid the influence of excess fluid in the flow channel and waste fluid in the component on the experimental results, the fluid removal task also needs to be considered. Therefore, the actual flow processing tasks include fluid input task, excess fluid removal task, waste fluid removal and source input task, fluid storage task, and fluid extraction task.

[0023] Secondly, this invention prioritizes the construction of storage channels, resulting in shorter overall line lengths and higher utilization of streaming channels. After constructing the storage channels, the streaming tasks are first sorted in ascending order based on their completion time. Then, to avoid resource conflicts, this invention employs an efficient scanline algorithm to group the streaming tasks, with tasks within the same group executing in parallel. After grouping, the streaming tasks within each group are routed sequentially. To better reuse already routed streaming channels, the cost of previously routed grid cells is reduced before routing tasks in this group. Then, it is determined whether the streaming port or waste port bound to the streaming task has been placed on the port layer. If not, an available location is selected on the port layer such that the streaming port or waste port is closest to the component of its bound streaming task. Finally, the A* algorithm is used to construct the corresponding streaming channel according to the type of streaming task. During the construction of the streaming channel, the A* algorithm continuously selects the grid cell with the lowest routing cost for routing. The flow path corresponding to the flow channel to the cell mesh Cost function The calculation is as follows:

[0024] In formula (2), for Manhattan distance to the target location and distance from the starting location to the target location The sum of Manhattan distances, in formula (3), for When wiring The current cost, for The initial cost, For one variable, For all currently wired stream processing tasks, the flow path passes through... The number of tasks, For the process The number of storage channels, For an infinite quantity, The size will depend on The system will dynamically adjust based on the wiring conditions. Case 1: If If there is no flow path routed here, then for Scenario 2: In the same group of stream processing tasks, to avoid resource conflicts between parallel stream processing tasks within the same group, after a stream processing task completes its routing, the cost of the mesh cells it has routed needs to be set to... Right now Otherwise, in order to better achieve channel multiplexing, According to The number of streaming channels and storage channels deployed on the device can be dynamically adjusted. .

[0025] 3. Generate nearest neighbor solutions After the initial solution is obtained, the initial solution is defined as follows: This allows the component to participate in the generation of subsequent nearest-neighbor solutions. In this stage, the current layout and routing solution will be changed using the following three adjustment strategies: Strategy 1: Randomly select a component and change its position; Strategy 2: Randomly select a component and rotate it clockwise. Strategy 3: Randomly select two components to swap their positions. In DCSA (Distributed Dynamic Automated Storage and Retrieval), the location of the storage channel has a crucial impact on the flow path length of both fluid storage and retrieval tasks. Simultaneously, the volume of the storage channel directly determines its length. If its location is improperly chosen, the flow paths of other stream processing tasks may be obstructed, forcing them to choose longer paths. Therefore, this invention abstracts the storage channel as a [length missing]. A component with a width of 1 unit participates in the first two adjustment strategies mentioned above. The generation of the nearest neighbor solution is defined as... It will choose an adjustment strategy with equal probability and make corresponding adjustments.

[0026] Before implementing the adjustment strategy, the flow paths of the selected components and other flow paths that intersect with them should be removed. Moving, swapping, or rotating components may cause overlap with other components, resulting in invalid generated solutions; such schemes need to be discarded. After implementing the adjustment strategy, the removed flow paths need to be rerouted according to the storage-driven flow path planning algorithm.

[0027] 4. Rescheduling In generation During the process, removing some flow paths and adjusting the strategy, and then rearranging these flow paths, may cause resource conflict issues. In addition, the actual total completion time of the biochemical reaction needs to be determined by calculating the start time and completion time of each flow processing task. Therefore, the above problems need to be solved in the rescheduling phase. In the rescheduling phase, the algorithm of this invention determines the prerequisite tasks for each stream processing task based on the timing diagram. That is, the current stream processing task can only be executed after all prerequisite tasks are completed. After calculating the prerequisite tasks, the actual completion time of each stream processing task needs to be recalculated. If the current stream processing task... start time Earlier than the maximum completion time of all prerequisite tasks Therefore, the start time of the task needs to be postponed to [date to be filled in]. After that. And then regarding... If there are overlapping tasks in the flow paths, their start times need to be postponed to avoid resource conflicts. After. For End time The calculation formula is as follows:

[0028] in, for Execution time, for fluid volume, For the unit volume of the flow channel, express The length of the flow channel, This indicates the flow velocity of the fluid in the flow channel.

[0029] 5. Cost Function After the rescheduling is completed, a comparison is required. and Regarding the quality of the solution, this invention uses the following formula as an evaluation function:

[0030] in, This represents the completion time of the biochemical reaction at the current nearest solution. The total length of the flow channel. The number of intersections, For three given constants.

[0031] like Having more Lower cost value, i.e. If so, then accept the solution. During the generation of nearest neighbor solutions, it may be possible to... It has higher cost value, that is To enhance global search capabilities and avoid getting trapped in local optima, simulated annealing algorithms must use probabilistic... To accept the worse solution, among which .

[0032] This invention uses the results of four benchmark tests of real-world biochemical reactions and five synthetic benchmark tests to validate the proposed algorithm. Table 1 shows the basic information of the above benchmark tests, where column 2 represents the number of operands included in each benchmark test, column 3 represents the components allocated to each benchmark test in the format (mixer, heater, filter, detector, separator), columns 4 and 5 represent the maximum storage volume and number of storage cycles required for each benchmark test, respectively, column 6 represents the completion time of each benchmark test under high-level synthesis, and column 7 represents the chip area required for each benchmark test.

[0033] Table 1. Detailed information on the benchmarks used.

[0034] The comparative literature implements a CFMBs architecture for distributed channel storage, but does not take into account actual fluid volume management. In order to verify the effectiveness of the algorithm of the present invention, the present invention introduces actual fluid volume management on the basis of the algorithm idea of ​​the comparative literature for comparison.

[0035] Table 2 shows the comparison results between the methods of the present invention and the modified comparative literature in terms of channel length and number of intersections. It can be seen that the method of the present invention optimizes the channel length by an average of 21.57%, with a maximum optimization of 32.35% on the benchmark instance Synthetic5. Regarding the number of intersections, the method of the present invention optimizes by an average of 33.12%, with a maximum optimization of 40.00% on the benchmark instance IVD1. This optimization is a result of comprehensively considering placement and routing, resulting in a more compact chip architecture. Furthermore, it benefits from the careful consideration of the initial solution and the strategy of abstracting memory channels into components for adjustment.

[0036] Table 2. Comparison results with comparative literature on flow channel length and number of intersections.

[0037] Therefore, to verify the impact of the memory-driven flow path planning algorithm introduced in the initial unrouting stage on the final result, a routing scheme that does not consider memory priority was designed based on the method of this invention. Table 3 shows the comparison between the method of this invention and the routing sequence that does not consider memory priority under the same parameters and number of iterations. Since the three benchmark test cases IVD1, IVD2, and IVD3 do not involve memory, only the remaining six benchmark test cases are compared here. It can be seen that the memory-priority routing algorithm of this invention has an average optimization of 18.96% in flow path length and 12.35% in the number of intersections.

[0038] Table 3 Comparison results of routing algorithms without considering storage priority in terms of flow path length and number of intersections.

[0039] Regarding storage channels, to illustrate the impact of storage channel location on flow channel length and the number of crossovers, the method of this invention abstracts storage channels as components during the simulated annealing solution generation stage, continuously rotating and translating them during the annealing process to optimize the final result. Table 4 shows a comparison between the method of this invention and a strategy that does not consider storage channels in the adjustment. Experimental results show that the method of this invention achieves an average optimization of 9.58% in flow channel length and an average optimization of 14.99% in the number of crossovers. In particular, in benchmark instances with a high number of storage iterations, such as Synthetic3 and Synthetic5, the method of this invention can achieve optimizations of 20.00% and 12.21% in the number of crossovers, respectively.

[0040] Table 4 Comparison results of flow channel length and number of crossovers with those without considering storage channel adjustment strategies.

[0041] The method of this invention also uses the completion time of the biochemical reaction as an important indicator for evaluating the physical design results during the simulated annealing process. Therefore, Tables 5-7 show the comparison of the completion time of the biochemical reaction with the modified comparative literature method, the flow path planning algorithm without considering storage drive, and the method without considering storage channel adjustment, respectively. It can be seen that the method of this invention achieves an average optimization of 20.33%, 16.49%, and 8.75% respectively compared with the above three methods.

[0042] Table 5. Comparison results of biochemical reaction completion time with the revised comparative literature.

[0043] Table 6. Comparison of biochemical reaction completion time with flow path planning algorithms that do not consider memory-driven approaches.

[0044] Table 7 Comparison of biochemical reaction completion time with that without considering storage channel adjustment

Claims

1. A microfluidic biochip physical design method considering volume management and channel storage, characterized in that, Includes the following steps: Step 1: Initial layout stage; Set a component-intensive expansion area Used in the middle of the chip to constrain unplaced components. Inside; Then calculate each component. The priority is calculated using the following formula: in, Representation Component With components The number of connections between them Representation Component With components The number of connections between them Indicates the allocated component library; Insert all components into the priority queue, and then dequeue them in sequence. inside, if If there is not enough space to place the current component, it needs to be expanded. Until the placement conditions are met; Step 2: Initial unrouting stage; Step 2-1: Set up the flow processing tasks, including fluid input task, excess fluid removal task, waste fluid removal task, source input task, fluid storage task, and fluid extraction task; Step 2-2: Construct storage channels. First, sort the stream processing tasks in ascending order based on their completion time. Then, group the stream processing tasks using the scan-line algorithm, with tasks within the same group executed in parallel. After grouping, route the stream processing tasks within each group sequentially. Before routing, reduce the cost of previously routed grid cells. Next, determine if the stream ports or waste ports bound to the stream processing task have been placed on the port layer. If not, select an available location on the port layer such that the stream port or waste port is closest to the component of its bound stream processing task. Finally, construct the corresponding stream channel using the A* algorithm based on the type of stream processing task. During the stream channel construction process, the A* algorithm continuously selects the grid cell with the lowest routing cost for routing. The current stream processing task... The flow path corresponding to the flow channel to the cell mesh Cost function The calculation is as follows: In formula (2), for Manhattan distance to the target location and distance from the starting location to the target location The sum of Manhattan distances; in formula (3), for During wiring The current cost, for The initial cost, For one variable; For all currently wired stream processing tasks, the flow path passes through... The number of tasks; For the process The number of storage channels; It is an infinite quantity; The size will depend on The system will dynamically adjust based on the wiring conditions. Case 1: If If there is no flow path routed here, then for Scenario 2: In the same group of stream processing tasks, to avoid resource conflicts between parallel stream processing tasks within the same group, after a stream processing task completes its routing, the cost of the mesh cells it has routed needs to be set to... Right now ,otherwise, According to The number of streaming channels and storage channels deployed on the device can be dynamically adjusted. ; Step 3: Generate nearest neighbor solutions; After the initial solution is obtained, the initial solution is defined as follows: This allows them to participate in the generation of subsequent nearest-neighbor solutions; The current layout and routing solution can be changed using the following three adjustment strategies: Strategy 1: Randomly select a component and change its position; Strategy 2: Randomly select a component and rotate it clockwise. Strategy 3: Randomly select two components and swap their positions; Abstract the storage channel into a length of... A component with a width of 1 unit participates in both Strategy 1 and Strategy 2; the generation of nearest neighbor solutions is defined as... Each adjustment strategy is selected with equal probability and adjusted accordingly. Before implementing the adjustment strategy, the flow paths of the selected components and other flow paths that intersect with these flow paths are removed; after the adjustment strategy is implemented, the removed flow paths are rerouted according to the flow path planning algorithm of the storage driver. Step 4: Rescheduling; Based on the timing diagram, determine the prerequisite tasks for each stream processing task; that is, the current stream processing task can only be executed after all prerequisite tasks are completed. After calculating the prerequisite tasks, the actual completion time of each stream processing task needs to be recalculated. If the current stream processing task... start time Earlier than the maximum completion time of all prerequisite tasks Therefore, the start time of the task needs to be postponed to [date to be filled in]. After that; and then regarding and If there are overlapping tasks in the flow paths, their start times need to be postponed to avoid resource conflicts. After; for End time The calculation formula is as follows: in, for Execution time, for fluid volume, For the unit volume of the flow channel, express The length of the flow channel, This indicates the velocity of the fluid in the flow channel; Step 5: Cost function; After the rescheduling is completed, a comparison is required. and The quality of the solution is evaluated using the following formula: in, This represents the completion time of the biochemical reaction at the current nearest solution. The total length of the flow channel. The number of intersections, For three given constants; like Having more Lower cost value, i.e. If the solution is correct, then accept the solution; however, during the generation of nearest neighbor solutions, it may cause... It has higher cost value, that is To enhance global search capabilities and avoid getting trapped in local optima, simulated annealing algorithms must use probabilistic... To accept the worse solution, among which .

2. The microfluidic biochip physical design method considering volume management and channel storage according to claim 1, characterized in that, The initial area of ​​the component dense expansion region is: , and For the length and width of the chip, and It is an integer greater than 1.

Images