A digital model-based filter topology iterative design method

Abstract

The application discloses a filter topology structure iterative design method based on a digital model and particularly relates to the field of computer-aided circuit design, and is used to solve the problem that after multiple rounds of topology adjustment, it is difficult to accurately limit the current topology modification range due to the historical structure change; by obtaining a filter digital model sequence and a simulation response, target deviation distance, topology difference data and target deviation change data are formed at common frequency sampling positions, connection changes of continuously associated shared connection endpoints are merged into local topology change units and local structure maintenance intervals are determined, local change response records are formed, then continuous improvement coverage sections and shared node incremental coverage sections are formed according to a response section to be improved, candidate topology modification ranges are determined by using an overlapping state and candidate digital models are formed, the candidate digital models are simulated, new design reference digital models are determined by combining residual deviation amounts of the section to be improved and total target deviation amounts, and filter digital model sequences are updated.

Description

Technical Field

[0001] This invention relates to the field of computer-aided circuit design, and more specifically, to an iterative design method for filter topology based on digital models. Background Technology

[0002] In the process of computer-aided design of filters, a digital model representing the resonant unit and its connection relationship is usually established first. Then, the digital model is simulated according to a unified frequency range. The difference between the simulation response and the design target is used to determine the local structure that needs to be adjusted, and the local connection relationship is changed in the digital model. This forms a new filter topology and the design comparison continues.

[0003] Continuous topology adjustments will form a sequence of filter digital models arranged according to the design rounds. The connection relationship between adjacent digital models can be clearly reflected in the current round of changes. The simulation response of each digital model can reflect the changes in the distance of each frequency position from the design target after the structural changes in the current round. Therefore, the completed digital model sequence actually preserves the process information between local topology changes and frequency response changes.

[0004] In actual topology iteration, the same design round may change multiple connections located in the same local region. The response improvement caused by a change in a local connection may continue to exist while the local structure is maintained. When a connection change occurs again near a shared node, the same response segment may also generate new improvements. If all response changes within a longer iteration range are recorded directly according to a single connection, the response effect after the change in the local structure state is likely to continue to be attributed to the original connection object, and the scope of topology modification will expand or shift accordingly.

[0005] Therefore, after multiple rounds of topology adjustments, how to clarify the interaction between historical structural changes and response changes from the already formed filter digital model sequence, and accurately limit the current topology modification range, remains a problem that needs to be solved in filter topology iterative design. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a filter topology iterative design method based on digital models, which addresses the problem that historical structural changes after multiple rounds of topology adjustments are difficult to accurately define the current topology modification range. By acquiring the filter digital model sequence and simulation response, target deviation distance, topology differential data, and target deviation change data are formed at the common frequency sampling location. Continuously associated connection changes at shared connection endpoints are merged into local topology change units, and local structure preservation intervals are determined, forming local change response records. Then, based on the response segment to be improved, a continuously improved coverage segment and a shared node incremental coverage segment are formed. The overlapping state is used to determine the candidate topology modification range and form candidate digital models. The candidate digital models are simulated, and a new design baseline digital model is determined by combining the remaining deviation of the segment to be improved and the total target deviation, thus updating the filter digital model sequence and solving the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] S1: Obtain the filter digital model sequence and simulation response, determine the target deviation distance at the common frequency sampling position, and compare adjacent filter digital models to form topological differential data and target deviation change data;

[0009] S2: Merge connection changes in the same design round into local topology change units based on shared connection endpoints, track the connection relationships of nodes involved in the local topology change units to determine the local structure preservation interval, and associate target deviation change data to form a local change response record;

[0010] S3: Determine the response segment to be improved based on the current filter digital model, determine the candidate local topology change unit from the local change response record, form the continuous improvement coverage segment and the shared node incremental coverage segment, determine the candidate topology modification range based on the overlap state of the two types of coverage segments and the response segment to be improved, and form the candidate digital model within the candidate topology modification range;

[0011] S4: Simulate candidate digital models, determine new design baseline digital models based on target deviation distance, and update filter digital model sequences based on the new design baseline digital models.

[0012] Furthermore, topological difference terms that share a common connection endpoint and are continuously associated through the common connection endpoint in the same design round are merged to form local topological change units. Starting from the design round in which the local topological change unit occurs, the connection relationships of the nodes involved are checked round by round. The design round in which any node undergoes a connection change again is determined as the new local structural change starting point. The continuous design rounds from the design round in which the local topological change unit occurs to the new local structural change starting point are determined as the local structural maintenance interval.

[0013] Furthermore, the filter digital model immediately adjacent to the candidate local topology change unit before the design round is determined as the first comparison benchmark. Common frequency sampling positions where the target deviation distance after the candidate local topology change unit occurs is less than the target deviation distance of the first comparison benchmark at the same common frequency sampling position are merged to form a direct improvement segment. Within the local structure preservation interval, the target deviation distance of each common frequency sampling position in the direct improvement segment is compared with the target deviation distance of the first comparison benchmark at the same common frequency sampling position in each design round. Common frequency sampling positions that are continuously less than the target deviation distance of the first comparison benchmark at the same common frequency sampling position are retained to form a continuously improved coverage segment.

[0014] Furthermore, along the filter digital model sequence, the shared node local topology change unit involved in the node connection relationship of the first candidate local topology change unit is identified. The filter digital model immediately adjacent to it before its occurrence is determined as the second comparison benchmark. The common frequency sampling positions where the target deviation distance after its occurrence is less than the target deviation distance of the second comparison benchmark at the same common frequency sampling position are merged to form an incremental improvement segment. Within its corresponding local structure preservation interval, the common frequency sampling positions where the target deviation distance is continuously less than that of the second comparison benchmark at the same common frequency sampling position are retained to form a shared node incremental coverage segment.

[0015] Furthermore, the design baseline digital model located at the current end of the filter digital model sequence is determined as the current filter digital model. Common frequency sampling positions in the current filter digital model with target deviation distance greater than zero and consecutive sequence numbers are merged to form the response segment to be improved. The target deviation change data corresponding to the rounds of local topology change units in each local change response record are checked. Local topology change units with common frequency sampling positions where the target deviation change value is less than zero within the response segment to be improved are determined as candidate local topology change units.

[0016] Furthermore, the nodes involved in the candidate topology modification range are matched according to the node identifier of the current filter digital model, and node pairs consisting of two different nodes are determined one by one; node pairs in the connected state are changed to the disconnected state, and node pairs in the disconnected state are changed to the connected state. The formation of each candidate digital model only changes the connection state of one node pair while maintaining the connection states of the rest. Candidate digital models are formed by different single local connection relationship changes.

[0017] Furthermore, within the response segment to be improved corresponding to each candidate digital model, the target deviation distance is accumulated according to the frequency interval of the common frequency sampling position to form the remaining deviation of the segment to be improved. Candidate digital models whose remaining deviation of the segment to be improved is less than the remaining deviation of the current filter digital model in the same response segment to be improved are determined as valid candidate digital models. The current filter digital model and the valid candidate digital models constitute the design baseline candidate set. The target deviation distance is accumulated at all common frequency sampling positions to form the total target deviation. The digital model corresponding to the smallest total target deviation is determined as the new design baseline digital model.

[0018] Furthermore, each connection relationship is represented by a node pair consisting of two different nodes and the connection state of the node pair. The connection state includes the existence of connection and the disconnection of connection. Node pairs with the same node identifier in adjacent filter digital models are compared, and node pairs whose connection state changes are identified as topology difference terms. The connection endpoints, connection state changes, and subsequent design rounds of the topology difference terms are then established to form a correspondence. All topology difference terms formed by the same pair of adjacent filter digital models constitute topology difference data.

[0019] Furthermore, candidate digital models are established for different single local connectivity changes within the candidate topology modification range. Each candidate digital model is associated with the response segment to be improved within the candidate topology modification range and the node pair where the connectivity state change occurs. Simulations are performed on each candidate digital model using the same simulation response type, port definition, excitation conditions, and common frequency sampling position as the current filter digital model. The target deviation distance corresponding to each common frequency sampling position is then determined according to the method for determining the target deviation distance.

[0020] Furthermore, the connection relationship between the new design baseline digital model and the current filter digital model is compared. If there are differences in the local connection relationship formed in this design round, the new design baseline digital model and its simulation response are added to the filter digital model sequence according to the new design round, and it is determined as the current filter digital model for the next design round. When the new design baseline digital model is the current filter digital model, the current filter digital model is retained as the design result of this topology iteration.

[0021] The technical effects and advantages of the filter topology iterative design method based on digital models of the present invention are as follows:

[0022] This invention merges connection changes in the same design round according to shared connection endpoints, and uses the change in the connection relationship of the involved nodes as the end boundary of the local structure preservation interval. This allows structural changes that occur together in the same local area to participate in response analysis according to the actual topological relationship, reduces the mixing of the scope of action between different local topological states, and makes the corresponding range of historical structural changes and target deviation changes clearer.

[0023] This invention tracks the persistent improvement locations after the formation of local topology changes and the incremental improvement locations generated by local topology changes of shared nodes, and uses the overlap of the two types of coverage segments and the response segments to be improved to delineate the candidate topology modification range. This concentrates the current structural adjustment on the local connection regions that have already undergone target-direction changes in the same frequency range, reducing the computational burden of repeatedly generating and verifying candidate structures in a large topology range.

[0024] This invention first filters out candidate digital models that do not improve the overall response of the current segment by using the remaining deviation of the segment to be improved. Then, it uses the total target deviation to compare the overall deviation in the complete design target frequency domain. This allows the new design benchmark digital model to take into account both the current problem frequency band and the complete design target, reducing the situation where structures that improve local frequency bands but increase deviations in other frequency positions are continued to be used, and keeping the sources of structural and response changes clear between design rounds. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of an iterative design method for filter topology based on a digital model.

[0026] Figure 2 To continuously improve the coverage area and the incremental coverage area of ​​shared nodes, a block diagram is formed;

[0027] Figure 3 A flowchart for determining the scope of candidate topology modifications for overlapping states;

[0028] Figure 4 A flowchart for screening candidate digital models and updating design benchmarks. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Please see Figure 1 This invention provides an iterative design method for filter topology based on a digital model, comprising:

[0031] S1: Obtain the filter digital model sequence and simulation response, determine the target deviation distance at the common frequency sampling position, and compare adjacent filter digital models to form topological differential data and target deviation change data;

[0032] S2: Merge connection changes in the same design round into local topology change units based on shared connection endpoints, track the connection relationships of nodes involved in the local topology change units to determine the local structure preservation interval, and associate target deviation change data to form a local change response record;

[0033] S3: Determine the response segment to be improved based on the current filter digital model, determine the candidate local topology change unit from the local change response record, form the continuous improvement coverage segment and the shared node incremental coverage segment, determine the candidate topology modification range based on the overlap state of the two types of coverage segments and the response segment to be improved, and form the candidate digital model within the candidate topology modification range;

[0034] S4: Simulate candidate digital models, determine new design baseline digital models based on target deviation distance, and update filter digital model sequences based on the new design baseline digital models.

[0035] This invention uses a sequence of filter digital models arranged according to design rounds and the simulation response of each filter digital model as the processing basis. First, the target deviation distance is determined at the common frequency sampling position, and topological differential data and target deviation change data are formed from adjacent digital models. Then, according to the shared connection endpoints, the connection changes of the same design round are merged into local topological change units. The change in the connection relationship of the involved nodes is used as the boundary of the local structure preservation interval, forming a local change response record. For the response segment to be improved in the current filter digital model, candidate local topological change units are determined from the local change response record, and continuous improvement coverage segments and shared node incremental coverage segments are formed respectively. The candidate topology modification range is determined according to the overlap state of the two types of coverage segments and the response segment to be improved, and a candidate digital model is formed within this range. After simulation, the candidate digital model is used to determine a new design baseline digital model through the remaining deviation of the segment to be improved and the total target deviation, and it is incorporated into the filter digital model sequence.

[0036] After successive design rounds of adjustments to the filter topology, a sequence of filter digital models arranged according to the design rounds has been formed. Each filter digital model has a corresponding simulation response, and the design objectives also specify the response requirements to be achieved at each frequency position. Step S1 needs to establish a comparative basis for each design round under unified design objectives and simulation conditions, and convert the changes in connection relationships and the changes in frequency positions toward or away from the design objective into data that can be directly accessed in step S2. However, only when different design rounds use the same response type, port definition, excitation conditions, and common frequency sampling position can there be a unified comparative basis between structural changes and target deviation changes. Therefore, step S1 sequentially completes the determination of the common frequency sampling position and frequency interval, the calculation of the target deviation distance, and the formation of topology differential data and target deviation change data.

[0037] A filter digital model is a data object that digitally represents the filter topology and the connection relationships required for simulation within a design cycle. It is used to store the filter structure state of that design cycle during computer-aided design. Specifically, each resonant unit in the filter is assigned a unique node identifier. Node pairs consisting of two different nodes represent the possible connection relationships between two resonant units, and the connection status of the node pairs indicates whether the corresponding connection relationship is currently in a connected state or disconnected. The filter digital model for that design cycle is organized according to the connection status of each node and node pair within that design cycle, and a design cycle identifier is determined for this filter digital model. Thus, different design cycles use the same node identifier to represent the same resonant unit, allowing adjacent filter digital models to directly compare their connection status based on node pairs.

[0038] In one implementation, a filter includes multiple resonant units arranged sequentially. A computer-aided design program determines node identifiers for each resonant unit and establishes connections between nodes based on the current design scheme. For example, in the current design round, there is a connection between the first and second resonant units, and between the second and third resonant units. Corresponding node pairs are formed in the filter digital model and recorded as connected states. When the connection between the second and third resonant units is removed in the next design round, and a connection is established between the first and third resonant units, the node identifiers of each resonant unit remain unchanged; only the connection states of the corresponding node pairs are changed. The changed node connections form the filter digital model for the next design round. Through this construction method, the structural differences between adjacent filter digital models can directly represent the actual changes in connection relationships that occur in the current design round.

[0039] The logic of step S1 is as follows:

[0040] After each filter digital model is formed, simulations are performed according to a unified simulation response type, port definition, excitation conditions, and simulation sweep position. This ensures that each filter digital model simultaneously corresponds to the topology state of a design round and the simulation response generated by that topology state. After completing one topology adjustment and obtaining a new simulation response, the new filter digital model is added to the filter digital model sequence according to the design round. Step S101 arranges these filter digital models according to the design round and organizes the simulation responses of each design round at the common frequency sampling position. Step S103 further forms topology differential data by comparing the connection states of the same node pairs in adjacent filter digital models, so that changes in the filter topology and corresponding changes in the simulation response can be established according to the same design round.

[0041] S101: Determine the common frequency sampling location and organize the design round data:

[0042] The filter digital models are arranged according to their design cycle identifiers, so that adjacent positions correspond to two consecutive design cycles. The design objective is determined by the current filter design task before the topology iteration begins, and the target response value, allowable response range, or response boundary that belongs to the same response quantity as the simulation response is given within the frequency range covered by the design objective. Each filter digital model uses the same node identifier system, the same simulation response type, the same port definition, the same excitation conditions, and the same set of simulation sweep positions to obtain the simulation response. The frequency range covered by the design objective is matched with the unified simulation sweep positions, and the sampling positions that simultaneously have the design objective and the simulation response are determined as common frequency sampling positions. The common frequency sampling positions are then arranged in a fixed order from low to high frequency.

[0043] After the common frequency sampling positions are determined, the response values ​​corresponding to each common frequency sampling position are obtained from the simulation response of each filter digital model in ascending order of frequency, so that the same common frequency sampling position always represents the same frequency position in different design rounds. In order to make step S4 adopt a unified deviation accumulation rule for different sampling methods, the corresponding frequency interval is determined for each common frequency sampling position. The midpoint between two adjacent common frequency sampling positions is used as the boundary of the frequency range represented by the adjacent sampling positions. The frequency interval corresponding to the common frequency sampling position located inside is the frequency width between its left boundary and its right boundary. The frequency interval corresponding to the common frequency sampling positions located at both ends of the design target coverage area is the frequency width between the endpoint and the midpoint of the adjacent common frequency sampling positions. Each filter digital model uses the same set of frequency intervals, and the frequency intervals are directly used by step S4 to accumulate the remaining deviation of the section to be improved and the total target deviation.

[0044] In one implementation, the filter digital model undergoes continuous topology adjustments to form several design cycles. One design cycle changes the connection relationships between resonant units, while another design cycle continues to adjust other local connections. Each simulation cycle obtains a response at the same frequency sweep position. The computer-aided design program arranges these filter digital models sequentially according to the design cycles and then places the simulation response of each cycle at the same common frequency sampling position, allowing direct comparison of response changes under different topology states at the same frequency position.

[0045] S102: Calculate the target deviation distance of each filter's digital model:

[0046] After determining the common frequency sampling locations and the arrangement of design rounds, the digital models of the filters are processed one by one. At each common frequency sampling location, the simulation response of the filter's digital model is obtained, and the target response requirement given by the design objective at the same common frequency sampling location is obtained. Then, the target deviation distance is calculated according to the expression form of the design objective. The target deviation distance always represents the response gap between the simulation response and the design objective, and its dimension is the same as the response dimension used in the simulation response.

[0047] When the design objective provides the target response value at the common frequency sampling location, let the first... The digital models of the filters in each design round at the common frequency sampling position The simulation response is The target response value at this location is designed to be... Target deviation distance Determined according to the following formula:

[0048]

[0049] in, This indicates the design round of the filter digital model in the filter digital model sequence. This indicates the common frequency sampling positions arranged in frequency order. This indicates the sampling position at the common frequency in this design round. The simulated response, Indicates the design objective at the common frequency sampling location. The given target response value, This indicates the sampling position at the common frequency in this design round. The target deviation distance.

[0050] When the design objective provides an allowable response range at a common frequency sampling location, the lower and upper boundaries of the allowable response range at that common frequency sampling location are obtained. When the simulated response is within the allowable response range, the target deviation distance is set to zero. When the simulated response is below the lower boundary of the allowable response range, the distance from the simulated response to the lower boundary is set to the target deviation distance. When the simulated response is above the upper boundary of the allowable response range, the distance from the simulated response to the upper boundary is set to the target deviation distance. When the design objective only provides the upper boundary of the response, the target deviation distance is set to zero when the simulated response is within the upper boundary and to zero when it exceeds the upper boundary. When the design objective only provides the lower boundary of the response, the target deviation distance is set to zero when the simulated response reaches or exceeds the lower boundary and to zero when it is below the lower boundary. After calculating all common frequency sampling locations, each filter digital model obtains a target deviation distance corresponding to each common frequency sampling location.

[0051] For example, when the filter design goal is to use an allowable response range in the passband and a response boundary requirement in the stopband, the computer-aided design program determines the target range of the simulated response at each frequency position. The frequency positions that have fallen into the allowable range are recorded as the zero target deviation distance, and the frequency positions that are still outside the allowable range are calculated according to the nearest target boundary. In this way, even when the passband and stopband use different target expression methods, they can still participate in the design round comparison through the same target deviation distance.

[0052] S103: Generate topological difference data and target deviation change data:

[0053] After the target deviation distance of each filter digital model is determined, two adjacent design rounds are selected sequentially for comparison according to the filter digital model sequence. The same filter digital model sequence adopts a unified node identification system. Each connection relationship is represented by a node pair consisting of two different nodes and the current connection status of the node pair. The connection status is used to indicate whether there is a connection between the node pairs or whether the node pairs are in a disconnected state. For two adjacent design rounds, the nodes in the two filter digital models are first matched according to the node identification, and then the connection status is compared according to the node pairs. The node pairs whose connection status changes from disconnected to connected or from connected to disconnected are identified as topology difference terms. The connection endpoints, connection status changes and the subsequent design rounds are then established to correspond to the topology difference terms generated by the same pair of adjacent filter digital models. All topology difference terms generated by the same pair of adjacent filter digital models constitute the topology difference data of the design round.

[0054] For example, when a design cycle establishes a new connection between two existing resonant units and simultaneously decouples another connection that shares one of the resonant units, the two connection changes each form a topology difference term, which is then collectively attributed to the topology difference data of that design cycle. Step S1 maintains the connection endpoints and design cycle relationships of the two connection changes during this stage. Step S2 continues to merge locally changed topology units based on whether these connection endpoints share nodes.

[0055] For the target deviation distance, the target deviation distance is compared one by one between the digital models of the same pair of adjacent filters according to the common frequency sampling position. The target deviation distance of the later design round is subtracted from the target deviation distance of the previous design round to obtain the target deviation change value at the corresponding common frequency sampling position:

[0056]

[0057] in, Indicates the first The design round is relative to the first Each design round at a common frequency sampling location The target deviation change value, Indicates the first The target deviation distance for each design round Indicates the first The target deviation distance for each design cycle. When the target deviation change is less than zero, the common frequency sampling position moves closer to the design target after the topology change in this cycle; when the target deviation change is equal to zero, the distance from the design target remains unchanged; when the target deviation change is greater than zero, the distance from the design target increases. All target deviation changes are retained in a predetermined order of the common frequency sampling positions to form the target deviation change data for this design cycle.

[0058] The topology differential data and target deviation change data are established in correspondence according to the same design round, so that each design round simultaneously records how the connection relationship has changed in this round and how the distance between the sampling positions of each common frequency and the design target has changed in this round. Step S2 directly calls the topology differential data of each design round, merges them into local topology change units based on the shared connection endpoints, and calls the target deviation change data within the same design round and the local structure preservation interval to form a local change response record; the target deviation distance of each filter digital model is further determined by step S3 to determine the response segment to be improved, and step S4 calculates the remaining deviation of the segment to be improved and the total target deviation.

[0059] After step S1 is completed, the filter digital model sequence has formed a directly comparable data foundation under a unified node identification system, simulation response type, port definition, excitation conditions, and common frequency sampling positions. Each common frequency sampling position has a defined frequency interval, and each filter digital model has a target deviation distance determined for each frequency position. Adjacent design rounds generate corresponding topology difference data and target deviation change data. The topology difference data is used in step S2 to merge local topology change units according to shared connection endpoints. The target deviation change data is used in step S2 to establish local change response records. The target deviation distance is used in step S3 to determine the response segment to be improved. The frequency interval and target deviation distance corresponding to the common frequency sampling position are used in step S4 to calculate the remaining deviation and total target deviation of the segment to be improved.

[0060] The logic of step S2 is as follows:

[0061] Multiple connection changes in the same design round may form the same local area by sharing connection endpoints. It is necessary to first merge them according to the actual connection relationship, then limit the continuous design rounds that maintain the same connection state along the filter digital model sequence, and establish the index relationship between the design round and the corresponding filter digital model. This forms a local change response record that can be directly obtained in step S3 for target deviation change data and target deviation distance.

[0062] S201: Merge local topology change units according to shared connection endpoints:

[0063] For the topology difference data generated in each design round in step S1, the topology difference items are obtained round by round according to the design round. Each topology difference item has clear connection endpoints and changes in connection relationships. The computer-aided design program takes the topology difference items within the same design round as the objects to be processed in this round, and performs connected component partitioning according to the shared relationships between the connection endpoints, so that the connection changes that occur together in the same local connection region form a structural processing object.

[0064] Specifically, each topological difference item in the same design round is used as the basic object for connected component partitioning. For any two topological difference items, their connection endpoints are compared. If two topological difference items have at least one common connection endpoint, a direct association is established between them. If the first topological difference item shares a connection endpoint with the second topological difference item, and the second topological difference item shares a connection endpoint with the third topological difference item, the three topological difference items are assigned to the same connected component along the continuous association relationship formed by the shared connection endpoints. After completing the association check of all topological difference items in this round, each connected component corresponds to a local topological change unit. Connection changes belonging to different connected components within the same design round form different local topological change units.

[0065] In practical implementation, a disjoint-set data structure (DFS) algorithm can be used to partition connected components. First, an independent set is established for each topological difference item. Then, the connection endpoints are compared item by item. Topological difference items with shared connection endpoints are merged into sets until all topological difference items in this round have been compared. Finally, topological difference items with the same root node are merged into the same local topological change unit. After the local topological change unit is formed, the nodes involved in the local topological change unit are obtained according to the connection endpoints of all topological difference items within it. These nodes serve as the objects for S202 to track the local connectivity state.

[0066] For example, in a design round, two connections might be changed simultaneously. One connection is between the first and second resonant units, and the other is between the second and third resonant units. These two connection changes form a continuous relationship through the second resonant unit. The computer-aided design program will merge these two connection changes into the same local topology change unit. If, in the same design round, another connection change occurs between a different set of resonant units, and its connection endpoints do not have a continuous sharing relationship with the nodes involved in this local topology change unit, then this connection change will form another local topology change unit. Through this process, connection changes occurring together in the same local region are tracked as a whole according to the actual topological connections.

[0067] S202: Determine the local structural retention interval along the design cycle:

[0068] Once the local topology change units are determined, the computer-aided design program traces the connection relationships of the nodes involved in each local topology change unit in the filter digital model sequence according to the design round in which each local topology change unit occurs. The tracing process is executed round by round in ascending direction of the design round, and the topology difference data of each design round formed in step S1 is used to determine whether a new connection state has entered the vicinity of these nodes.

[0069] For an event that occurs in the first The local topology change unit of the design round will be the first The design round is determined as the design round in which the current connection state of the local topology change unit takes effect, and all nodes involved in the local topology change unit are obtained; from the first design round... Starting with the first design round, the topology difference data of each design round is checked sequentially, and the connection endpoints of each topology difference item in that design round are matched with the nodes involved in the local topology change unit. When at least one topology difference item in a design round has any connection endpoint belonging to a node involved in the local topology change unit, that design round is determined as the design round for that local region to enter the next connection state, and the first design round is set as follows: The consecutive design rounds from the current design round to the previous design round are defined as the local structure preservation interval.

[0070] When no new connection changes involving the node of the local topology change unit have occurred in the currently recorded design rounds of the filter digital model sequence, the local structure preservation interval extends to the design round at the current end of the filter digital model sequence. Thus, the local structure preservation interval always corresponds to a definite local connection state, its starting point determined by the round in which the local topology change unit is formed, and its ending point determined by the location where a connection change occurs again within the same local node range.

[0071] For example, after a local topology change unit is formed, several consecutive design rounds only adjust other local regions of the filter. The connection relationships involving the nodes of this local topology change unit remain unchanged, and these design rounds are continuously classified into the same local structure preservation interval. When a design round changes the connection relationship involving one of the nodes again, this design round becomes the starting point of a new local structure change, and the original local structure preservation interval ends before this design round. Based on this, step S3 can distinguish the model range in which a local connection state continues to exist and the time position of the shared node entering the next connection state.

[0072] S203: Establish a local change response record:

[0073] After each local topology change unit obtains its local structure preservation interval, the computer-aided design program organizes the local topology change unit, the local structure preservation interval, and the target deviation change data formed in step S1 into a local change response record according to the design round correspondence. This organization process uses the local topology change unit as an index, enabling a local structure object to correspond to its own formation position, preservation range, and the target deviation change process that has occurred within that preservation range.

[0074] Specifically, first, the design round in which the local topology change unit occurs is determined. Then, the filter digital model immediately preceding the design round is identified in the filter digital model sequence, and a correspondence is established between the filter digital model and the local topology change unit. According to the design rounds included in the local structure preservation interval, an index relationship is established between each design round and the filter digital model of the same design round in the filter digital model sequence. Then, the target deviation change data formed in step S1 is obtained according to the design round order. A correspondence relationship is established between the connections and nodes involved in the local topology change unit, the local structure preservation interval, the filter digital model before the local topology change unit occurs, the filter digital model corresponding to each design round within the local structure preservation interval, and the target deviation change data, forming a local change response record. In step S3, according to the design round index in the local change response record, the target deviation distance of the same design round and the same common frequency sampling position is obtained from the target deviation distance of each filter digital model formed in step S1.

[0075] The design cycle relationship in the local change response record is directly established according to the filter digital model sequence. For example, if a local topology change unit is formed in one design cycle and its local structure preservation interval covers multiple consecutive design cycles, then the record sequentially saves the target deviation change data corresponding to these design cycles. Step S3 can use this to track the change of target deviation distance with design cycles according to the sampling position at the same common frequency. At the same time, the local structure preservation interval determined in S202 limits the model range of the record for structural action tracking. When a new local topology change unit is formed near the shared node, the new local topology change unit re-establishes the corresponding local change response record according to its own formation cycle.

[0076] In one implementation, once a local topology change unit is formed, the connection relationships of the nodes involved remain unchanged, while other locations of the filter undergo several topology adjustments. The computer-aided design program establishes a correspondence between the target deviation change data of each design round within the local structure maintenance interval and the local topology change unit in chronological order. When a connection change occurs again near the shared node, the maintenance range of the original local change response record ends before the change. The newly occurring connection change redefines the local topology change unit according to S201, and then determines a new local structure maintenance interval according to S202. This process enables step S3 to form a continuously improved coverage segment and a shared node incremental coverage segment by utilizing the response trajectory during the duration of a local connection state and the response trajectory after entering the next connection state near the shared node, respectively.

[0077] After step S2 is completed, the connection changes of continuous local regions formed by shared connection endpoints in the same design round have been merged into local topology change units. Each local topology change unit obtains a local structure preservation interval defined by the position of the next change in node connection relationship, and forms a local change response record indexed by the local topology change unit. The local change response record simultaneously maintains the correspondence between the filter digital model before the local topology change unit occurs, the filter digital model corresponding to each design round within the local structure preservation interval, and the target deviation change data. Based on this, step S3 first filters candidate local topology change units in the response segment to be improved, and then directly obtains the target deviation distance of the corresponding design round to form a continuous improvement coverage segment and a shared node incremental coverage segment.

[0078] The logic for step S3 is as follows:

[0079] The response segment to be improved in the current filter digital model still needs to first screen out candidate local topology change units that have previously generated target-direction changes in this frequency range from the historical iteration records. Then, the improvement positions continuously left by the candidate local topology change units and the newly added improvement positions of the shared node local topology change units are tracked. Based on this, step S3 determines the target local topology object, the candidate topology modification range, and the candidate digital model that establishes a corresponding relationship with the response segment to be improved. The following processing is completed by the computing equipment running the filter digital model and simulation program.

[0080] S301: Determine the response segment to be improved and establish a frequency location comparison range:

[0081] The filter digital model that was determined as the design benchmark when entering this design round and is located at the current end of the filter digital model sequence is identified as the current filter digital model. The target deviation distance formed in step S1 of the current filter digital model is obtained, and the common frequency sampling positions are checked one by one from low to high. The common frequency sampling positions with a target deviation distance greater than zero are identified as the frequency positions that still need to be adjusted. When two common frequency sampling positions have adjacent numbers in the common frequency sampling position sequence formed by frequency arrangement, the two sampling positions are identified as continuous frequency positions. The common frequency sampling positions that satisfy the condition of a target deviation distance greater than zero and consecutive numbers are merged into a response segment to be improved. When the current filter digital model forms multiple separate response segments to be improved, local topology action tracing is performed for each response segment to be improved. When the target deviation distances corresponding to all common frequency sampling positions of the current filter digital model are zero, the current filter digital model is identified as the design result of this topology iteration.

[0082] For each response segment to be improved, the local change response record formed in step S2 is checked item by item according to the common frequency sampling positions contained in the response segment to be improved, and the target deviation change data corresponding to the design round of each local topology change unit is obtained; when there is a common frequency sampling position in the target deviation change data of the design round that is located in the response segment to be improved and the target deviation change value is less than zero, the local topology change unit is determined as the candidate local topology change unit corresponding to the response segment to be improved, and the filter digital model before the candidate local topology change unit occurs, the local structure preservation interval, and the filter digital model corresponding to each design round within the local structure preservation interval are obtained, so that the candidate local topology change unit originates from the historical local structure that has already undergone target direction change in the current response segment to be improved.

[0083] For example, if the current filter digital model still has a target deviation distance in a continuous stopband frequency range, the computing device will form a response segment to be improved in this continuous frequency range, and then check the local change response records saved in step S2 according to the same common frequency sampling position; if another frequency range also has a continuous target deviation, then this frequency range will form another response segment to be improved, and local topology action analysis will be performed with an independent frequency position range, so that different frequency ranges correspond to their own candidate topology modification ranges.

[0084] S302: Forming a continuously improving coverage section:

[0085] For each candidate local topology change unit determined in step S301, the local change response record corresponding to the candidate local topology change unit is obtained. The filter digital model immediately adjacent to the design round in which the candidate local topology change unit occurs is determined as the first comparison benchmark. The filter digital model corresponding to the design round in which the candidate local topology change unit occurs is obtained. The target deviation distance between the two filter digital models is compared one by one at all common frequency sampling positions. When the target deviation distance at a certain common frequency sampling position after the candidate local topology change unit occurs is less than the target deviation distance of the first comparison benchmark at the same common frequency sampling position, the common frequency sampling position is determined as the direct improvement position. Then, the direct improvement positions with consecutive serial numbers are merged to form a direct improvement segment according to the arrangement number of the common frequency sampling positions.

[0086] After the direct improvement segment is formed, the common frequency sampling positions contained in the direct improvement segment are used as the continuous tracking range. The filter digital model is obtained in each design round according to the local structure preservation interval determined in step S2. For each common frequency sampling position in the direct improvement segment, the target deviation distance of that design round is compared with the target deviation distance of the first comparison benchmark at the same common frequency sampling position. When a common frequency sampling position satisfies the condition that its target deviation distance is less than the target deviation distance of the first comparison benchmark at the same common frequency sampling position in all design rounds contained in the local structure preservation interval, that common frequency sampling position is retained. After completing the comparison of all design rounds in the local structure preservation interval, the retained common frequency sampling positions are merged according to frequency continuity to form a continuous improvement coverage segment.

[0087] Therefore, the direct improvement segment represents the target improvement position that appears when the local topology change is just formed, and the continuous improvement coverage segment further defines the portion of these improvement positions that persist during the local connection state. For example, after a local topology change unit is formed, a continuous stopband frequency position first shows a decrease in target deviation distance, and the computing device determines this frequency position as a direct improvement segment; within the local structure maintenance interval of the local topology change unit, only a portion of the frequency positions in the direct improvement segment always maintain a smaller target deviation distance relative to the first comparison reference, then this portion of continuous frequency positions constitutes the continuous improvement coverage segment, and the local topology change unit that forms the continuous improvement coverage segment continues to participate in the overlap judgment of S303.

[0088] S303: Forming shared node incremental coverage segments and determining candidate topology modification ranges:

[0089] After the continuous improvement coverage segment is formed, the filter digital model sequence is checked again according to the end position of the local structure preservation interval given in step S2. The first local topology change unit that changes the node connection relationship of the candidate local topology change unit is obtained, and this local topology change unit is identified as the shared node local topology change unit. The shared node local topology change unit and the original local topology change unit have at least one common node. Its design round is adjacent to the original local structure preservation interval in the design round arrangement, which indicates that the same local connection area has changed from the original connection state to a new connection state. When the local structure preservation interval extends directly to the current design round, the shared node incremental coverage segment is recorded as an empty frequency set, so that the continuously improved coverage segment directly participates in the determination of the target local topology object.

[0090] When a shared node local topology change unit exists, the filter digital model immediately adjacent to the design round in which the shared node local topology change unit occurred is obtained and used as the second comparison benchmark. Then, the filter digital model after the shared node local topology change unit occurred is obtained, and the target deviation distance is compared one by one according to the common frequency sampling position. The common frequency sampling position where the target deviation distance after the shared node local topology change unit occurred is less than the target deviation distance of the second comparison benchmark at the same common frequency sampling position is determined as the incremental improvement position. The incremental improvement positions with continuous frequency are merged to form an incremental improvement segment. Subsequently, the local structure preservation interval corresponding to the shared node local topology change unit is compared round by round. For each common frequency sampling position in the incremental improvement segment, it is checked whether the target deviation distance of each design round is continuously less than the target deviation distance of the second comparison benchmark at the same common frequency sampling position. Only the common frequency sampling positions that continuously satisfy the comparison relationship within the local structure preservation interval are retained, and they are merged according to the frequency continuity relationship to form the shared node incremental coverage segment.

[0091] See Figure 2 As shown, after the continuous improvement coverage segment and the shared node incremental coverage segment are formed, each response segment to be improved is matched with the continuous improvement coverage segment and the shared node incremental coverage segment according to the common frequency sampling position. When the response segment to be improved and any coverage segment share at least one common frequency sampling position, they are determined to overlap, and the common frequency sampling positions that are shared and have consecutive sequence numbers are merged into the overlapping segment. When the response segment to be improved and any coverage segment do not share a common frequency sampling position, they are determined to separate. When the local structure preservation interval corresponding to the candidate local topology change unit extends to the current end of the filter digital model sequence but has not yet formed a shared node local topology change unit, the shared node incremental coverage segment is determined as an empty frequency set, and participates in the determination of the target local topology object according to the state of separation between the empty frequency set and the response segment to be improved.

[0092] When the response section to be improved overlaps with the continuous improvement coverage section, and the shared node incremental coverage section separates from the response section to be improved, the candidate local topology change unit forming the continuous improvement coverage section is identified as the target local topology object, and the connections and nodes involved in the candidate local topology change unit are identified as the candidate topology modification range. When the response section to be improved overlaps with both the continuous improvement coverage section and the shared node incremental coverage section, the candidate local topology change unit and the shared node local topology change unit are jointly identified as the target local topology object. The two local connection areas are connected according to the shared nodes of the two local topology change units, and the connections and shared nodes involved in the two local topology change units are jointly identified as the candidate topology modification range. When the response section to be improved overlaps with the shared node incremental coverage section, and the continuous improvement coverage section separates from the response section to be improved, the shared node local topology change unit is identified as the target local topology object, and the connections and nodes involved in the shared node local topology change unit are identified as the candidate topology modification range.

[0093] See Figure 3 As shown, for example, a continuously improved coverage segment formed by a candidate local topology change unit intersects with the current stopband response segment to be improved, while the shared node incremental coverage segment formed by a shared node local topology change unit is located in another frequency range. The computing device uses the local connection region involved by the candidate local topology change unit as the candidate topology modification range. When both types of coverage segments fall into the same response segment to be improved, the candidate topology modification range is extended along the shared node to the continuous local connection region involved by the two local topology change units. When the continuously improved coverage segment has left the current response segment to be improved, while the shared node incremental coverage segment still overlaps with the response segment to be improved, the candidate topology modification range is transferred to the local connection region involved by the shared node local topology change unit.

[0094] S304: Forming candidate digital models within the scope of candidate topology modifications:

[0095] After the candidate topology modification range is determined, the current filter digital model is used as the common structural basis of the candidate digital models. The nodes involved in the candidate topology modification range are matched according to the node identifiers of the current filter digital model. Within the candidate topology modification range, node pairs consisting of two different nodes are determined one by one. For node pairs in the current filter digital model that are in a connected state, the connection state of the node pair is changed to a disconnected state, forming a local connection relationship change. For node pairs in the current filter digital model that are in a disconnected state, the connection state of the node pair is changed to a connected state, forming a local connection relationship change. The formation of each candidate digital model only changes the connection state of one node pair, and the other connection states in the current filter digital model remain the connection states of the current design round.

[0096] For each different single local connectivity change that can be formed within the candidate topology modification range, a candidate digital model is established, and a correspondence is established between each candidate digital model and the response segment to be improved that generated the candidate topology modification range and the node pair whose state has changed this time, so that step S4 can evaluate the candidate digital model according to the response segment to be improved; after the connection state change is completed, a candidate digital model is formed according to the node identifier, connection relationship representation method and simulation configuration adopted by the current filter digital model, so that each candidate digital model has a clear difference in local connectivity relationship with the current filter digital model.

[0097] In one implementation, when the candidate topology modification range includes several connections formed around the shared resonant unit, the computing device, based on the current filter digital model, changes one of the connection relationships to form multiple candidate digital models. Each candidate digital model reflects only one local connection change, thereby enabling the response change of the candidate digital model to be clearly mapped to the local connection change after the simulation is completed in step S4, and to continue to form new topology differential data and target deviation change data.

[0098] After step S3 is completed, the current filter digital model, located at the end of the current filter digital model sequence and serving as the benchmark for this design round, has formed a response segment to be improved. Each response segment to be improved is screened using target deviation change data to obtain candidate local topology change units, and further forms a continuous improvement coverage segment and a shared node incremental coverage segment. The overlap state between the two types of coverage segments and the response segment to be improved has been used to determine the target local topology object and the candidate topology modification range. Different node pairs within the candidate topology modification range form candidate digital models that only change one connection state. Each candidate digital model establishes a correspondence with the response segment to be improved that generated the candidate topology modification range and the node pair whose state changed. Step S4 directly performs simulation and design benchmark selection on the candidate digital models according to this correspondence.

[0099] The logic for step S4 is as follows:

[0100] Step S1 has unified the common frequency sampling position, frequency interval, and target deviation distance of each filter digital model. Step S2 forms a local change response record. Step S3 determines the response segment to be improved based on this record and forms a candidate digital model within the candidate topology modification range that has only one local connection change and corresponds to the response segment to be improved. However, the candidate digital model still needs to verify whether the local connection change actually reduces the remaining deviation of the response segment to be improved under the same simulation response type, port definition, excitation conditions, and common frequency sampling position as the current filter digital model. Combined with the complete design objectives, a new design baseline digital model is determined. Therefore, step S4 completes the candidate scheme selection in the order of candidate simulation, local deviation screening, full frequency domain comparison, and design cycle update.

[0101] S401: Simulate candidate digital models and generate the remaining deviation of the section to be improved:

[0102] Each candidate digital model formed in step S3 is based on the current filter digital model. Each candidate digital model has only one change in local connectivity within the corresponding candidate topology modification range. The same simulation response type, port definition, excitation conditions, and common frequency sampling position as the current filter digital model are used to simulate each candidate digital model one by one. The simulation program uses a circuit simulation solver or an electromagnetic simulation solver that can obtain the simulation response based on the nodes and connectivity of the filter digital model, and obtains the simulation response of each candidate digital model according to the unified simulation conditions. Following the processing method of determining the target deviation distance in step S1, the target deviation distance of each candidate digital model at each common frequency sampling position is further obtained.

[0103] After the target deviation distance of the candidate digital model is formed, the response segment to be improved corresponding to the candidate digital model is generated according to step S3. The common frequency sampling positions contained in the response segment to be improved are extracted, and the frequency interval corresponding to each common frequency sampling position is obtained. The target deviation distance is accumulated sequentially according to the frequency interval to form the remaining deviation of the segment to be improved of the candidate digital model. For any digital model and response sections to be improved Remaining deviation in the section to be improved It can be represented as:

[0104]

[0105] in, The common frequency sampling position determined in step S1, For digital models At common frequency sampling location The target deviation distance at that location, The frequency interval corresponding to the sampling position of this common frequency. This refers to the response segment to be improved as determined in step S3. The target deviation distance has the response dimension used in the simulation response, and the frequency interval has the frequency dimension. Therefore, the remaining deviation of the segment to be improved is the product of the response dimension and the frequency dimension, and its value reflects the target deviation that a digital model still retains throughout the entire response segment to be improved.

[0106] The computing device uses the same frequency position and accumulation method to calculate the remaining deviation of the current filter digital model in the response segment to be improved, so that the candidate digital model and the current filter digital model have a consistent comparison range. For example, after forming several candidate digital models for a stopband response segment to be improved in step S3, the computing device performs frequency sweep simulation of these candidate digital models respectively, and accumulates the target deviation distance only at the common frequency sampling positions included in the stopband response segment to be improved. This yields the remaining response gap in the stopband segment that has not reached the design target after each individual change of a local connection relationship.

[0107] S402: Screening valid candidate digital models and forming a design basis candidate set:

[0108] After each candidate digital model and the current filter digital model have generated the remaining deviation of the segment to be improved, the computing device performs numerical comparisons item by item for the same response segment to be improved. When the remaining deviation of the segment to be improved of a candidate digital model is less than that of the current filter digital model, it indicates that the single change in local connectivity contained in the candidate digital model has reduced the cumulative target deviation of the response segment to be improved, and the candidate digital model is determined as a valid candidate digital model.

[0109] When step S3 generates candidate digital models for multiple response segments to be improved, each candidate digital model is compared according to the response segment to be improved corresponding to which it was generated, so that the effectiveness of a candidate digital model is always determined by the frequency range it targets. After the comparison of all candidate digital models is completed, the current filter digital model and all effective candidate digital models are grouped into the same design basis selection set. Each digital model in the design basis selection set retains its simulation response and the target deviation distance of each common frequency sampling position determined by the rules adopted in step S1, so that S403 can continue to compare within the complete design target coverage.

[0110] For example, multiple single-connection changes are formed within the candidate topology modification range. Some candidate digital models reduce the cumulative target deviation of the corresponding response segment to be improved, and these candidate digital models are included in the design basis selection set. Other candidate digital models change the local connection relationship, but do not achieve a smaller comparison result for the cumulative target deviation of the corresponding response segment to be improved. The computing device completes the candidate range shrinkage according to the comparison relationship of the remaining deviation of the segment to be improved, so that the comparison of the complete design target is concentrated on the digital model that has made actual corrections to the current response segment to be improved.

[0111] S403: Calculate the total target bias and update the filter digital model sequence:

[0112] After the design baseline selection set is formed, the computing device expands the comparison range from the response segment to be improved determined in step S3 to all common frequency sampling positions covered by the design target. For each digital model in the design baseline selection set, the target deviation distance at each common frequency sampling position is obtained, and these deviations are accumulated sequentially according to the frequency intervals determined in step S1 to form the total target deviation of that digital model. For any digital model in the design baseline selection set... Total target deviation It can be represented as:

[0113]

[0114] in, This represents the set of all common frequency sampling locations covered by the design target. and The target deviation distance and frequency interval determined in step S1 are used respectively. The total target deviation and the remaining deviation of the section to be improved adopt the same accumulation rule and the same dimension. The difference between the two is that the calculation range is different. The remaining deviation of the section to be improved is used to confirm the actual correction of the candidate digital model to the current response section to be improved. The total target deviation is used to compare the overall result after the change of the local connection relationship is applied to the whole design frequency range.

[0115] After calculating the total target deviation of all digital models in the design baseline candidate set, they are compared from smallest to largest. The digital model with the smallest total target deviation is determined as the new design baseline digital model, and the simulation response generated by this digital model is obtained. Through this comparison process, a candidate digital model is only compared with the current filter digital model for the complete design target coverage after the cumulative target deviation of the response segment to be improved has been reduced. This ensures that changes in local connectivity are evaluated uniformly at other common frequency sampling locations while improving the target segment.

[0116] For example, in step S402, a candidate digital model has reduced the remaining deviation of a certain stopband response segment to be improved, but the change in its local connectivity has also changed the target deviation distance near the passband. In S403, the computing device includes the stopband, passband, and other common frequency sampling positions covered by the design target into the total target deviation accumulation. When another candidate digital model also improves the stopband segment and retains a small cumulative target deviation in the entire design frequency range, the candidate digital model is determined as the new design reference digital model through the comparison of the total target deviation.

[0117] After the new design baseline digital model is determined, the connection relationship between the new design baseline digital model and the current filter digital model is compared. If there is a difference between the connection relationship corresponding to the new design baseline digital model and the current filter digital model in the local connection relationship formed in this design cycle, the new design baseline digital model and its simulation response are added to the filter digital model sequence according to the new design cycle, so that the new design baseline digital model becomes the current filter digital model in the next design cycle, and new topology differential data and target deviation change data are formed according to step S1. When the new design baseline digital model is the current filter digital model, the current filter digital model is retained as the design result of this topology iteration, so that the filter digital model sequence maintains the design cycle relationship that has been formed.

[0118] See Figure 4 As shown, after step S4, each candidate digital model has obtained its simulation response and target deviation distance under the same simulation conditions as the current filter digital model. Valid candidate digital models are then selected by filtering the remaining deviation of the section to be improved. The current filter digital model and the valid candidate digital models form a design baseline selection set. A new design baseline digital model is determined by the total target deviation. If the new design baseline digital model differs from the current filter digital model in the local connectivity relationships formed in this design cycle, the new design baseline digital model and its simulation response are added to the filter digital model sequence and enter the next design cycle. If the new design baseline digital model is still the current filter digital model, the current filter digital model is retained as the design result of this topology iteration, ensuring that each new design cycle entering the filter digital model sequence has a clear source of structural and response changes.

[0119] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A filter topology iterative design method based on a digital model, characterized in that, include: S1: Obtain the filter digital model sequence and simulation response, determine the target deviation distance at the common frequency sampling position, and compare adjacent filter digital models to form topological differential data and target deviation change data; S2: Merge connection changes in the same design round into local topology change units based on shared connection endpoints, track the connection relationships of nodes involved in the local topology change units to determine the local structure preservation interval, and associate target deviation change data to form a local change response record; S3: Determine the response segment to be improved based on the current filter digital model, determine the candidate local topology change unit from the local change response record, form the continuous improvement coverage segment and the shared node incremental coverage segment, determine the candidate topology modification range based on the overlap state of the two types of coverage segments and the response segment to be improved, and form the candidate digital model within the candidate topology modification range; S4: Simulate candidate digital models, determine new design baseline digital models based on target deviation distance, and update filter digital model sequences based on the new design baseline digital models.

2. The filter topology iterative design method based on digital models according to claim 1, characterized in that, Topological difference terms that share a common connection endpoint and are continuously associated through the shared connection endpoint in the same design round are merged to form local topological change units; Starting from the design round in which the local topology change unit occurs, examine the connection relationships of the nodes involved in each round. The design round in which any node undergoes a connection change again is determined as the new starting point of the local structural change. The continuous design rounds from the design round in which the local topology change unit occurs to the new starting point of the local structural change are determined as the local structural preservation interval.

3. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, The filter digital model immediately adjacent to the candidate local topology change unit before the design round is determined as the first comparison benchmark. The common frequency sampling positions where the target deviation distance of the candidate local topology change unit after the occurrence of the unit is less than the target deviation distance of the first comparison benchmark at the same common frequency sampling position are merged to form a direct improvement segment. Within the local structure preservation interval, the target deviation distance of each common frequency sampling position in the segment is compared with the target deviation distance of the first comparison benchmark at the same common frequency sampling position in each design round. Common frequency sampling positions that are continuously smaller than the target deviation distance of the first comparison benchmark at the same common frequency sampling position are retained to form a continuously improved coverage segment.

4. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, The first shared node local topology change unit involving the node connection relationship of the candidate local topology change unit is determined along the filter digital model sequence. The filter digital model immediately adjacent to it before its occurrence is determined as the second comparison benchmark. The common frequency sampling positions where the target deviation distance after its occurrence is less than the target deviation distance of the second comparison benchmark at the same common frequency sampling position are merged to form an incremental improvement segment. Within its corresponding local structure preservation interval, the common frequency sampling positions that are continuously less than the target deviation distance of the second comparison benchmark at the same common frequency sampling position are retained to form a shared node incremental coverage segment.

5. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, The design baseline digital model located at the current end of the filter digital model sequence is determined as the current filter digital model. Common frequency sampling positions in the current filter digital model with target deviation distance greater than zero and consecutive sequence numbers are merged to form the response segment to be improved. The target deviation change data corresponding to the round of occurrence of local topology change unit in each local change response record is checked. Local topology change units with common frequency sampling positions with target deviation change values ​​less than zero in the response segment to be improved are determined as candidate local topology change units.

6. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, The nodes involved in the candidate topology modification range are matched according to the node identifier of the current filter digital model, and node pairs consisting of two different nodes are determined one by one; node pairs in the connected state are changed to the disconnected state, and node pairs in the disconnected state are changed to the connected state. The formation of each candidate digital model only changes the connection state of one node pair while maintaining the connection states of the rest. Candidate digital models are formed by different single local connection relationship changes.

7. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, Within the response segment to be improved corresponding to each candidate digital model, the target deviation distance is accumulated according to the frequency interval of the common frequency sampling position to form the remaining deviation of the segment to be improved. The candidate digital model whose remaining deviation of the segment to be improved is less than the remaining deviation of the current filter digital model in the same response segment to be improved is determined as a valid candidate digital model. The current filter digital model and the valid candidate digital models constitute the design baseline candidate set. The target deviation distance is accumulated at all common frequency sampling positions to form the total target deviation. The digital model corresponding to the smallest total target deviation is determined as the new design baseline digital model.

8. The filter topology iterative design method based on digital models according to claim 1, characterized in that, Each connection is represented by a node pair consisting of two different nodes and the connection state of the node pair. The connection state includes the existence of connection and the disconnection of connection. Node pairs with the same node identifier in adjacent filter digital models are compared. Node pairs whose connection state changes are identified as topology difference terms. The connection endpoints, connection state changes, and subsequent design rounds of the topology difference terms are then correlated. All topology difference terms formed by the same pair of adjacent filter digital models constitute topology difference data.

9. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, For different single local connectivity changes within the candidate topology modification range, candidate digital models are established, and a correspondence is established between each candidate digital model and the response segment to be improved that generates the candidate topology modification range and the node pair where the connectivity status change occurs. The candidate digital models are simulated using the same simulation response type, port definition, excitation conditions, and common frequency sampling positions as the current filter digital model. The target deviation distance corresponding to each common frequency sampling position is then determined according to the method for determining the target deviation distance.

10. The filter topology iterative design method based on a digital model according to claim 1, characterized in that, Compare the connectivity between the new design baseline digital model and the current filter digital model; When there are differences in the local connectivity relationships formed in this design cycle, the new design baseline digital model and its simulation response are added to the filter digital model sequence according to the new design cycle, and it is determined as the current filter digital model for the next design cycle; when the new design baseline digital model is the current filter digital model, the current filter digital model is retained as the design result of this topology iteration.

Images