Topological electrical parameter mixed fault diagnosis method and device for blasting networking

Abstract

The present application relates to the technical field of data processing and analysis, and further relates to a topological electrical parameter mixed fault diagnosis method and device for blasting networking, comprising the following steps: step S1: abstracting the blasting networking as a directed graph structure and deploying intelligent acquisition nodes containing Kelvin double-arm bridge arrays and switching matrices at each vertex; step S2: applying multiple rounds of excitation at the root node position and synchronously acquiring the node voltage values of each vertex, and constructing a topological voltage distribution matrix according to the vertex number; step S3: calculating the measured resistance value sequence of each edge according to the topological voltage distribution matrix, inputting the edge abnormal scoring model according to the layer number of hops to generate an abnormal edge candidate set; and step S4: performing switching matrix isolation measurement and insulation detection on the abnormal edge candidate set according to the abnormal scoring order. The present application realizes systematic modeling and fine perception of the electrical state of complex blasting networking, effectively improving the accuracy and localizability of pre-explosion fault diagnosis.

Description

Technical Field

[0001] This invention belongs to the field of data processing and analysis technology, specifically relating to a method and device for diagnosing hybrid topology electrical parameters in blasting networks. Background Technology

[0002] Blasting operations typically rely on a network structure consisting of numerous detonators and connecting wires to achieve synchronous or sequential detonation. The safety and reliability of this operation largely depend on the detection and confirmation of the electrical condition of the entire blasting network before detonation. In current engineering practice, blasting networks are susceptible to various factors during construction, transportation, laying, and backfilling, including mechanical pulling, loose joints, insulation aging, moisture corrosion, and human error, leading to faults such as open circuits, short circuits, poor contact, or insulation damage. Failure to accurately identify and locate these hidden faults before detonation can result not only in misfires, misfires, or unbalanced blasting effects, but also potentially serious safety accidents. Therefore, comprehensive, reliable, and locatable electrical fault diagnosis of blasting networks without applying detonation energy has always been a key concern in the engineering field.

[0003] In existing technologies, a common approach is to use overall resistance measurement or simple segmented measurement to detect the total resistance of the blasting circuit, or to measure the circuit resistance at local nodes to see if it falls within the empirically permissible range. These methods are simple to implement and have low equipment costs, but they are essentially overall or coarse-grained detection methods, often only able to determine "whether an anomaly exists," but struggling to pinpoint the specific segment of the conductor or node where the anomaly occurred. When the blasting network is large-scale and the branch structure is complex, changes in total resistance may be masked by normal fluctuations in multiple conductors, leading to minor contact defects or early insulation degradation going undetected. Furthermore, conductors of different lengths and materials have significant resistance differences, and relying solely on fixed thresholds can easily result in misjudgments or missed detections. To improve location capabilities, some existing solutions introduce segmented testing or segment-by-segment troubleshooting, such as manually disconnecting conductors segment by segment and measuring their resistance, or manually dividing the wiring into testing sections. While these methods improve location accuracy to some extent, they rely on manual operation, are inefficient, and in complex field environments and large-scale networks, the testing process is time-consuming and prone to introducing new wiring errors. At the same time, frequent manual disconnection may also disturb the original contact state, reducing the stability and repeatability of the test results. Summary of the Invention

[0004] The main objective of this invention is to provide a method and apparatus for topological electrical parameter hybrid fault diagnosis of blasting networks. By abstracting the blasting network into a directed graph structure and introducing intelligent acquisition nodes with Kelvin double-arm bridge arrays and switching matrices at each vertex, a systematic modeling and refined perception of the electrical state of complex blasting networks is achieved, effectively improving the accuracy and localization of fault diagnosis before detonation.

[0005] To solve the above problems, the technical solution of the present invention is implemented as follows: A method for diagnosing hybrid electrical parameter faults in blast-type networks includes the following steps: Step S1: Abstract the blasting network into a directed graph structure and deploy intelligent acquisition nodes containing Kelvin double-arm bridge arrays and switching matrices at each vertex. Scan each edge to obtain the initial resistance measurement values ​​of the segmented conductors and generate a topological resistance reference map. Step S2: Apply multiple rounds of excitation at the root node and simultaneously collect the node voltage values ​​of each vertex, and construct the topology voltage distribution matrix according to the vertex number; Step S3: Calculate the sequence of measured resistance values ​​for each side based on the topological voltage distribution matrix, differentiate it with the topological resistance reference map and extract the fluctuation feature vector, and generate a candidate set of abnormal sides by inputting it into the side anomaly scoring model according to the number of jumps. Step S4: Perform switching matrix isolation measurement and insulation detection on the candidate set of abnormal edges according to the abnormality score, and determine the fault type by combining the isolation state resistance value and leakage current value and output the fault diagnosis result.

[0006] Furthermore, in step S1, each detonator node in the directed graph structure is a vertex, each connecting wire segment is an edge, and the root node corresponds to the blasting network bus entry point; vertex numbers are set for vertices and segment wire numbers are set for edges, and a topology connection table is established based on vertex numbers and segment wire numbers.

[0007] Furthermore, the Kelvin dual-arm bridge array includes a first bridge arm and a second bridge arm. The first bridge arm is connected to a precision reference resistor group, and the second bridge arm is connected to the segmented conductor to be measured. The switching matrix includes a row gating module and a column gating module. The row gating module selects the current segmented conductor number to be measured, and the column gating module selects the bridge unit number of the Kelvin dual-arm bridge array participating in the measurement and establishes the connection configuration of the first bridge arm and the second bridge arm.

[0008] Further, in step S1, the control switching matrix scans the segmented wires corresponding to each edge of the directed graph structure and obtains the initial resistance measurement value of each segmented wire, storing the initial resistance measurement value of each segmented wire to form a topological resistance reference map; obtaining the initial resistance measurement value of the segmented wire includes: sending an initialization command to each intelligent acquisition node; after the intelligent acquisition node responds to the initialization command, the row selection module of the switching matrix sequentially selects each segmented wire according to the preset scanning order, and the column selection module of the switching matrix connects the selected segmented wire to the second bridge arm of the Kelvin double-arm bridge array; the Kelvin double-arm bridge array injects a constant measurement current into the second bridge arm through the current injection port of the first bridge arm, and the Kelvin double-arm bridge array collects the voltage value across the precision reference resistor group through the voltage sensing port of the first bridge arm and collects the voltage value across the segmented wire through the voltage sensing port of the second bridge arm, and calculates the initial resistance measurement value of the segmented wire based on the voltage value across the precision reference resistor group, the voltage value across the segmented wire, and the nominal resistance value of the precision reference resistor group; the initial resistance measurement value of each segmented wire is stored in the topological resistance database to form a topological resistance reference map.

[0009] Furthermore, the construction of the topology voltage distribution matrix in step S2 includes: sending a voltage divider excitation command to the intelligent acquisition node at the root node position; after the intelligent acquisition node at the root node position responds to the voltage divider excitation command, it switches the matrix row selection module to connect the excitation voltage source to the bus input and outputs the first round of DC excitation voltage; under the state of the first round of DC excitation voltage application, a synchronous acquisition command is broadcast to all intelligent acquisition nodes; after all intelligent acquisition nodes respond to the synchronous acquisition command, they acquire the node voltage value through the voltage sensing port of the second bridge arm of the Kelvin double-arm bridge array and transmit the node voltage value back through the bus; the node voltage value is the voltage value of the location of the intelligent acquisition node relative to the common reference ground; the node voltage values ​​are arranged in order of vertex number to form the first round of node voltage vector; the excitation voltage source is controlled to output the second round of DC excitation voltage to multiple rounds of DC excitation voltage in sequence and the process of broadcasting the synchronous acquisition command, transmitting the node voltage value back and constructing the node voltage vector is repeated; the multiple rounds of node voltage vectors are stacked row by row to form a topology voltage distribution matrix and stored in the voltage divider response database.

[0010] Furthermore, step S3 involves solving the sequence of estimated measured resistance values, which includes: obtaining the difference between node voltage values ​​between adjacent vertices based on the topological voltage distribution matrix and obtaining the current estimate of the current flowing through the edge; generating the estimated measured resistance values ​​of the edge under each round of DC excitation voltage conditions based on the difference between node voltage values ​​and the current estimate, and forming the sequence of estimated measured resistance values; the current estimate is obtained based on the output state of the excitation voltage source and the topological resistance reference spectrum.

[0011] Furthermore, step S3 involves extracting the fluctuation feature vector as follows: extracting the initial resistance measurement value corresponding to the edge from the topological resistance reference map as the reference value; obtaining the resistance deviation value by subtracting the reference value from each measured resistance estimate in the measured resistance estimate sequence; and forming a resistance deviation sequence by calculating the sequence mean, sequence range, and sequence variance of the resistance deviation sequence and forming a fluctuation feature vector from the sequence mean, sequence range, and sequence variance.

[0012] Furthermore, step S3, which involves generating a candidate set of abnormal edges by inputting the edge anomaly scoring model in layers according to the number of hops, includes: dividing the edges into multiple layers based on the number of hops from the root node according to the directed graph structure, and processing each layer sequentially starting from the root node layer; reading the fluctuation feature vector for each edge and inputting it into the edge anomaly scoring model to obtain an anomaly score value; marking the edges as high-confidence abnormal edges, medium-confidence abnormal edges, or normal edges based on the relationship between the anomaly score value and the first preset threshold and the second preset threshold; and summarizing the high-confidence abnormal edges and medium-confidence abnormal edges to form a candidate set of abnormal edges.

[0013] Further, step S4, determining the fault type and outputting the fault diagnosis result, includes: sorting the abnormal edge candidate set from high to low according to the abnormal edge score value to form a queue of abnormal edges to be verified, and selecting the current edge to be verified in sequence; determining that the current edge to be verified connects the first endpoint and the second endpoint; sending an isolation measurement command to the intelligent acquisition node at the first endpoint; after the intelligent acquisition node at the first endpoint responds to the isolation measurement command, switching the matrix row selection module to select the segmented conductor corresponding to the current edge to be verified, and switching the matrix column selection module to disconnect the segmented conductor from the overall circuit of the blasting network; and performing a four-wire precision resistance measurement on the segmented conductor in the disconnected state using the Kelvin double-arm bridge array to obtain the isolation state resistance value; sending an insulation detection command to the intelligent acquisition node at the second endpoint; after the intelligent acquisition node at the second endpoint responds to the insulation detection command, switching the matrix row selection module to select the segmented conductor corresponding to the current edge to be verified, and switching... The switching matrix column gating module connects the insulation test source to the insulation gap between the segmented conductor and the adjacent conductor. The micro-current detection unit collects the leakage current value flowing through the insulation gap. Based on the comparison results of the isolation state resistance value and the initial resistance measurement value, as well as the comparison results of the leakage current value and the insulation qualification threshold, the fault type is determined. The fault types include open circuit fault, poor contact fault, insulation damage fault, and short circuit fault. A restore connection command is sent to the intelligent acquisition node at the first endpoint position to cause the switching matrix column gating module to reconnect the segmented conductor to the overall loop of the blasting network. The fault type of the current edge to be verified is recorded to the fault diagnosis result list and the current edge to be verified is removed from the queue of abnormal edges to be verified. The process of selecting the queue of abnormal edges to be verified, sending the isolation measurement command, sending the insulation detection command, determining the fault type, and sending the restore connection command is repeated until the queue of abnormal edges to be verified is empty. The fault diagnosis result list is then output as the fault diagnosis result.

[0014] A hybrid fault diagnosis device for topology-electrical parameters in blasting network configurations includes a main control unit, multiple intelligent acquisition nodes, a bus, a topology resistance database, and a voltage divider response database. Multiple intelligent acquisition nodes are deployed at each vertex of the directed graph structure of the blasting network. Each intelligent acquisition node includes a Kelvin double-arm bridge array and a switching matrix. The switching matrix includes row selection and column selection modules. The Kelvin double-arm bridge array includes a first bridge arm and a second bridge arm. The first bridge arm connects to a precision reference resistor group, and the second bridge arm connects to the conductor segment to be tested. The intelligent acquisition node at the root node includes an excitation voltage source. Each intelligent acquisition node also includes an insulation test source and a micro-current detection unit. The main control unit is connected to the bus and configured... The process involves: constructing a directed graph structure and controlling the row and column gating modules to scan segmented conductors to obtain initial resistance measurements, which are then written into the topology resistance database to form a topology resistance reference map; controlling the excitation voltage source to output multiple rounds of DC excitation voltage and controlling the intelligent acquisition node to synchronously acquire node voltage values ​​to form a topology voltage distribution matrix, which is then written into the voltage divider response database; generating an abnormal edge candidate set based on the topology resistance reference map and the topology voltage distribution matrix; controlling the switching matrix to disconnect segmented conductors and controlling the Kelvin double-arm bridge array to obtain the isolation state resistance value; and controlling the insulation test source and the micro-current detection unit to obtain the leakage current value; and outputting fault diagnosis results based on the isolation state resistance value and the leakage current value.

[0015] The method and apparatus for topology electrical parameter hybrid fault diagnosis of blasting network of the present invention have the following beneficial effects: Compared with the traditional method that relies solely on overall resistance or manual segment detection, the present invention establishes a unified topology constraint at the network level, so that the initial resistance measurement value of segmented conductors can be stored and reused in the form of a topology resistance reference spectrum for a long time, thereby effectively distinguishing the structural differences and abnormal changes caused by conductors of different lengths and materials.

[0016] During the multi-round excitation and synchronous acquisition phase, a topological voltage distribution matrix is ​​constructed to transform the potential state of all network nodes into a calculable and alignable data structure. This allows for a complete characterization of the response characteristics of each conductor segment under multi-round excitation conditions, providing sufficient information redundancy for subsequent analysis. Based on this, the invention introduces the concepts of measured resistance estimation sequence and resistance deviation sequence, combining static benchmarks with dynamic responses. Multi-round information is compressed using fluctuation feature vectors, amplifying early, hidden faults such as poor contact at the statistical feature level, preventing them from being masked by normal fluctuations in the overall circuit. Anomaly scoring is performed hierarchically based on the number of jumps, ensuring consistency between the diagnostic process and the spatial structure of the network. This facilitates the priority identification of critical conductor segments with a greater impact on the overall circuit, thereby improving diagnostic efficiency and reducing interference from irrelevant segments.

[0017] During the verification phase, this invention utilizes a switching matrix to achieve controllable isolation of individual segmented conductors. Without disrupting the overall network structure, it obtains the isolation resistance value and combines it with leakage current values ​​obtained from insulation testing for joint judgment. This allows open-circuit faults, poor contact faults, insulation damage faults, and short-circuit faults to be clearly distinguished within the same judgment framework. By promptly restoring the connection after each verification, subsequent testing is ensured to be conducted under conditions close to the original operating conditions, enhancing the consistency and repeatability of diagnostic results. Overall, this invention has significant advantages in safety, automation, diagnostic accuracy, and engineering applicability. It enables rapid, reliable, and traceable fault diagnosis of large-scale, multi-branch blasting networks without applying initiation energy, providing strong technical support for the safe implementation of blasting operations. Attached Figure Description

[0018] Figure 1 A circuit diagram illustrating the connection principle between the Kelvin double-arm bridge array and the switching matrix provided in an embodiment of the present invention; Figure 2 A schematic diagram of the resistance evolution characteristics of open circuit faults and poor contact faults provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the characteristic curves of insulation damage fault and short circuit fault provided in an embodiment of the present invention. Detailed Implementation

[0019] A method for diagnosing hybrid electrical parameter faults in blast-type networks includes the following steps: Step S1: Abstract the blasting network into a directed graph structure and deploy intelligent acquisition nodes containing a Kelvin double-arm bridge array and a switching matrix at each vertex. Scan each edge to obtain the initial resistance measurement values ​​of the segmented conductors and generate a topological resistance reference map. Step S2: Apply multiple rounds of excitation at the root node and simultaneously collect the node voltage values ​​of each vertex, and construct the topology voltage distribution matrix according to the vertex number; Step S3: Calculate the sequence of measured resistance values ​​for each side based on the topological voltage distribution matrix, differentiate it with the topological resistance reference map and extract the fluctuation feature vector, and generate a candidate set of abnormal sides by inputting it into the side anomaly scoring model according to the number of jumps. Step S4: Perform switching matrix isolation measurement and insulation detection on the candidate set of abnormal edges according to the abnormality score, and determine the fault type by combining the isolation state resistance value and leakage current value and output the fault diagnosis result.

[0020] In one implementation, the main control unit first receives the wiring list and on-site identification information of the blasting network, mapping the detonator nodes as vertices of a directed graph structure and the segmented conductors as edges. The reason for using directed edges in the directed graph structure is that subsequent electrical parameter acquisition requires a unified agreement on the current injection direction and voltage induction direction. Using directed edges fixes the measurement polarity from the starting vertex to the ending vertex, ensuring consistent voltage signs for the same segmented conductor across different rounds and intelligent acquisition nodes, avoiding inconsistencies in the initial resistance measurement values ​​of the same segmented conductor in the topology resistance database due to polarity reversal. Vertex numbers use consecutive integer encoding; for example, the root node is numbered 1, and the remaining vertices are numbered sequentially according to the on-site construction or wiring order, from 2, 3, 4 to the end vertex. Segmented conductor numbers also use consecutive integer encoding; for example, edges are numbered 1, 2, 3, up to the end segmented conductor number. The main control unit establishes a topology connection table based on the vertex number and segmented traverse number. The topology connection table contains at least the fields of "segmented traverse number, starting vertex number, ending vertex number, and set of adjacent segmented traverse numbers" to facilitate the rapid location of the correspondence between segmented traverses and physical channels during subsequent scanning.

[0021] Subsequently, intelligent acquisition nodes are deployed at each vertex. Each intelligent acquisition node comprises a Kelvin dual-arm bridge array and a switching matrix. The first arm of the Kelvin dual-arm bridge array connects to a precision reference resistor group, and the second arm connects to the segmented conductor under test. The reason for using a Kelvin dual-arm bridge array here is that the resistance of the segmented conductor is often on the order of magnitude. When the additional resistance introduced by the connecting terminals, posts, and contact points is on the same order of magnitude as the conductor resistance, two-wire measurement will mix the contact point resistance with the conductor resistance. The Kelvin dual-arm bridge array separates the current injection port and the voltage sensing port, so that the voltage sensing port carries almost no current. This significantly reduces the impact of the voltage drop on the voltage sensing port leads and contact points on the measurement results, resulting in better repeatability of the initial resistance measurement values ​​of the segmented conductor and a more stable topological resistance reference spectrum. The switching matrix includes a row selection module and a column selection module. The row selection module selects the current segmented conductor number under test, and the column selection module selects the bridge unit number of the Kelvin dual-arm bridge array participating in the measurement and establishes the connection configuration of the first and second bridge arms. The scanning order of the row selection module can be consistent with the segmented conductor number, such as scanning from 1 to 256, or it can generate a scanning order of "expanding outward from the root node" according to the topology connection table. The latter is more intuitive when troubleshooting wiring errors on site because the scanning progress is consistent with the wiring direction.

[0022] refer to Figure 1To clearly illustrate the circuit topology, the conventional power supply module and communication interface are omitted from the diagram, focusing instead on the precision measurement circuitry of the analog front end. The core logic of this circuit architecture lies in eliminating the influence of contact resistance and lead resistance on measurement accuracy through a four-wire measurement method, particularly for segmented conductors with low resistance in long-distance transmission within explosive network deployments. For example... Figure 1 As shown, the measurement circuit mainly consists of two parts: a constant measurement current injection circuit and a high-impedance voltage sensing circuit. Their physical connection is dynamically reconfigured through a switching matrix. In the constant measurement current injection circuit, a constant current source controlled by the main control unit generates a constant measurement current. The current flows out from the current injection port and sequentially through the first and second bridge arms. The first bridge arm is connected in series with a precision reference resistor group, as shown in the figure. This indicates the equivalent resistance value of the precision reference resistor group. The precision reference resistor... A low-temperature drift, high-precision alloy resistor is selected, whose resistance value is known and serves as the traceability reference for the entire measurement system. The second bridge arm dynamically connects to the segmented conductor to be measured via the column and row selection modules of the switching matrix, as shown in the figure (labeled). This represents the equivalent resistance of the segmented conductor. The switching matrix acts as a multiplexer in the circuit, closing specific switching channels according to instructions in the topology connection table, thereby physically connecting specific edges distributed in the directed graph structure of the blasting network to the position of the second bridge arm. When the switch is closed, a constant measuring current is maintained. To form a complete series circuit, i.e., current Flowing sequentially without splitting and Then it flows back to signal ground or current return port. Because and Since they are in the same series branch, according to Kirchhoff's current law, the current flowing through them is exactly equal, both being [missing value]. .

[0023] In the voltage sensing loop, the system is configured with two independent high-impedance voltage acquisition channels. The first channel is connected across a precision reference resistor. The two ends are used to collect reference voltage values. Please note that in the image... The sampling point is located inside the resistor terminals, close to the resistor body. This connection method ensures that the sampled voltage drop is only caused by the resistor itself. It is generated by the conductor itself, excluding the voltage drop across the connecting wires. The second channel is connected across the segmented conductor. Both ends are used to collect the voltage value of the conductor. The dashed lines in the diagram represent voltage sensing lines, which are directly connected to the two ends of the segmented conductor. Because the voltage acquisition channel has extremely high input impedance (typically in the megaohm range), the current flowing through the voltage sensing lines approaches zero. According to Ohm's law, the voltage drop across the sensing lines is almost zero; therefore, the potential difference acquired at the voltage sensing port accurately reflects... The actual voltage across the two ends. This four-wire connection effectively separates the injection circuit carrying large current from the measurement circuit carrying weak signals in space, ensuring that the contact resistance introduced by the switching matrix contacts and the lead resistance introduced by the long wires exist only in the current loop and are not superimposed on the voltage measurement result. Based on the above circuit connection, the main control unit executes the resistance calculation logic. According to Ohm's law, the voltage across the precision reference resistor satisfies the equation... The voltage across the segments of the conductor to be measured satisfies the equation In practical engineering scenarios, constant measurement current may be affected by power fluctuations or changes in ambient temperature. Slight drift may occur if used directly. Calculations are performed, and the current is... Errors will be directly transmitted to the final result. However, this embodiment uses the ratio measurement method, and by solving the two equations simultaneously, we can obtain... By modifying this formula, the formula for calculating the resistance to be measured is: During this calculation process, a constant measuring current is used. As a common factor, it is mathematically eliminated. This means that even if the output of the current source fluctuates at the instant of measurement, as long as that fluctuation simultaneously acts on... and Its effect on the final resistance measurement value The effects will cancel each other out. This characteristic is particularly important in field blasting environments, as battery voltage drops or drastic temperature changes often lead to instability of the excitation source. The circuit topology in this embodiment ensures the robustness of the measurement data through a hardware-level differential comparison mechanism. Furthermore, the figure also illustrates the gating action of the switching matrix. When it is necessary to measure the next segmented conductor, the column gating module will disconnect the current... The connection is established and the next wire is quickly connected, maintaining a constant state throughout the process. The connection state in the first bridge arm remains unchanged, ensuring that all segmented conductor measurements are based on the same physical reference. This achieves consistent normalization of data across the entire network in the topological resistance reference map.

[0024] To ensure that each intelligent acquisition node outputs initial resistance measurement values ​​under the same measurement reference, the main control unit sends an initialization command to each intelligent acquisition node. The initialization command carries at least the nominal resistance value of the precision reference resistor group, the segmented conductor number and physical channel mapping, and the scanning strategy field. The nominal resistance value of the precision reference resistor group can be selected from multiple ranges, such as four ranges of 100, 200, 500, and 1000 connected in parallel, or a range switch can be used to form an equivalent resistance value set. This ensures that the voltage value of the first bridge arm falls within the high-resolution acquisition range when facing segmented conductors of different lengths and materials, thereby improving measurement resolution. For example, the main control unit pre-assigns one precision reference resistor group range for each segmented conductor. The assignment rules can be based on historical engineering statistics, such as dividing common length segmented conductors into 8 categories, each corresponding to a more suitable range; alternatively, a quick pre-scan can be performed during the initialization phase to select a more suitable range using coarse measurement results.

[0025] After responding to the initialization command, the intelligent acquisition node enters the scanning process for the initial resistance measurement of the segmented conductors. The row selection module of the switching matrix sequentially selects each segmented conductor according to a preset scanning order, while the column selection module connects the selected segmented conductor to the second arm of the Kelvin dual-arm bridge array, simultaneously connecting the corresponding precision reference resistor group to the first arm. Subsequently, the Kelvin dual-arm bridge array injects a constant measurement current into the second arm through the current injection port of the first arm, and acquires the voltage value across the precision reference resistor group through the voltage sensing port of the first arm, and the voltage value across the segmented conductor through the voltage sensing port of the second arm. The constant measurement current is output in an energy-constrained manner, for example, each injection lasts for 20 to 50 milliseconds, with an injection interval set to 50 to 200 milliseconds. The purpose is to make the slight temperature rise of the conductor and contact point negligible within the measurement cycle, making the initial resistance measurement value closer to the static value and reducing the deviation caused by thermal drift. To suppress the impact of on-site electromagnetic interference on voltage acquisition, voltage sampling adopts a strategy of multiple sampling and taking the median or truncated mean. For example, 32 sampling points are collected for each configuration, the four largest and four smallest are discarded, and the mean of the remaining sampling points is calculated. In this way, a stable voltage value can still be obtained when there are occasional spike interferences.

[0026] In one specific calculation method, the intelligent acquisition node or main control unit calculates the initial resistance measurement value of the segmented conductor based on the voltage values ​​across the precision reference resistor set, the voltage values ​​across the segmented conductor, and the nominal resistance value of the precision reference resistor set. The calculation can be performed using a bridge ratio in the following form: in, This represents the initial resistance measurement value of the segmented conductor. This indicates the nominal resistance value of the precision reference resistor group. This indicates the voltage value across the segmented conductor. This represents the voltage across the precision reference resistor array. The advantage of using a ratio form is that a constant measurement current may exhibit slight deviations under different intelligent acquisition nodes, different temperatures, and power supply fluctuations. and The current deviations experienced during the same injection have the same direction. The effect of these current deviations can be canceled out by using a ratio, thus ensuring consistency between the currents obtained from different intelligent acquisition nodes. More consistent. To avoid problems caused by shorter segmented conductor lengths. To minimize sampling noise and amplify relative errors, a "voltage sensing port bridging length verification" can be added to the switching matrix access configuration. This involves configuring an additional pair of voltage sensing points closer to the terminals outside the two ends of the same segmented conductor for retesting. If the deviation between the two results exceeds a preset consistency condition, resampling is triggered, and the median of the three results is taken as the mean. .

[0027] To ensure the topological resistance reference spectrum can serve as a long-term benchmark, measurement-related status data is recorded synchronously during the scanning process. A common practice is to record the following fields for each segment: "measurement round, precision reference resistance group setting, number of sampling points, truncation strategy identifier, smart acquisition node number, and ambient temperature sampling value." The ambient temperature sampling value can be provided by a temperature sensor within the smart acquisition node, for example, a 0.1 resolution sample, and is collected once before each segment measurement begins. The benefit of this approach is that if subsequent measurements of the same segment's resistance at different times show drift, the precision reference resistance group setting and ambient temperature sampling value at that time can be traced back to distinguish between "normal changes caused by material temperature drift" and "abnormal changes caused by poor contact," thereby reducing false alarms.

[0028] After scanning is complete, the main control unit summarizes the initial resistance measurements of all segmented conductors and writes them into the topology resistance database, forming a topology resistance baseline map. The topology resistance baseline map must contain at least the fields "segmented conductor number, starting vertex number, ending vertex number, initial resistance measurement value, measurement timestamp, intelligent acquisition node number, and precision reference resistance group setting". To ensure the topology resistance baseline map is consistent with the directed graph structure, the main control unit performs a consistency check before data entry: it checks whether each segmented conductor number exists in the topology connection table, whether the starting vertex number matches the ending vertex number, and whether there are duplicate records for the same segmented conductor number. When duplicate records are found, they are not directly overwritten; instead, multiple records are retained and marked with the cycle number, facilitating the selection of a more reliable record later, for example, prioritizing the record with a higher number of sampling points and a smaller variance after truncation.

[0029] In terms of alternative implementation methods, obtaining the initial resistance measurement value of the segmented conductor is not limited to a single scan. In one implementation, the main control unit controls the intelligent acquisition node to perform three independent scans on each segmented conductor, and then processes the results from the three scans. The median value is written into the topological resistance database to suppress outliers caused by occasional contact jitter. In another implementation, the column gating module of the switching matrix selects not only the bridge unit number but also different combinations of voltage sensing ports, forming two configurations: "terminal inner sensing" and "terminal outer sensing." If there is a stable difference in the initial resistance measurements obtained from the two configurations, the main control unit records the difference as an estimate of the additional resistance at the contact point, providing a more sensitive benchmark for subsequent contact failure assessment. In yet another implementation, the precision reference resistor set uses two sets of independent nominal resistance values ​​for parallel measurement. The same segment of the conductor yields two values ​​under the two sets of nominal resistance values. The main control unit uses the difference between the two as a linear consistency index. The measurement result with better linear consistency is written into the topological resistance reference spectrum, while the measurement result with poorer linear consistency triggers a retest and records the reason for the retest.

[0030] The above implementation process binds the "topological constraints of the directed graph structure" and the "high-precision segmented conductor resistance measurement of the Kelvin double-arm bridge array" into the same data link: on the one hand, the topology connection table fixes the physical conductors and edge numbers to avoid benchmark confusion; on the other hand, the ratio calculation and multiple sampling strategy make the repeatability and cross-node consistency of the initial resistance measurement values ​​stronger, so that the topology resistance benchmark spectrum can stably support the subsequent calculation of the measured resistance estimation sequence and differential diagnosis based on the topology voltage distribution matrix.

[0031] In one implementation, after generating the topological resistance reference map, the main control unit enters a multi-round excitation and synchronous acquisition phase. To ensure that the node voltage values ​​of different vertices are comparable under the same time reference, the main control unit first defines one acquisition process as an acquisition round and assigns a round number to each acquisition round. The round number uses a consecutive integer encoding, for example, starting from 1 and incrementing sequentially to 6. The setting of the round number brings two direct benefits: first, it facilitates the concatenation of node voltage values ​​from the same round into a node voltage vector according to the vertex number; second, it facilitates the stacking of node voltage vectors from multiple rounds into a topological voltage distribution matrix, enabling the subsequent calculation of the measured resistance estimation value sequence to trace the output state of the excitation voltage source using the round as an index.

[0032] When multiple rounds of excitation are applied at the root node, the main control unit sends a voltage divider excitation command to the intelligent acquisition node at the root node. After responding to the voltage divider excitation command, the intelligent acquisition node at the root node switches the matrix row selection module to connect the excitation voltage source to the bus input and outputs the first round of DC excitation voltage. The reason for using the root node as the access point is that the topology connection table is organized into hop count levels starting from the root node. Applying a DC excitation voltage at the root node allows the potential to propagate outward along the directed graph structure, making the potential decrease relationship between the vertex number and the hop count more stable. This facilitates the subsequent layered processing by hop count, making it easier for anomaly scoring to exhibit spatial clustering characteristics. The DC excitation voltage maintains a stable window in each round, for example, 120 milliseconds, so that the sampling after the synchronous acquisition command is triggered can cover the stable segment. At the same time, an interval window is inserted between rounds, for example, an interval of 80 milliseconds, to wait for the node voltage value to fall back to the initial state that can be repeated in the next round, thereby reducing the impact of residual charge or electromagnetic disturbance on the next round. When the excitation voltage source outputs multiple rounds of DC excitation voltage, the output state of each round is recorded as the excitation voltage source output state field and stored in association with the round number, which is convenient to call in the subsequent calculation of the current estimate.

[0033] In the first round of DC excitation voltage application, the main control unit broadcasts a synchronization acquisition command to all intelligent acquisition nodes. The synchronization acquisition command includes a round number, an acquisition start time marker, and a sampling configuration field. The sampling configuration field can include the number of sampling points and the sampling interval. For example, if the number of sampling points is 64 and the sampling interval is 1 millisecond, each intelligent acquisition node obtains a sequence of node voltage values ​​within a 64-millisecond sampling window. The reason for sampling within the same sampling window is that there is a slight difference between the fieldbus propagation delay and the local clock of each intelligent acquisition node. If only single-point voltage values ​​are collected, occasional interference will significantly affect the single-point results. Representing the node voltage values ​​as statistics within a window can suppress transient disturbances to a lower level. A common practice is to perform a truncated mean on the 64 sampling points for each intelligent acquisition node, discarding the four largest and four smallest values, and then averaging the remaining 56 sampling points to obtain the node voltage values ​​for that round. The node voltage value is defined as the voltage value of the location of the intelligent acquisition node relative to the common reference ground. The common reference ground is maintained in a consistent manner by the intelligent acquisition node at the root node position, and a common reference ground verification flag is broadcast through the bus before the synchronous acquisition command is issued, so that each intelligent acquisition node outputs the node voltage value under the same common reference ground framework.

[0034] After all intelligent acquisition nodes have completed sampling, they transmit the round number, vertex number, and node voltage value back to the main control unit via the bus. The main control unit arranges the node voltage values ​​according to the vertex number order to construct the first round node voltage vector. If the blast network contains 128 vertices, the first round node voltage vector contains 128 elements. The main control unit then controls the excitation voltage source to sequentially output the second round DC excitation voltage to the multiple rounds of DC excitation voltage, and repeats the process of broadcasting synchronous acquisition commands, transmitting node voltage values, and constructing node voltage vectors to obtain multiple rounds of node voltage vectors. Finally, the main control unit stacks the multiple rounds of node voltage vectors row-wise to form a topology voltage distribution matrix and stores it in the voltage divider response database. To facilitate subsequent fast indexing by round and vertex, the topology voltage distribution matrix can be represented by row and column indexes. ,in Indicates the round number is The vertex number in the node voltage vector is The node voltage value; For round numbering, Number the vertices; The values ​​are derived from the corresponding intelligent data collection nodes in each round. The node voltage values ​​are output and returned. The advantage of using matrix representation is that the voltage difference between adjacent vertices of any subsequent edge can be directly obtained through two matrix elements in the same round, making the calculation path clear and suitable for batch processing.

[0035] Regarding optional implementation methods, the number of rounds of DC excitation voltage can be adaptively selected according to the network scale. For example, 4 rounds are used when the number of vertices is less than 64, 6 rounds are used when the number of vertices is between 64 and 256, and 8 rounds are used when the number of vertices is greater than 256. Increasing the number of rounds can enhance statistical stability and observability, because the node voltage value of the same edge in different rounds will be affected by the output state of the bus entrance excitation voltage source and the overall loop current distribution relationship. Multiple rounds of data make it easier for the subsequent resistance deviation sequence to show the difference between "stable offset" and "fluctuation with round", thereby improving the distinction between poor contact faults and open circuit faults. Another approach is to execute two synchronous acquisition commands in the stable section after each round of DC excitation voltage application, obtain two node voltage values, and take the median as the node voltage value of that round. In this way, the intra-row consistency of the topology voltage distribution matrix is ​​better when there is intermittent bus interference.

[0036] During the stage of calculating the estimated resistance sequence for each edge, the main control unit reads the topological voltage distribution matrix from the voltage divider response database, the topological resistance reference spectrum from the topological resistance database, and the starting and ending vertex numbers corresponding to each edge from the topological connection table. Since the directed graph structure defines the direction of the voltage difference for each edge, the main control unit calculates the difference in node voltage values ​​between adjacent vertices for each edge in each round. For ease of writing, it can be... Indicates that the vertex number is In round number The node voltage values ​​below, where From the elements of the topological voltage distribution matrix Obtain; Use The edge number is The starting vertex number, using The edge number is The endpoint vertex number is given. Then the edge number is given. In round number The difference in node voltage values ​​can be expressed as ,in This represents the voltage difference along the edge. This represents the nodal voltage value at the starting vertex. This represents the node voltage value at the endpoint vertex. Number the edges. The rounds are numbered. By adopting this consistent voltage difference definition, the resistance calculation for the same edge in all rounds is expressed using the same notation system, resulting in more stable statistical characteristics of the subsequent resistance deviation sequence.

[0037] After obtaining the difference in node voltage values, the main control unit generates an estimate of the current flowing through the edge. This current estimate is calculated based on the output state of the excitation voltage source and the topology resistance reference map. The core consideration is that in multi-branch networks, the edge current cannot be directly obtained solely from the difference in node voltage values. The topology resistance reference map provides the initial resistance measurement value for each segmented conductor, which can serve as an approximate resistance prior for estimating current distribution. Specifically, the main control unit uses the initial resistance measurement value from the topology resistance reference map as the initial resistance value for each edge, and combines this with the output state of the excitation voltage source to solve for the DC network current splitting, obtaining the current estimate for each edge in each round. The current estimate can be denoted as... ,in The edge number is In round number The estimated current value is as follows; The sign of the edge is consistent with the edge direction, making... and The correspondence remains consistent. (This is the result.) Subsequently, the main control unit generates estimated values ​​of the measured resistance of the edge under each round of DC excitation voltage conditions based on the difference in node voltage values ​​and the estimated current values, and forms a sequence of estimated measured resistance values. The estimated measured resistance values ​​can be expressed as follows: ,in The edge number is In round number The calculated value of the measured resistance is as follows. This represents the difference in node voltage values ​​for that round. This represents the estimated current value for that round. Using the estimated current value as the denominator allows for a unified conversion of voltage changes across multiple excitation rounds to a resistance space, enabling direct comparison of results from different rounds. If a contact failure exists on a certain edge, the difference in node voltage values ​​under multiple excitation rounds often exhibits nonlinear fluctuations; this needs to be converted to... It is then more easily captured by the sequence range and sequence variance.

[0038] After obtaining the sequence of measured resistance estimates for each edge, the main control unit extracts the corresponding initial resistance measurement value from the topological resistance reference map as the reference value. The difference between each measured resistance estimate in the measured resistance estimate sequence and the reference value yields the resistance deviation value, which is then used to form the resistance deviation sequence. The significance of the resistance deviation sequence lies in separating the "nominal difference of the conductor itself" from the "increment caused by anomalies." Different segments of the conductor in the same network have different lengths and materials, resulting in naturally different initial resistance measurements. This difference can be directly addressed by using... Anomaly scoring often treats structural differences as anomalies, while the fluctuation feature vector of the resistance deviation sequence, after aligning with the benchmark value, better reflects the dynamic attributes of anomalies. The main control unit calculates the sequence mean, sequence range, and sequence variance of the resistance deviation sequence and combines them into a fluctuation feature vector. The sequence mean characterizes long-term shifts, suitable for identifying continuous increases caused by open-circuit faults; the sequence range characterizes the span between cycles, suitable for capturing jumps caused by poor contact faults; the sequence variance characterizes the fluctuation intensity, suitable for distinguishing between "stable shifts" and "random jitters." For example, under a 6-cycle DC excitation voltage, if the resistance deviation sequence of a certain edge shows "closeness in the first 3 cycles and a significant upward shift in the last 3 cycles," the sequence mean and sequence range will increase simultaneously, and the fluctuation feature vector is more likely to be judged as an anomaly in the edge anomaly scoring model. If the resistance deviation sequence of a certain edge has basically the same number of cycles but is higher than the benchmark overall, the sequence mean increases while the sequence range and sequence variance are small, which is closer to the situation of open-circuit faults or stability degradation.

[0039] refer to Figure 2 , Figure 2 This graph displays the resistance evolution characteristics of open-circuit faults and poor contact faults. The curve on the left depicts the evolution of the isolation resistance value on the open-circuit fault side over time. The horizontal axis represents time in days, ranging from 0 to 100 days, and the vertical axis represents the isolation resistance value. The unit is ohms, ranging from 0 to 3 ohms. The curve is plotted as a solid red line. In the initial stage, the isolation resistance value is close to the reference value corresponding to 0.5 ohms, and it shows an exponential growth trend over time, reaching approximately 2.5 ohms at 100 days. The blue dashed line in the figure represents the initial resistance reference value, which extends horizontally at the 0.5 ohm position on the vertical axis. The orange dashed line represents the circuit breaker threshold, which is the threshold for determining complete circuit disconnection. The calculated resistance value is Ohms, the orange dashed line extends horizontally at the 1.5 ohm position on the vertical axis. The green translucent filled area covers the range of 0 to 0.575 ohms on the vertical axis; this area represents the allowable tolerance range, and the allowable tolerance percentage... Take 0.15 as the upper limit of the allowable deviation range. Calculated as The resistance value remained within the normal green range for the first 20 days. Between 20 and 40 days, the resistance gradually exceeded the allowable range but did not reach the open circuit threshold. After 40 days, the resistance exceeded the 1.5 ohm open circuit threshold and continued to increase. This characteristic indicates that the open circuit fault exhibits a gradual development process: initially, a slight increase in resistance; in the middle stage, exceeding the allowable range; and in the later stage, reaching a complete disconnection state. Random fluctuations were superimposed on the curve to simulate noise interference in actual measurements, with the fluctuation amplitude controlled within the 0.05 ohm range to ensure a clear and discernible evolution trend.

[0040] The graph on the right depicts the resistance fluctuation characteristics of the faulty contact side. The horizontal axis represents the number of test cycles, ranging from 1 to 20, and the vertical axis represents the isolation resistance value. The unit is ohms, with values ​​ranging from 0.4 to 0.65 ohms. The solid orange line connecting the circular markers represents the isolation resistance values ​​obtained from each measurement; these markers are filled with yellow for easy identification. The blue dashed line extending horizontally at the 0.5 ohm position on the vertical axis represents the reference resistance value. The green semi-transparent filler area covering the 0.425 to 0.575 ohm range represents the permissible deviation range, with the lower limit of this range being... Calculated as Ohm, upper limit The resistance is 0.575 ohms. The measurement curve exhibits periodic fluctuations, with the center of fluctuation oscillating around the baseline value. The 6th measurement reaches a maximum of approximately 0.58 ohms, exceeding the upper limit of the allowable range, while the 14th measurement reaches a minimum of approximately 0.47 ohms, within the allowable range but close to the lower limit. The red arrow in the graph points to the maximum value and is labeled 0.580 ohms, while the blue arrow points to the minimum value and is labeled 0.470 ohms. The difference between these two points constitutes the series range, which is 0.11 ohms. The fluctuation characteristics of poor contact faults stem from the instability of contact resistance at the contact point under mechanical vibration or temperature cycling. In some measurement cycles, increased pressure at the contact point causes a decrease in resistance, while in others, oxidation or loosening of the contact point causes an increase in resistance. This irregular fluctuation contrasts sharply with the monotonic growth trend of open circuit faults. In the fluctuation characteristic vector, the series range and series variance are significantly higher on the poor contact side than on the normal side but lower than on the open circuit side. The side anomaly scoring model utilizes this difference to distinguish fault types.

[0041] When generating a candidate set of abnormal edges using a hierarchical edge anomaly scoring model based on hop count, the main control unit divides the edges into multiple levels according to the hop count from the root node based on the directed graph structure, and processes each level sequentially starting from the root node level. The advantage of hierarchical processing is that edges near the root node have a greater impact on the overall potential distribution; if these edges are abnormal, they will form a wider chain reaction in the topological voltage distribution matrix. Processing the lower hop count levels first can identify key branches earlier, and the anomaly scoring of subsequent levels is less affected by upstream uncertainties. For each edge, the main control unit reads the fluctuation feature vector and inputs it into the edge anomaly scoring model to obtain an anomaly score value. The edge anomaly scoring model can be implemented in various forms. A common choice is to map the fluctuation feature vector to a normalized range of anomaly score values, with the anomaly score value ranging from 0 to 1; a larger value indicates a higher probability of an anomaly. To facilitate clearer explanations during field deployment, the edge anomaly scoring model can establish a sample set based on historical network data during the training phase. Each sample in the set contains a fluctuation feature vector and a corresponding fault label, including open-circuit faults, poor contact faults, insulation damage faults, short-circuit faults, and normal edges. After training, the edge anomaly scoring model outputs anomaly scores for the input fluctuation feature vectors and is allowed to simultaneously output the most likely fault label as reference information. This reference information is used for ranking subsequent validation strategies, but the generation of the candidate set of anomaly edges is still based on the relationship between the anomaly score and the threshold.

[0042] The main control unit marks edges as high-confidence abnormal edges, medium-confidence abnormal edges, or normal edges based on the relationship between the abnormal score value and the first and second preset thresholds. It then aggregates the high-confidence and medium-confidence abnormal edges to form a candidate set of abnormal edges. To ensure both stability and adjustability of the threshold settings, the first and second preset thresholds can be loaded by the main control unit before the start of a project. For example, the first preset threshold can be set to 0.75, and the second preset threshold to 0.45. A higher first preset threshold helps to prioritize the verification of a small number of the most suspicious edges, while a lower second preset threshold is used to include edges with slight abnormal signs in the candidate range, thus balancing speed and coverage in subsequent isolation measurement and insulation detection stages. If the focus on "minimizing missed detections" is greater on-site, the first and second preset thresholds can be lowered simultaneously; if the focus is on "minimizing false alarms," ​​the second preset threshold can be raised, making the candidate set of abnormal edges more refined.

[0043] In terms of alternative implementation methods, the current estimate can be calculated by first approximating and then correcting. Specifically, the first round of current estimates is calculated based on the topological resistance baseline map, resulting in a preliminary sequence of measured resistance estimates. Then, the mean of the measured resistance estimates for each edge is merged with the initial resistance measurement to form an updated resistance estimate. Based on this updated resistance estimate, the current estimate is recalculated to obtain a second round of current estimates that more closely reflects the actual current shunt. A second sequence of measured resistance estimates is then generated. The difference between the two calculations can be stored as a data consistency index. Edges with significant differences often correspond to upstream anomalies or deviations in topological connections; these edges are more likely to receive higher anomaly scores in the edge anomaly scoring model. Another approach is to additionally store the range before truncation for each node voltage value when constructing the topological voltage distribution matrix, as a data acquisition quality marker. When the data acquisition quality marker indicates an anomaly in a certain round of fluctuations, the main control unit reduces the contribution of that round to the calculation of the resistance deviation sequence for all edges, making the fluctuation feature vector more reflective of real structural changes rather than occasional interference.

[0044] Through the above process, multiple rounds of DC excitation voltage and synchronous acquisition transform the potential state of the entire network into a topological voltage distribution matrix; the topological voltage distribution matrix is ​​combined with the topological resistance reference map to make the measured resistance estimation sequence and resistance deviation sequence of the edge level explicit; the fluctuation feature vector compresses the multi-round information into a statistical representation sensitive to anomalies; the edge anomaly scoring model layered by hop count makes the abnormal edge candidate set have an interpretable spatial structure and facilitates the subsequent verification stage to proceed according to priority.

[0045] In one implementation, after obtaining the candidate set of abnormal edges, the main control unit first converts the candidate set into a queue of abnormal edges to be verified. The construction of the queue of abnormal edges to be verified follows a sorting rule from high to low anomaly score. During the sorting process, the anomaly score is used as the primary sorting key, and the hop count of the edge in the directed graph structure is used as the secondary sorting key. The reason for including the hop count in the secondary sorting key is that, under the same anomaly score, the edge closer to the root node has a stronger impact on the potential distribution of the overall loop of the blast network. Verifying edges with low hop counts first makes it easier to fix problems with a large impact range first, so that the isolation action in the subsequent verification process has less measurement disturbance to the remaining edges to be verified. To reduce the impact of sorting on real-time performance, the main control unit can store the candidate set of abnormal edges as a fixed-capacity array structure, for example, with a capacity of 512, and only perform insertion sort on newly added edges each time the candidate set of abnormal edges is updated, thereby keeping the overall sorting time within a predictable range.

[0046] After the queue of abnormal edges to be verified is established, the main control unit retrieves the current edge to be verified from the head of the queue. To ensure consistency in referencing the same edge during isolation measurement and insulation detection, the main control unit generates a one-time verification record for the current edge to be verified. The verification record includes at least the segment conductor number, first endpoint, second endpoint, abnormal score value, verification round number, and timestamp fields. The first and second endpoints are directly obtained from the topology connection table, and their definitions remain fixed: the first endpoint corresponds to the starting vertex number, and the second endpoint corresponds to the ending vertex number. Fixing the directionality of the first and second endpoints ensures the stability of the subsequent reconnection command issuance position, avoiding inconsistencies in the switching matrix access configuration between different intelligent acquisition nodes due to inconsistent directions.

[0047] The main control unit sends an isolation measurement command to the intelligent acquisition node at the first endpoint. The isolation measurement command carries configuration fields such as the segment wire number, the number of measurement repetitions, the number of sampling points, and the sampling interval. The number of measurement repetitions can be set to 3, the number of sampling points can be set to 64, and the sampling interval can be set to 1 millisecond. The advantage of setting the number of measurement repetitions to 3 is that the switching matrix may experience slight jitter at the moment of disconnection and connection. The Kelvin double-arm bridge array is sensitive to transient effects at the contact point. Using 3 repetitions and taking the median value can significantly reduce the impact of occasional jitter on the isolation state resistance value, while not introducing excessively long verification delays.

[0048] After responding to the isolation measurement command, the intelligent acquisition node at the first endpoint position switches the matrix row selection module to select the segmented conductor corresponding to the current edge to be verified, and switches the matrix column selection module to disconnect the segmented conductor from the overall blasting network loop. After the disconnection action is completed, the intelligent acquisition node at the first endpoint position performs a disconnection status confirmation. The disconnection status confirmation can be achieved by acquiring the voltage change trend at both ends of the segmented conductor once in low-energy detection mode using a Kelvin double-arm bridge array: if the voltage value at both ends of the segmented conductor quickly converges to a stable level within a preset observation window after disconnection, the disconnection status is more reliable; if the voltage value at both ends of the segmented conductor still shows the same fluctuation as the overall blasting network loop after disconnection, it indicates that the column selection module has not completed the disconnection or that there is a bypass coupling path. The intelligent acquisition node at the first endpoint position can trigger a repeated disconnection action and record the disconnection status confirmation mark. The reason for adding the disconnection status confirmation is that the measurement of the isolation status resistance value must be carried out under the condition that the segmented conductor does not participate in the parallel current shunting. If the segmented conductor still has a parallel path with the overall circuit of the blasting network, the isolation status resistance value will be pulled down by the parallel resistance, thus misjudging the open circuit fault as a normal side or poor contact fault.

[0049] After the disconnection status is confirmed, the Kelvin double-arm bridge array performs a four-wire precision resistance measurement on the segmented conductor in the disconnected state to obtain the isolation resistance value. The four-wire precision resistance measurement can be solved using the bridge ratio method, which facilitates the suppression of minor fluctuations in the measurement current. The isolation resistance value can be expressed as... ,in This indicates the resistance value in isolation mode; the nominal resistance value of a precision reference resistor group can be expressed as... ,in This indicates the nominal resistance value of the precision reference resistor group; the voltage across the precision reference resistor group can be expressed as... ,in This represents the voltage value acquired through the voltage sensing port of the first bridge arm; the voltage value across the segmented conductor can be expressed as... ,in This represents the voltage value acquired through the voltage sensing port of the second bridge arm. The solution for the isolation state resistance value can be written as: This method allows for the combined effects of factors in a single measurement. and The measured current fluctuation is eliminated in the ratio, allowing different intelligent acquisition nodes to obtain data under different power supply fluctuation conditions. More comparable. In order to make and More stable, the intelligent acquisition node at the first endpoint position performs measurements in each measurement. and Sixty-four sampling points were collected, and the mean of the remaining 56 sampling points was truncated by discarding the four largest and four smallest samples. The mean of these 56 samples was then calculated to obtain the solution. and The isolation resistance value is obtained from three repeated measurements. The intelligent acquisition node at the first endpoint takes the median of the three results as the isolation resistance value and sends it back to the main control unit. At the same time, it sends back a disconnection status confirmation flag and a measurement repetition completion flag.

[0050] The main control unit sends insulation detection commands to the intelligent acquisition node at the second endpoint location in parallel or subsequently. The insulation detection command carries configuration fields such as the segment conductor number, insulation gap selection flag, number of leakage current sampling points, and sampling interval. The insulation gap selection flag indicates the specific selection method for the intelligent acquisition node at the second endpoint location regarding the insulation gap between the segment conductor and adjacent conductors. The selection of adjacent conductors can be based on the set of other edges sharing vertices with the currently verified edge in the topology connection table, prioritizing the conductors corresponding to adjacent edges that rank higher in the anomaly score ranking. This allows for faster confirmation of whether there is mutual influence between abnormal edges, reducing the number of times insulation detection is repeatedly performed on adjacent edges.

[0051] After responding to the insulation detection command, the intelligent acquisition node at the second endpoint switches the matrix row selection module to select the segmented conductor corresponding to the current side to be verified, and switches the matrix column selection module to connect the insulation test source to the insulation gap between the segmented conductor and the adjacent conductor. The output state of the insulation test source is uniformly configured by the main control unit in the insulation detection command. The intelligent acquisition node at the second endpoint applies the insulation test voltage according to the output state of the insulation test source, and the micro-current detection unit collects the leakage current value flowing through the insulation gap. The leakage current value can be expressed as... ,in This indicates the leakage current value; the microcurrent detection unit collects data during a single insulation test. One sampling point, This represents the number of sampling points; the sampling point sequence can be represented as... ,in Indicates the first Leakage current sampling value at each sampling point This indicates the sampling point number. To suppress the impact of peak current caused by the transient connection of the switching matrix on the judgment, the leakage current value is obtained using a median strategy. ,in This represents the median of the sampling point sequence. The median is not sensitive to a small number of spikes, but it can still reflect the steady-state leakage current trend of the insulation gap even in the presence of occasional spikes. For example, A value of 64 can be used, and the sampling interval can be 1 millisecond, so that the insulation test is completed within 64 milliseconds; if the interference in the field environment is stronger, It can be increased to 128 for greater stability. .

[0052] The main control unit receives the isolation status resistance value transmitted back by the intelligent acquisition node at the first endpoint position. Leakage current value transmitted back by the intelligent acquisition node at the second endpoint Then, the initial resistance measurement value corresponding to the segment conductor number in the topological resistance reference map is retrieved. The initial resistance measurement value can be expressed as: ,in This represents the initial resistance measurement value recorded in the topological resistance reference map. To make the determination process of the "allowable deviation range" reproducible, the main control unit represents the allowable deviation range as a proportional interval, where the allowable deviation ratio can be expressed as... ,in This indicates the allowable deviation ratio and is pre-configured by the main control unit, for example... We take 0.15. This gives us the lower limit of the allowable deviation range. upper limit of allowable deviation range ,in This indicates the lower limit of the allowable deviation range. This indicates the upper limit of the allowable deviation range. The reason for using a proportional interval instead of a fixed difference is that there are natural differences in the initial resistance measurements of different conductor segments. Using a proportional interval allows short conductors and long conductors to be judged under the same relative standard, thereby reducing the situation where "short conductors are misjudged by fixed differences".

[0053] At the same time, the main control unit represents the insulation qualification threshold as a fixed leakage current threshold, which can be expressed as: ,in Indicates the insulation pass threshold; insulation testing is judged based on... and The magnitude of the fault is used as the basis. To distinguish between open circuit faults and poor contact faults, the main control unit also uses a complete disconnection judgment threshold. The complete disconnection judgment threshold can be expressed as: ,in This represents the threshold for determining complete disconnection and is used to characterize the deviation of the isolated resistance value from the initial resistance measurement. The degree of deviation can be expressed as the normalized deviation ratio. ,in Indicates the degree of deviation. A positive value indicates that the isolation resistance is higher than the initial resistance measurement. The advantage of using a normalized deviation ratio is that it establishes a clear threshold for determining complete disconnection. It allows for the reuse of conductors across different segments, eliminating the need to set different thresholds for each conductor. For example, A value of 3 can be chosen, indicating that when the isolation resistance value reaches three times or more of the initial resistance measurement, it is closer to an open-circuit fault; when Outside the allowable deviation range but not reached At the corresponding level, it is closer to a poor contact fault.

[0054] During the fault type determination process, the main control unit first combines... and Complementary judgments are formed. The isolation resistance value reflects the conduction characteristics of the segmented conductor in the open state, while the leakage current value reflects the insulation status of the insulation gap between the segmented conductor and adjacent conductors. The benefit of using both simultaneously is that open-circuit faults and poor contact faults often manifest as degraded conduction characteristics while the insulation remains acceptable; insulation damage faults often manifest as insulation degradation while the conduction characteristics may still be close to the baseline; and short-circuit faults often change both conduction characteristics and insulation status simultaneously. Based on this, the main control unit identifies the fault type in the following manner.

[0055] when Exceeding the allowable deviation range and Below the insulation qualification threshold At that time, the main control unit determines the fault type as either an open circuit fault or a poor contact fault. The main control unit then calculates the degree of deviation. ,when Greater than or equal to the complete disconnection determination threshold When the corresponding determination condition is met, the main control unit determines that the fault type is an open circuit fault; when When the determination condition corresponding to the complete disconnection determination threshold is not reached, the main control unit determines that the fault type is a poor contact fault. Placing the open circuit fault and the poor contact fault under the same branch and then further subdividing can avoid directly classifying the situation of "increased contact resistance but not completely disconnected" into the open circuit fault, so as to give a prompt more in line with the on-site status during subsequent construction processing.

[0056] When Is within the allowable deviation range and Exceeds the insulation qualification threshold When, the main control unit determines that the fault type is an insulation damage fault. At this time, the conduction characteristic is close to the reference and continuous leakage occurs in the insulation gap, indicating that the insulation isolation ability between the wires has deteriorated. This deterioration is more likely to stably present as an increase in leakage current under DC excitation. Therefore, using The threshold for determination is more reliable.

[0057] Refer to Figure 3 , Figure 3 Show the characteristic curves of insulation damage faults and short circuit faults. The left curve graph describes the relationship between the insulation damage degree and the leakage current value. The horizontal axis represents the insulation damage degree and the unit is percentage, with a value range from 0 to 100 percentage, and the vertical axis represents the leakage current value And the unit is microampere, with a value range from 0 to 50 microampere. The purple solid line connecting the circular marked points represents the leakage current values measured under different damage degrees, and the marked points are filled with light purple. The red dashed line extends horizontally at the 10 microampere position on the vertical axis to represent the insulation qualification threshold , this threshold divides the insulation state into two intervals: qualified and unqualified. The green semi-transparent filled area covers the range of 0 to 10 microampere on the vertical axis and is marked as the qualified area, and the red semi-transparent filled area covers the range of 10 to 50 microampere on the vertical axis and is marked as the unqualified area. The leakage current curve remains below 5 microampere in the insulation damage degree interval of 0 to 20 percentage, located in the insulation qualified area. In the insulation damage degree interval of 20 to 40 percentage, the leakage current value rises from 5 microampere to about 15 microampere, crossing the insulation qualification threshold and entering the unqualified area. In the insulation damage degree interval of 40 to 100 percentage, the leakage current value accelerates growth and reaches about 45 microampere at 100 percentage damage. The curve growth law conforms to the non-linear characteristic, and uses To describe the relationship Represents the percentage of damage degree, the linear term Characterizes the slow growth in the initial stage of damage, and the non-linear term Characterizes the accelerated deterioration in the middle and late stages of damage. The insulation damage fault is in the isolation resistance value While maintaining a value close to the baseline, the leakage current value significantly exceeds the insulation qualification threshold. This combination of characteristics enables the fault judgment logic to distinguish insulation damage faults from open circuit faults and poor contact faults, because the latter two usually manifest as abnormal isolation resistance values ​​while the leakage current value is still within the qualification range.

[0058] The scatter plot on the right describes the combined distribution characteristics of isolation resistance and leakage current values ​​during a short-circuit fault. The horizontal axis represents the isolation resistance value. The unit is ohms, with values ​​ranging from 0.05 to 0.45 ohms. The vertical axis represents the leakage current value. The unit is microamps, ranging from 0 to 50 microamps. The scatter plot uses a color mapping scheme, transitioning from cool blue to warm red. The color intensity corresponds to the resistance value; scatter plots near a resistance of 0.1 ohms appear blue, and those near a resistance of 0.4 ohms appear red. The overall distribution of the scatter plots shows a negative correlation: when the isolation resistance is in the low range of 0.1 to 0.2 ohms, the leakage current is concentrated in the high range of 30 to 40 microamps; when the isolation resistance increases to 0.3 to 0.4 ohms, the leakage current decreases to the range of 15 to 25 microamps. The black dashed line represents a cubic polynomial fitting curve, expressed by the fitting formula: The exponential decay term describes the physical law that the leakage current increases sharply when the short circuit severity worsens, i.e., the resistance value decreases. The orange dashed line extending vertically at the 0.425 ohm position on the horizontal axis represents the lower limit of the resistance, which is the lower limit of the allowable deviation range. According to ohm calculations, the horizontal extension of the purple dashed line at the 10 microamp position on the vertical axis represents the leakage current threshold. A red, semi-transparent filled area covers a rectangular region ranging from 0.1 to 0.425 ohms on the horizontal axis and 10 to 50 microamps on the vertical axis. This region is marked as the short-circuit fault zone. Sample points within this zone must simultaneously meet two criteria: an isolation resistance value below the lower limit of the allowable deviation range and a leakage current value exceeding the insulation qualification threshold. The short-circuit fault determination logic requires… and The simultaneous fulfillment of these two conditions prevents a simple low resistance or a simple high leakage current from being misjudged as a short circuit fault. Colored bars on the right side of the graph indicate the resistance value range, with scales from 0.1 to 0.4 ohms, allowing for intuitive reading of the corresponding resistance value through the scatter color.

[0059] when Below the lower limit of the allowable deviation range and Exceeding the insulation qualification threshold When the fault type is determined to be short circuit, the main control unit determines it to be a short circuit fault. The introduction of the lower limit of the allowable deviation range provides a clear standard for the judgment of "below the benchmark" and avoids uncertain statements such as "slightly lower". At the same time, the leakage current value is included in the short circuit fault judgment so that the scenario of "conductivity becomes stronger but insulation is still qualified" will not be mistakenly classified as a short circuit fault.

[0060] Once the fault type is determined, the main control unit sends a reconnection command to the intelligent acquisition node at the first endpoint. After responding to the reconnection command, the intelligent acquisition node at the first endpoint switches the matrix column selection module to reconnect the segmented conductor to the overall loop of the blast network. Performing the reconnection immediately after each edge verification reduces the impact of prolonged disconnection on the node voltage distribution of subsequent edges to be verified, allowing the isolation measurement and insulation detection of subsequent edges to be completed under conditions closer to the original topology. After the reconnection is completed, the main control unit records the segmented conductor number, first endpoint, second endpoint, anomaly score, and isolation resistance value of the current edge to be verified. Leakage current value The fault type, disconnection status confirmation flag, and measurement repetition count completion flag are written into the fault diagnosis result list, and the current edge to be verified is removed from the queue of abnormal edges to be verified.

[0061] The main control unit repeatedly executes the above process on the queue of abnormal sides to be verified until the queue is empty, and then outputs a list of fault diagnosis results as the fault diagnosis results. In order to make the output results directly usable for on-site verification, the list of fault diagnosis results can be sorted by abnormal score from high to low, and the verification timestamp is retained for easy tracing of the verification order.

[0062] In terms of optional implementation methods, isolation measurement commands and insulation detection commands can be issued in parallel. During parallel issuance, the main control unit simultaneously issues isolation measurement commands to the intelligent acquisition node at the first endpoint and insulation detection commands to the intelligent acquisition node at the second endpoint for the same current edge to be verified, and sets a unified verification round number, so that the two types of returned data naturally correspond to the same verification round number. The advantage of the parallel strategy is that it shortens the total verification time for a single edge, especially when the size of the candidate set of abnormal edges is large, it can significantly reduce the overall verification time. Another optional approach is to add an adjacent conductor rotation strategy to the insulation detection command: for the same segment of conductor, two adjacent conductors are selected sequentially to form two insulation gaps and obtain two leakage current values ​​respectively. The main control unit takes the larger of the two leakage current values ​​as the leakage current value. This allows for the inclusion of situations where "leakage to a particular adjacent conductor is more pronounced" in the judgment, thereby increasing the sensitivity to insulation damage faults and short-circuit faults.

[0063] Another optional approach is used to handle cases where the anomaly score output by the edge anomaly scoring model is close to a threshold. When constructing the queue of anomaly edges to be verified, the main control unit marks edges whose anomaly scores fall near a second preset threshold as low-priority verification edges. During the verification phase, an additional round of retesting is added to these low-priority verification edges; for example, the number of measurement repetitions is increased from 3 to 5. Furthermore, the interquartile range is calculated based on the median of the isolation state resistance values ​​from the 5 retests, and this interquartile range is used as a measurement stability reference and written into the fault diagnosis result list. This approach improves the measurement stability of boundary samples and reduces "judgment oscillations caused by fluctuations near the threshold" without altering the fault type determination logic.

[0064] Through the above process, the candidate set of abnormal edges is transformed into an executable verification sequence. The switching matrix provides controllable disconnection and connection actions, enabling the isolation state resistance value to reflect the conductivity characteristics of the segmented conductor itself. The insulation test source and micro-current detection unit provide independent observation of the insulation gap, enabling the leakage current value to reflect the insulation status. The combined judgment of the isolation state resistance value and leakage current value distinguishes between open circuit faults, poor contact faults, insulation damage faults, and short circuit faults. Furthermore, the connection is restored after each verification, ensuring that subsequent verifications proceed under consistent overall loop conditions. The final fault diagnosis result output has consistent fields and traceable verification records.

[0065] In summary, the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for diagnosing hybrid topology electrical parameters in blasting network configurations, characterized in that, Includes the following steps: Step S1: Abstract the blasting network into a directed graph structure and deploy intelligent acquisition nodes containing Kelvin double-arm bridge arrays and switching matrices at each vertex. Scan each edge to obtain the initial resistance measurement values ​​of the segmented conductors and generate a topological resistance reference map. Step S2: Apply multiple rounds of excitation at the root node and simultaneously collect the node voltage values ​​of each vertex, and construct the topology voltage distribution matrix according to the vertex number; Step S3: Calculate the sequence of measured resistance values ​​for each side based on the topological voltage distribution matrix, differentiate it with the topological resistance reference map and extract the fluctuation feature vector, and generate a candidate set of abnormal sides by inputting it into the side anomaly scoring model according to the number of jumps. Step S4: Perform switching matrix isolation measurement and insulation detection on the candidate set of abnormal edges according to the abnormality score, and determine the fault type by combining the isolation state resistance value and leakage current value and output the fault diagnosis result.

2. The method as described in claim 1, characterized in that, In step S1, each detonator node in the directed graph structure is a vertex, each connecting wire segment is an edge, and the root node corresponds to the blasting network bus entry point; vertex numbers are set for vertices and segment wire numbers are set for edges, and a topology connection table is established based on vertex numbers and segment wire numbers.

3. The method as described in claim 2, characterized in that, The Kelvin dual-arm bridge array includes a first bridge arm and a second bridge arm. The first bridge arm is connected to a precision reference resistor group, and the second bridge arm is connected to the segmented conductor to be measured. The switching matrix includes a row gating module and a column gating module. The row gating module selects the current segmented conductor number to be measured, and the column gating module selects the bridge unit number of the Kelvin dual-arm bridge array to participate in the measurement and establishes the connection configuration of the first bridge arm and the second bridge arm.

4. The method as described in claim 3, characterized in that, In step S1, the control switching matrix scans the segmented wires corresponding to each edge of the directed graph structure and obtains the initial resistance measurement value of each segmented wire. The initial resistance measurement value of each segmented wire is stored to form a topological resistance reference map. Obtaining the initial resistance measurement value of the segmented conductor includes: sending an initialization command to each intelligent acquisition node; after the intelligent acquisition node responds to the initialization command, the switching matrix row selection module sequentially selects each segmented conductor according to a preset scanning order, and the switching matrix column selection module connects the selected segmented conductor to the second arm of the Kelvin double-arm bridge array; the Kelvin double-arm bridge array injects a constant measurement current into the second arm of the bridge array through the current injection port of the first arm; the Kelvin double-arm bridge array acquires the voltage value across the precision reference resistor group through the voltage sensing port of the first arm and the voltage value across the segmented conductor through the voltage sensing port of the second arm; the initial resistance measurement value of the segmented conductor is calculated based on the voltage value across the precision reference resistor group, the voltage value across the segmented conductor, and the nominal resistance value of the precision reference resistor group; and the initial resistance measurement value of each segmented conductor is stored in the topology resistance database to form a topology resistance reference map.

5. The method as described in claim 4, characterized in that, Step S2, constructing the topology voltage distribution matrix, includes: sending a voltage divider excitation command to the intelligent acquisition node at the root node position; after the intelligent acquisition node at the root node position responds to the voltage divider excitation command, it switches the matrix row selection module to connect the excitation voltage source to the bus input and outputs the first round of DC excitation voltage; under the first round of DC excitation voltage applied state, a synchronous acquisition command is broadcast to all intelligent acquisition nodes; after all intelligent acquisition nodes respond to the synchronous acquisition command, they acquire node voltage values ​​through the voltage sensing port of the second bridge arm of the Kelvin double-arm bridge array and transmit the node voltage values ​​back through the bus; the node voltage values ​​are the voltage values ​​of the location of the intelligent acquisition node relative to the common reference ground; the node voltage values ​​are arranged in order of vertex number to form the first round of node voltage vector; the excitation voltage source is controlled to output the second round of DC excitation voltage to multiple rounds of DC excitation voltage in sequence, and the process of broadcasting the synchronous acquisition command, transmitting the node voltage values, and constructing the node voltage vector is repeated; the multiple rounds of node voltage vectors are stacked row by row to form the topology voltage distribution matrix and stored in the voltage divider response database.

6. The method as described in claim 5, characterized in that, Step S3 involves solving the sequence of estimated measured resistance values, which includes: obtaining the difference between node voltage values ​​between adjacent vertices based on the topological voltage distribution matrix and obtaining the current estimate of the edge flowing through it; generating the estimated measured resistance values ​​of the edge under each round of DC excitation voltage conditions based on the difference between node voltage values ​​and the current estimate, and forming the sequence of estimated measured resistance values; and obtaining the current estimate based on the output state of the excitation voltage source and the topological resistance reference spectrum.

7. The method as described in claim 6, characterized in that, Step S3 involves extracting the fluctuation feature vector as follows: extracting the initial resistance measurement value corresponding to the edge from the topological resistance reference map as the reference value; obtaining the resistance deviation value by subtracting the reference value from each measured resistance estimate in the measured resistance estimate sequence; and forming a resistance deviation sequence by calculating the sequence mean, sequence range, and sequence variance of the resistance deviation sequence and forming a fluctuation feature vector from the sequence mean, sequence range, and sequence variance.

8. The method as described in claim 7, characterized in that, Step S3, which involves generating a candidate set of abnormal edges by inputting the edge anomaly scoring model in layers according to the number of hops, includes: dividing the edges into multiple layers based on the number of hops from the root node according to the directed graph structure, and processing each layer sequentially starting from the root node layer; reading the fluctuation feature vector for each edge and inputting it into the edge anomaly scoring model to obtain an anomaly score value; marking the edges as high-confidence abnormal edges, medium-confidence abnormal edges, or normal edges based on the relationship between the anomaly score value and the first preset threshold and the second preset threshold; and summarizing the high-confidence abnormal edges and medium-confidence abnormal edges to form a candidate set of abnormal edges.

9. The method as described in claim 8, characterized in that, Step S4, determining the fault type and outputting the fault diagnosis result, includes: sorting the abnormal edge candidate set from highest to lowest abnormal edge score to form a queue of abnormal edges to be verified, and sequentially selecting the current edge to be verified; determining that the current edge to be verified connects the first endpoint and the second endpoint; sending an isolation measurement command to the intelligent acquisition node at the first endpoint; after the intelligent acquisition node at the first endpoint responds to the isolation measurement command, switching the matrix row selection module to select the segmented conductor corresponding to the current edge to be verified, and switching the matrix column selection module to disconnect the segmented conductor from the overall circuit of the blasting network; and performing a four-wire precision resistance measurement on the segmented conductor in the disconnected state using the Kelvin double-arm bridge array to obtain the isolation state resistance value; sending an insulation detection command to the intelligent acquisition node at the second endpoint; after the intelligent acquisition node at the second endpoint responds to the insulation detection command, switching the matrix row selection module to select the segmented conductor corresponding to the current edge to be verified, and switching the matrix column selection module to disconnect the segmented conductor from the overall circuit of the blasting network; and performing a four-wire precision resistance measurement on the segmented conductor using the Kelvin double-arm bridge array in the disconnected state to obtain the isolation state resistance value; and sending an insulation detection command to the intelligent acquisition node at the second endpoint; after the intelligent acquisition node at the second endpoint responds to the insulation detection command, switching the matrix row selection module to select the segmented conductor corresponding to the current edge to be verified, and switching the matrix column selection module to disconnect the segmented conductor from the overall circuit of the blasting network. The array gating module connects the insulation test source to the insulation gap between the segmented conductor and the adjacent conductor. The micro-current detection unit collects the leakage current value flowing through the insulation gap. Based on the comparison results of the isolation state resistance value and the initial resistance measurement value, as well as the comparison results of the leakage current value and the insulation qualification threshold, the fault type is determined. The fault types include open circuit fault, poor contact fault, insulation damage fault, and short circuit fault. A restoration connection command is sent to the intelligent acquisition node at the first endpoint position to cause the switching matrix column gating module to reconnect the segmented conductor to the overall loop of the blasting network. The fault type of the current edge to be verified is recorded to the fault diagnosis result list and the current edge to be verified is removed from the queue of abnormal edges to be verified. The process of selecting the queue of abnormal edges to be verified, sending the isolation measurement command, sending the insulation detection command, determining the fault type, and sending the restoration connection command is executed cyclically until the queue of abnormal edges to be verified is empty. The fault diagnosis result list is then output as the fault diagnosis result.

10. A topology-parameter hybrid fault diagnosis device for implementing the method of any one of claims 1 to 9, characterized in that, The system includes a main control unit, multiple intelligent acquisition nodes, a bus, a topology resistance database, and a voltage divider response database. Multiple intelligent acquisition nodes are deployed at each vertex of the directed graph structure of the blasting network. Each intelligent acquisition node includes a Kelvin double-arm bridge array and a switching matrix. The switching matrix includes a row selection module and a column selection module. The Kelvin double-arm bridge array includes a first bridge arm and a second bridge arm. The first bridge arm connects to a precision reference resistor group, and the second bridge arm connects to the conductor segment to be tested. The intelligent acquisition node at the root node includes an excitation voltage source. Each intelligent acquisition node also includes an insulation test source and a micro-current detection unit. The main control unit is connected to the bus and configured to: construct the directed graph structure. The system controls the row and column selection modules to scan the segmented conductors, obtain initial resistance measurements, and write them into the topology resistance database to form a topology resistance reference map. It also controls the excitation voltage source to output multiple rounds of DC excitation voltage and controls the intelligent acquisition nodes to synchronously acquire node voltage values ​​to form a topology voltage distribution matrix, which is then written into the voltage divider response database. Based on the topology resistance reference map and the topology voltage distribution matrix, it generates a candidate set of abnormal edges. The system controls the switching matrix to disconnect the segmented conductors, controls the Kelvin double-arm bridge array to obtain the isolation state resistance value, and controls the insulation test source and micro-current detection unit to obtain the leakage current value. Based on the isolation state resistance value and the leakage current value, it outputs the fault diagnosis result.

Images