A method for intelligent assessment of contact ablation of a molded case circuit breaker

Abstract

The application discloses a kind of intelligent evaluation methods of contact ablation of moulded case circuit breaker, specifically related to low-voltage electrical apparatus condition monitoring and fault diagnosis technical field, for solving the problem that real ablation signs and non-ablation factors such as load fluctuation, environmental change, terminal connection anomaly, heat dissipation limitation, mechanism hysteresis are effectively distinguished in the process of existing moulded case circuit breaker contact ablation evaluation.By identifying opening event, closing event, tripping event and recovery power transmission event and establishing corresponding event measurement window, operating current, terminal voltage drop, near contact area temperature rise response and opening and closing action timing are collected in the window, then combined with load state, ambient temperature and event type to perform working condition normalization processing, and construct interference exclusion matrix to exclude event characteristics corresponding to terminal connection anomaly, heat dissipation limitation anomaly and mechanism hysteresis anomaly, so as to effectively distinguish contact real ablation signs and non-ablation interference factors, improve contact ablation identification accuracy and result reliability.

Description

Technical Field

[0001] This invention relates to the field of low-voltage electrical appliance condition monitoring and fault diagnosis technology, specifically to an intelligent assessment method for contact erosion of molded case circuit breakers. Background Technology

[0002] Molded case circuit breakers (MCCBs) are widely used in low-voltage distribution cabinets, power distribution boxes, charging pile power distribution units, and end-of-line power distribution branches in industrial production lines, undertaking functions such as branch connection, disconnection, and protection isolation. During long-term operation under load, frequent start-stop operations, fault isolation, and power restoration, the contact parts are prone to burning, contact deterioration, and decreased conductivity. Current on-site maintenance typically assesses the circuit breaker's health status through infrared thermography, single-time current detection, terminal voltage sampling, manual inspection, or power outage inspection. Some solutions combine auxiliary contacts, mechanical positions, or protection action records to monitor the operating status. These methods can reflect the presence of abnormalities to some extent, but most remain at the level of single signal discrimination, single-time state comparison, or static threshold alarms, mainly addressing the questions of "is it abnormal?" or "does it need to be re-inspected?"

[0003] Existing technologies for assessing contact erosion in molded case circuit breakers (MCCBs) generally face a more specific and prominent difficulty: phenomena such as increased inter-terminal voltage drop, increased local temperature rise, and fluctuations in operating timing observed during operation are not necessarily caused by contact erosion itself. Factors such as changes in branch load conditions, ambient temperature variations, abnormal terminal connections, differences in heat dissipation conditions in the installation space, and sluggish mechanism movements can all cause the test results to exhibit external characteristics similar to contact erosion. Without distinguishing between different operating events, establishing a unified observation scope around the events, normalizing operating conditions by incorporating load conditions and ambient temperature, and eliminating non-erosion interference factors, the conclusions drawn from existing technologies are prone to misclassifying non-erosion factors as contact erosion and easily masking true erosion signs amidst complex operating condition fluctuations. This results in insufficient stability of the judgment results, often limiting their application in the field to only supplementary references.

[0004] Therefore, in the actual operating scenarios of molded case circuit breakers, how can existing technologies effectively distinguish between actual contact erosion signs and non-erosion interference factors from continuous field data collection records under different operating events such as opening, closing, tripping, and power restoration, as well as different loads and environmental conditions, to obtain more targeted and consistent assessment data? This problem is not simply about improving the accuracy of a single detection quantity, nor is it simply about adding a single monitoring method. Rather, it's about ensuring the comparability, distinguishability, and reliability of contact erosion assessment results for subsequent maintenance and repair under complex operating conditions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent assessment method for contact erosion in molded case circuit breakers, thereby resolving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a smart assessment method for contact erosion of molded case circuit breakers, comprising: S1. Identify the opening, closing, tripping, and power restoration events of the molded case circuit breaker, and establish an event measurement window around each target operation event; S2. Collect operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, and event type identifier within the event measurement window to form an event measurement dataset; S3. Perform operating condition normalization processing on the event measurement dataset based on load status, ambient temperature, and event type to generate a comparable event feature set; S4. Construct an interference elimination matrix based on the comparable event feature set to screen out event features corresponding to terminal connection abnormalities, heat dissipation limitation abnormalities, and mechanism hysteresis abnormalities. S5. Extract conductive degradation features, thermal response hysteresis features, and post-action recovery features from the event features after filtering out interference, and combine them to form an ablation evidence cluster. S6. Generate contact ablation grading results based on the evidence strength, consistency, and persistence of the ablation evidence cluster, and output the corresponding maintenance judgment.

[0007] Furthermore, S1 includes: Event identification is performed in a fixed sequence for tripping events, tripping events, closing events, and power restoration events. The confirmed target runtime event reference time point is the timestamp of the candidate state change point; Establish an event measurement window around a reference time point; The event measurement window includes the pre-event baseline interval corresponding to the preceding observation segment and the post-event response interval corresponding to the following observation segment.

[0008] Furthermore, S2 includes: Capture records within the window based on the event measurement window boundaries; Alignment is performed using a unified timestamp; Records are grouped into the same group within the window based on the device's unique identifier, branch's unique identifier, event type identifier, reference time point, sampling time, and phase. An event measurement dataset is formed by using the device unique identifier, branch unique identifier, event type identifier, reference time point, sampling time, phase, operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, record source version number, rule version number, and cause code.

[0009] Furthermore, S3 includes: First, group and normalize according to load status, then partition and normalize according to ambient temperature, and finally form corresponding normalized fields according to event type. The normalized operating current is recorded as a percentage of the rated current. The normalized inter-terminal voltage drop is the original inter-terminal voltage drop record with uniform load state boundary attached; The normalized near-contact temperature rise response is the original near-contact temperature rise response record with ambient temperature range markings and unified load state boundaries; The unified opening and closing action sequence is the original opening and closing action sequence record under the corresponding event type, which retains the original time difference direction and time difference unit.

[0010] Furthermore, S4 includes: The data are grouped together according to the equipment unique identifier, branch unique identifier, event type identifier, reference time point, phase and measurement window under the same event, including the unified operating current, unified inter-terminal voltage drop, unified near-contact temperature rise response, unified opening and closing action sequence, load status, and ambient temperature range identifier. An interference exclusion matrix is ​​formed by using a single event measurement window as the granularity and by phase. The interference exclusion matrix includes terminal connection anomaly markers, heat dissipation limitation anomaly markers, mechanism hysteresis anomaly markers, exclusion field identifiers, and reserved event feature identifiers.

[0011] Furthermore, S4 also includes: Terminal connection anomalies are only filtered out from the normalized inter-terminal voltage drop field. The heat dissipation limitation anomaly only filters out the normalized near-contact area temperature rise response field; The mechanism lag anomaly only filters out the normalized opening and closing action timing field; In the filtered records, the corresponding fields are marked as exclusion fields and are not included in the retained event characteristics when called downstream.

[0012] Furthermore, S5 includes: Extract conductivity degradation features, thermal response hysteresis features, and post-action recovery features based on the device's unique identifier, branch's unique identifier, event type identifier, reference time point, and phase. Using a single event measurement window and a single phase as the granularity, the successfully extracted conductivity degradation features, thermal response hysteresis features, and post-action recovery features are merged into ablation evidence clusters according to the unique equipment identifier, branch unique identifier, event type identifier, reference time point, and phase. The minimum condition for the ablation evidence cluster to be valid is that at least two of the three types of features are successfully extracted.

[0013] Furthermore, S6 includes: Retrieve conductivity degradation characteristics, thermal response hysteresis characteristics, post-action recovery characteristics, and corresponding missing reason codes by device unique identifier, branch unique identifier, event type identifier, reference time point, and phase; Using a single event measurement window and a single phase as the granularity, contact ablation grading results are generated in a fixed order of evidence strength, evidence consistency, and evidence persistence.

[0014] Furthermore, S6 also includes outputting corresponding maintenance judgments based on the contact erosion classification results, wherein: When the strength of evidence reaches the slight boundary and the consistency of evidence is established, a slight ablation is generated and a decision to continue running is output. When the strength of evidence reaches the moderate boundary, the consistency of evidence is established, and the persistence of evidence reaches the persistence threshold, a moderate ablation is generated and a planned maintenance judgment is output. When the strength of evidence reaches the severe boundary, the consistency of evidence is established, and the persistence of evidence reaches the severe persistence threshold, severe ablation is generated and a priority repair judgment is output.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By identifying opening events, closing events, tripping events, and power restoration events, and establishing an event measurement window around the target operating events, the system collects operating current, inter-terminal voltage drop, near-contact temperature rise response, and opening / closing action sequence within the event measurement window. Then, based on the load status, ambient temperature, and event type, it performs operating condition normalization processing and further constructs an interference elimination matrix to screen out event characteristics corresponding to abnormal terminal connections, heat dissipation limitations, and mechanism hysteresis. This achieves the goal of distinguishing actual contact erosion signs from non-erosion interference factors, improving the accuracy and reliability of contact erosion identification.

[0016] 2. By extracting conductivity degradation features, thermal response hysteresis features, and post-action recovery features from the event features after filtering out interference, and combining them to form an ablation evidence cluster, and then generating contact ablation grading results based on evidence strength, evidence consistency, and evidence persistence, and outputting corresponding maintenance judgments, the dispersed operation signs are transformed into gradeable, traceable, and executable maintenance basis, thereby improving the pertinence of maintenance judgments and reducing the risk of blind maintenance and missed inspections. Attached Figure Description

[0017] Figure 1 This is a flowchart of an intelligent assessment method for contact erosion in molded case circuit breakers. Figure 2 A schematic diagram illustrating the establishment of a target-based event identification and event measurement window; Figure 3 A schematic diagram illustrating the structure of the event measurement dataset; Figure 4 This is a schematic diagram of the normalization process for operating conditions; Figure 5 A schematic diagram illustrating the relationship between interference exclusion matrix construction and screening; Figure 6 This is a schematic diagram illustrating the formation and grading of ablation evidence clusters. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example: Combined with Appendix Figure 1-6 This embodiment provides a smart assessment method for contact erosion of molded case circuit breakers, including: S1. Identify the opening, closing, tripping, and power restoration events of the molded case circuit breaker, and establish an event measurement window for each target operational event. The specific implementation is as follows: First, the opening, closing, tripping, and power restoration events of the molded case circuit breaker are identified. Then, an event measurement window is established around each target operating event. This action occurs at the operating site of the molded case circuit breaker in low-voltage distribution cabinets, power distribution boxes, charging pile distribution units, and industrial production line terminal distribution branches. It is completed collaboratively by the field acquisition device and local controller connected to the molded case circuit breaker. The executing entity can be set as the event identification program in the cabinet control terminal. The time period covers the equipment's load operation period, load switching period, fault clearing period, and post-maintenance recovery period. The upstream receives the real-time operating records of the molded case circuit breaker, and the downstream provides the confirmed target operating events and their corresponding event measurement windows to the event measurement dataset formation stage.

[0020] A tripping event is a field action record of the molded case circuit breaker contacts changing from a closed state to an open state; a closing event is a field action record of the molded case circuit breaker contacts changing from an open state to a closed state; a tripping event is a record of the automatic tripping of the molded case circuit breaker under overload, short circuit, or protection linkage conditions; and a power restoration event is a field action record of the corresponding branch being re-energized and entering a stable power supply state after the previous trip. A stable power supply state is defined as the branch voltage continuously exceeding 90% of the rated voltage and the auxiliary contacts continuously remaining closed. For branches with loads after power restoration, the branch current must not be zero within three consecutive sampling periods, and the fluctuation amplitude must not exceed 20% of the average value of those three sampling periods. For branches in an unloaded standby state after power restoration, a non-zero branch current is not a necessary condition.

[0021] The branch voltage is recorded and fixedly taken from the sampling terminal on the load side of the circuit breaker. In a three-phase branch, the three phases are sampled separately, and the condition for the condition to be met is that the sampled values ​​of the three phases simultaneously reach more than 90% of the rated phase voltage. In a single-phase branch, the condition for the condition to be met is that the sampled value between the incoming and outgoing terminals on the load side reaches more than 90% of the rated voltage. The condition for the condition to be met is not that any single phase is met, nor is the condition for the condition to be met by the average value of multiple phases.

[0022] The record sources used for the aforementioned event identification are fixed as auxiliary contact status records, mechanical position status records, shunt trip circuit status records, trip indication contact records, branch voltage existence records, branch current records, and manual operation marks. The auxiliary contact status records are the circuit breaker auxiliary contact conduction status records. When using normally open auxiliary contacts, conduction is fixedly mapped to a closed state, and disconnection is fixedly mapped to an open state. When using normally closed auxiliary contacts, conduction is fixedly mapped to an open state, and disconnection is fixedly mapped to a closed state. The auxiliary contact wiring method, once locked in the record source version, cannot be changed within the same continuous operating cycle. The mechanical position status records are the circuit breaker operating mechanism position records. Status code 1 is fixed to indicate a closed state, and status code 0 is fixed to indicate an open state. Other status codes are all transferred to pending verification. When the auxiliary contact status records and mechanical position status records give opposite conclusions at the same time, the primary judgment source takes priority, and the primary judgment conclusion is not rewritten by the auxiliary verification source, but a conflict mark should be added. If the primary judgment source itself is invalid, it is directly transferred to pending verification.

[0023] The branch current record is a current sampling record of the sampling point on the load side of the circuit breaker. The position status unit is status code, the branch voltage unit is volt, the branch current unit is ampere, and the timestamp unit is millisecond. The sampling rhythm can be set from 10 milliseconds to 100 milliseconds, preferably 20 milliseconds. The timestamp alignment tolerance is no more than 5 milliseconds. All records are locked according to the same rule version and the same record source version, and cannot be switched within the same continuous operating cycle.

[0024] Before each record enters the identification process, it is first sorted and aligned according to the local acquisition timestamp of the unified clock source. Then, single-point transitions and backswing segments with durations shorter than the state maintenance threshold are removed. Next, forward padding is performed on the allowed padding fields, and finally, event merging is completed. The candidate state change point takes the timestamp of the first state flip record after timestamp alignment and jitter removal, excluding the timestamp generated after padding and the timestamp of stability determination completion. The reference time point of the confirmed target running event is fixed equal to the timestamp of the candidate state change point and does not change due to the establishment time of the auxiliary verification source, the establishment time of the stable power supply state, or the manual review time. State maintenance The threshold and noise reduction threshold use the same value. The allowed fields to be filled in are fixed as branch voltage existence records and branch current records. Forward filling is allowed when the continuous missing data length does not exceed 200 milliseconds. Auxiliary contact status records, mechanical position status records, shunt trip circuit status records, trip indication contact records, and manual operation marks must not be automatically filled in. If any of the following seven situations occur, the data will be stopped and the data will be transferred to pending verification or pending data entry: reversed timestamp, coexistence of closed and open states at the same time, coexistence of manual trip mark and trip mark, missing data exceeding the filling boundary, missing rule version, missing equipment unique identifier, or missing branch unique identifier.

[0025] Event identification is performed in a fixed order: tripping event, opening event, closing event, and power restoration event. When a previous event type is established, no record of the next event type is generated for the same reference time point. Specifically, after a tripping event is established, only the tripping event record is retained and the contact opening result is marked. No separate opening event record is generated for the same reference time point. The closing event is confirmed based on the change of the contact state from open to closed. The power restoration event is confirmed based on the branch regaining a stable power supply state after the previous disconnection. If the closing event is established but the branch has not formed a stable power supply state, the power restoration event is not confirmed.

[0026] The primary source for determining the tripping event is the mechanical position status record, and the secondary source for verification is the auxiliary contact status record and the branch current record; the primary source for determining the closing event is the mechanical position status record, and the secondary source for verification is the auxiliary contact status record and the branch voltage existence record; the primary source for determining the tripping event is the tripping indication contact record, and the secondary source for verification is the mechanical position status record and the manual operation mark; the primary source for determining the power restoration event is the branch voltage existence record, and the secondary source for verification is the auxiliary contact status record and the branch current record.

[0027] A pre-observation segment and a post-observation segment are established around the candidate state change point. The pre-observation segment is used to confirm that the state remains stable before the change, and the post-observation segment is used to confirm that the state remains stable after the change. The pre-observation segment can be set to 300 milliseconds to 3000 milliseconds, and the post-observation segment can be set to 500 milliseconds to 5000 milliseconds. During the first round of adjustment, the default value is determined based on the continuous 7-day operation record of the same type of molded case circuit breaker. The state holding threshold is the 95th percentile value of the stable holding duration record of the same type of event within the 7 days, rounded up. The pre-observation segment is the 95th percentile value of the continuous duration record of the stable state before the candidate state change point within the 7 days, rounded up. The post-observation segment is the 95th percentile value of the record of the time required for the state to complete the stability confirmation after the candidate state change point within the 7 days, rounded up. If there are fewer than 30 valid samples of any event type within 7 days, the default value of the rule version is adopted for that event type. No new thresholds can be generated before the first round of adjustment is completed.

[0028] The target operation event is confirmed when the primary judgment source is established and the secondary verification source meets the corresponding consistency conditions within the post-observation period. Specifically, the consistency conditions for a tripping event are: the disconnected state continuously reaches the state holding threshold, and the branch current drops below 5% of the rated current for three consecutive sampling cycles within the post-observation period. The consistency conditions for a closing event are: the closed state continuously reaches the state holding threshold, and the branch voltage recovers to above 90% of the rated voltage for three consecutive sampling cycles within the post-observation period. The consistency conditions for a tripping event are: the tripping mark is earlier than or synchronous with the establishment of the disconnected state, and the manual operation mark is missing. The consistency conditions for a power restoration event are: the branch voltage recovers and the auxiliary contact closes, both records are established within the post-observation period. For branches where the load current rises after power restoration, the branch current must reappear and remain for three consecutive sampling cycles. For branches in an unloaded standby state after power restoration, the condition for establishment is that the branch voltage is continuously higher than 90% of the rated voltage and continuously reaches the state holding threshold; the reappearance of the branch current is no longer a necessary condition.

[0029] The status maintenance requirement is that each sampling point maintains the same status, allowing one completed sampling point to exist, and reverse transitions are not allowed; manual maintenance lock is valid when the lock mark from the maintenance operation ticket record or the local maintenance switch record is in effect, and new events are not confirmed during the period when the manual maintenance lock is valid.

[0030] An event measurement window is established using the confirmed target operation event reference time point as the central reference point. The event measurement window consists of a pre-event baseline interval and a post-event response interval. The pre-event baseline interval corresponds to the pre-observation segment, and the post-event response interval corresponds to the post-observation segment. It retains the event type, reference time point, pre-boundary time, post-boundary time, unique equipment identifier, unique branch identifier, record source version number, rule version number, and manual review mark. The event measurement window is stored in the event recording area of ​​the field control terminal in a structured text format and is simultaneously written to the timing recording area of ​​the power distribution monitoring host. The next step directly calls the record according to the unique equipment identifier, unique branch identifier, event type, and reference time point. The original boundary cannot be rewritten. Only the addition of review records is allowed, and the original records cannot be overwritten.

[0031] The field control terminal and the power distribution monitoring host use a message interaction method with acknowledgment receipts. The maximum single write latency is 500 milliseconds, and the number of concurrent identifications can be set to 32 branches per terminal. The number of retries per record is 3, with a retry interval of 200 milliseconds. If the same device unique identifier, branch unique identifier, event type, and reference time point are repeatedly sent and the time deviation does not exceed 20 milliseconds, it is considered the same record, and only the communication status is updated. In the event of power interruption, clock loss, record source disconnection, or manual forced operation without leaving a trace, the original fragment is frozen. After freezing, it is prohibited to complete, merge, or overwrite. It is allowed to add reason codes and manual review conclusions. The reason codes are fixed as five categories: timestamp mismatch, mutual exclusion state conflict, source missing, manual locking, and write failure. The minimum set of interfaces includes device unique identifier, branch unique identifier, event type, reference time point, front boundary time, back boundary time, record source version number, rule version number, and reason code. If the device unique identifier is missing, it returns "device identifier missing"; if the rule version number is missing, it returns "version missing"; if the front and back boundary times are reversed, it returns "time boundary invalid".

[0032] Preferably, in the distribution branch of the molded case circuit breaker with a rated current of 250 amps, the sampling rhythm is set to 20 milliseconds, the timestamp alignment tolerance is set to 5 milliseconds, the state holding threshold is set to 100 milliseconds, the pre-observation segment is set to 500 milliseconds, and the post-observation segment is set to 1200 milliseconds. In the power restoration event, for the load restoration scenario, the branch voltage is required to recover to more than 90% of the rated voltage and be maintained continuously for 800 milliseconds, while the branch current reappears and continues for 3 sampling cycles; for the no-load standby scenario, the branch voltage is required to recover to more than 90% of the rated voltage and be maintained continuously for 800 milliseconds, the auxiliary contacts remain closed, and the branch current is no longer required to reappear. A total of 146 tripping events, 142 closing events, 11 tripping events, and 11 power restoration events were obtained through continuous on-site recording for 30 days. All event measurement windows were formed and called by the next stage. The event type consistency rate of 50 manually sampled segments was 100%, and the maximum boundary deviation was 40 milliseconds. Alternatively, when the mechanical position status record cannot be obtained, the auxiliary contact status record can be used as the main judgment source and the branch voltage existence record as the auxiliary verification source to complete the identification of opening and closing events. The identification order, boundary locking method, version locking method, deduplication method, and trace retention method of other events remain unchanged.

[0033] S2. Collect operating current, inter-terminal voltage drop, near-contact temperature rise response, opening and closing action sequence, and event type identifier within the event measurement window to form an event measurement dataset. The specific implementation is as follows: Within the event measurement window that has been established and its boundaries locked in the previous step, the operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, and event type identifier are collected sequentially, and an event measurement dataset is formed accordingly. This action occurs at the operating site of the molded case circuit breaker in the low-voltage distribution cabinet, power distribution box, charging pile distribution unit, and industrial production line terminal distribution branch. It is completed collaboratively by the field acquisition device, the cabinet control terminal, and the measurement circuit that is fixed to the circuit breaker. The execution entity can be set as the measurement recording program in the cabinet control terminal. The acquisition interval is fixed and constrained by the front boundary time and the back boundary time of the event measurement window. Records corresponding to the front boundary time and the back boundary time are included in this window. Records outside the boundary cannot be included in this window. The period of action covers the entire event measurement window corresponding to the opening event, closing event, tripping event, and power restoration event. Upstream, it receives the confirmed event type identifier, reference time point, front boundary time, and back boundary time. Downstream, it provides the event measurement dataset without rewriting the original boundary to the comparable event feature set generation stage.

[0034] The operating current is the actual current record of the circuit breaker load-side conductor within the event measurement window. In three-phase branches, it is collected and recorded separately for each phase. In single-phase branches, it is recorded through a single acquisition channel for each branch. The current is taken from the current transformer or Hall current sensor installed on the corresponding phase conductor on the outgoing side of the molded case circuit breaker. After being converted to the primary side value by the field acquisition device, it is written into the data. The unit is amperes. The event measurement dataset retains the phase-by-phase operating current record corresponding to each sampling moment. The average value within the window is not used as a substitute. The sampling rhythm can be set from 10 milliseconds to 100 milliseconds, preferably 20 milliseconds. The measurement tolerance can be set to no more than 1% of the rated range. The operating current is continuously collected from the start to the end of the event measurement window without splicing across windows. If there is a continuous lack of sampling for more than 200 milliseconds, this item will stop being processed and will be transferred to the pending record.

[0035] Inter-terminal voltage drop is a record of the potential difference between the incoming connection point and the corresponding outgoing connection point of the molded case circuit breaker body on the same phase, excluding the upstream busbar extension and the downstream cable extension. In three-phase branches, it is obtained after being bound one-to-one according to the phase sequence. In single-phase branches, it is obtained between the incoming and outgoing terminals. The binding relationship is locked in the record source version. The unit is millivolts. The sampling rhythm is consistent with the operating current. The timestamp uses the local unified clock source. The inter-terminal voltage drop is fixed by subtracting the corresponding outgoing terminal sampling value from the sampling value of the incoming terminal of the same phase and retaining the sign. It cannot be switched to other sampling directions within the same continuous operating cycle. Cross-cycle subtraction is not used, and multi-phase averaging is not used. If any sampling terminal is missing, phase is mismatched, or sampling channel is disconnected at the same sampling time, the inter-terminal voltage drop at that sampling time will not be formed.

[0036] The near-contact temperature rise response is a record of the temperature increase of the corresponding phase housing near-contact temperature measurement point relative to the average temperature of the baseline interval before the event at each sampling time within the event measurement window. In the three-phase branch, corresponding temperature measurement points are set and recorded separately for each phase. In the single-phase branch, a single corresponding temperature measurement point is set. The temperature measurement point is fixedly arranged at a preset position on the outer surface of the housing within the shortest straight-line distance range from the outer side of the corresponding phase contact cavity. The position coordinates, installation direction, and measurement point number are locked in the recording source version and cannot be replaced by the highest temperature point of the whole machine. The unit is degrees Celsius. The sampling rhythm can be set from 100 milliseconds to 1000 milliseconds, preferably 200 milliseconds. The measurement tolerance can be set to no more than 1 degree Celsius. The near-contact temperature rise response is formed point by point according to the sampling time. The average temperature of the baseline interval before the event is formed by all valid original temperature sampling points of the same temperature measurement point within the baseline interval before the event. The supplementary points do not participate in the formation of the average value. If there are fewer than 5 consecutive valid original temperature samples, the formation of the near-contact temperature rise response of this window is stopped and the process is put into the pending verification stage.

[0037] The timing sequence of opening and closing actions is a record of the sequence of actions formed around a reference time point within the event measurement window, based on the mechanical position status record and the auxiliary contact status record. It is only generated within the event measurement window corresponding to the opening, closing, and tripping events, and is not generated separately within the event measurement window corresponding to the power restoration event. The unit is milliseconds, and it includes at least the mechanical position status flip time, the auxiliary contact status flip time, and the time difference between the two. When there are multiple flips within the same event measurement window, the first valid flip time that is closest to the reference time point and finally enters a stable state is taken. The backflip segment is not used as the landing point of the action timing. The time difference is fixed and recorded as the auxiliary contact status flip time minus the mechanical position status flip time. When the opening and closing action timing cannot be formed, it does not affect the inclusion of the operating current, inter-terminal voltage drop, near-contact area temperature rise response, and event type identifier within the same event measurement window. However, the opening and closing action timing field should be set to null and a reason code should be added. Automatic recalculation is not allowed.

[0038] The event type identifier is inherited from the confirmed result of the previous step and will not be re-determined in this step. It is not allowed to be modified in reverse by the collected results. If the event type identifier of the previous step is missing, this step will stop forming the event measurement dataset and will proceed to the pending verification stage.

[0039] When processing records, the process begins by first extracting records within the event measurement window based on its boundaries. Then, these records are aligned using a unified timestamp. Next, distorted segments are removed, including those with values ​​exceeding the upper limit of the measurement range, those remaining continuously unchanged for longer than the set duration, those with adjacent sampling points exceeding the upper limit specified by the record source version, and those caused by phase misalignment. Forward padding is then performed on allowed padding fields. Finally, records are grouped into the same set within the window based on device unique identifier, branch unique identifier, event type identifier, reference time point, sampling time, and phase. All sampled records within the same event measurement window constitute a single event measurement dataset. Each sampling time and each... Each phase corresponds to one record. Forward completion is only allowed for continuous numerical sampling records on which the operating current, inter-terminal voltage drop, and near-contact temperature rise response depend. Completion is only allowed if the continuous missing sampling time does not exceed 200 milliseconds. After completion, new time points must not be formed across windows. Mechanical position status records, auxiliary contact status records, and event type identifiers must not be automatically completed. If any of the following seven situations occur, such as reversed timestamp, missing event measurement window boundary, missing rule version, missing record source version, missing sampling end mapping, inconsistent temperature measurement point version, or multiple conflicting values ​​at the same sampling time, the data processing will be stopped and the data will be transferred to pending verification or pending supplementation.

[0040] After data acquisition, an event measurement dataset is formed by the smallest set of fields: unique device identifier, unique branch identifier, event type identifier, reference time point, sampling time, phase, operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, record source version number, rule version number, and cause code. This dataset is stored in the event data recording area of ​​the field control terminal in a structured text format and simultaneously written to the timing recording area of ​​the power distribution monitoring host. The next step only calls the data based on the unique device identifier, unique branch identifier, event type identifier, and reference time point, without overwriting the records already archived in this step. The field control terminal and the power distribution monitoring host use a message interaction method with acknowledgment receipts. The maximum delay for a single data entry can be set to 500 milliseconds, the number of concurrent data acquisitions can be set to 32 branches per terminal, the number of retries can be set to 3, and the retry interval can be set to 200 milliseconds. Old version records must not overwrite newly archived records. If the same unique device identifier, unique branch identifier, event type identifier, and reference time point are repeatedly uploaded with a time deviation of no more than 20 milliseconds, it is considered the same record, and only the communication status is updated.

[0041] When sensor disconnection, sampling open circuit, temperature measurement point detachment, or insufficient samples in the event measurement window occur, the original fragments already obtained should be retained. The fabrication of missing fields is prohibited. Reason codes for missing source, measurement point failure, boundary violation, and value violation should be added and marked for manual review. The minimum interface set should include at least the device unique identifier, branch unique identifier, event type identifier, reference time point, sampling time, phase, operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, record source version number, rule version number, and reason code. If the device unique identifier is missing, return "device identifier missing"; if the event type identifier is missing, return "event type missing"; if the sampling time exceeds the event measurement window boundary, return "boundary violation".

[0042] During on-site inspection, the following parameters are used as references: operating current corresponding to a portable clamp meter, inter-terminal voltage drop corresponding to a millivoltmeter, near-contact temperature corresponding to a thermal imager, and opening / closing action sequence corresponding to an action record. The deviations in operating current, inter-terminal voltage drop, near-contact temperature rise response, opening / closing action sequence, and event type identification consistency rate are checked in no fewer than 300 event measurement window samples. Specifically, the operating current deviation can be set to no more than 2% of the rated range, the inter-terminal voltage drop deviation can be set to no more than 5 millivolts, the near-contact temperature rise response deviation can be set to no more than 1.5 degrees Celsius, the opening / closing action sequence deviation can be set to no more than 20 milliseconds, and the event type identification consistency rate can be set to no less than 99%.

[0043] Preferably, in a three-phase distribution branch of a molded case circuit breaker with a rated current of 250 amps, the event measurement window length is 1700 milliseconds, the sampling rhythm for operating current and inter-terminal voltage drop is 20 milliseconds, and the sampling rhythm for near-contact temperature is 200 milliseconds. In a certain closing event measurement window, the operating current of phase A is 186 amps, the operating current of phase B is 181 amps, and the operating current of phase C is 184 amps. The inter-terminal voltage drop of phase A is 42 millivolts, the inter-terminal voltage drop of phase B is 39 millivolts, and the inter-terminal voltage drop of phase C is 41 millivolts. The temperature rise response of the near-contact area of ​​phase A is 2.8 degrees Celsius. The auxiliary contact state reversal time minus the mechanical position state reversal time is 14 milliseconds. The event type is identified as a closing event. All the sampled records under the same event measurement window constitute a complete event measurement dataset and are called by the next step.

[0044] Alternatively, the AC branch uses a current transformer to obtain the operating current, and the branch containing the DC component uses a Hall current sensor to obtain the operating current. When the temperature rise response in the near-contact area cannot be obtained through the fixed temperature measurement point on the casing, infrared temperature measurement records with a fixed field of view, a fixed corresponding area, and recorded separately by phase can be used. These records are still formed according to the same event measurement window boundary and the same front baseline interval. The acquisition order, version locking method, disk loading method, and deduplication method of the remaining fields remain unchanged.

[0045] S3. Perform operating condition normalization processing on the event measurement dataset based on load status, ambient temperature, and event type to generate a comparable event feature set. The specific implementation is as follows: First, load status, ambient temperature, and event type are extracted sequentially from the event measurement dataset archived in the previous step. Then, based on load status, ambient temperature, and event type, operating condition normalization processing is performed on the operating current, inter-terminal voltage drop, near-contact temperature rise response, and opening / closing action sequence within the same event measurement window to form a comparable event feature set. Among them, the operating condition normalization processing refers to unifying the event measurement data obtained under different load statuses and different ambient temperatures within the same model of molded case circuit breaker, the same branch type, and the same event type to the same comparison boundary. After normalization, four types of fields are formed: normalized operating current, normalized inter-terminal voltage drop, normalized near-contact temperature rise response, and normalized opening / closing action sequence. These fields are not used for direct comparison across models, frame levels, or wiring methods.

[0046] This action occurs at the operating site of molded case circuit breakers in low-voltage distribution cabinets, power distribution boxes, charging pile power distribution units, and industrial production line end power distribution branches. It is completed by the unified recording program in the control terminal inside the cabinet. The action period is after the event measurement dataset is archived and before the comparable event feature set is written. Upstream, it receives the unique identifier of the equipment, the unique identifier of the branch, the event type identifier, the reference time point, and the corresponding event measurement dataset. Downstream, it provides the subsequent links with a comparable event feature set that does not rewrite the boundary of the original event measurement dataset.

[0047] The load status is a record of the load level of the corresponding branch of the molded case circuit breaker within the event measurement window. The rated current value is preferentially taken from the equipment ledger lock value corresponding to the rated current on the circuit breaker nameplate. If the equipment ledger is missing, the preset value of the rule version is taken. It cannot be switched within the same continuous operating cycle. In the three-phase branch, the rated current ratio is recorded separately for each phase, and the state corresponding to the largest rated current ratio among the three phases is used as the unified load status of the event measurement window. In the single-phase branch, it is formed by the unique acquisition channel of the single branch. The rated current ratio is in percentage form. It is obtained by comparing all valid original operating current sampling points between the end time of the baseline interval before the event and the reference time point with the rated current value point by point. It does not use external records to supplement or use manual estimation to replace. The light load state can be set to a rated current ratio of no more than 30%, the medium load state can be set to a ratio of more than 30% but no more than 70%, and the heavy load state can be set to a ratio of more than 70%. The boundary cannot be switched within the same rule version. If there are fewer than 5 consecutive sampling points of valid operating current in the event measurement window, the load status of the window is not formed and it is transferred to pending verification.

[0048] Ambient temperature is the recorded air temperature around the installation location of the molded case circuit breaker, taken from the ambient temperature measurement point inside the distribution cabinet or the common temperature measurement point of the circuit breaker installation cavity. The ambient temperature measurement point and the near-contact area temperature measurement point must not be shared. The unit is degrees Celsius. The sampling rhythm is consistent with or higher than the temperature rise response of the near-contact area. Within the same event measurement window, the ambient temperature record is formed by all valid original ambient temperature sampling points from the start point of the baseline interval before the event to the end point of the response interval after the event. The ambient temperature interval can be marked as low temperature interval, normal temperature interval, and high temperature interval. The temperature range can be set as follows: low temperature range, no higher than 15 degrees Celsius; normal temperature range, higher than 15 degrees Celsius but no higher than 35 degrees Celsius; high temperature range, higher than 35 degrees Celsius. In actual field conditions, for example, when multiple molded case circuit breakers are installed side by side in the same cabinet, the common temperature measuring point is only allowed to be shared by branches within the same installation cavity and within a set distance from the outer edge of the target molded case circuit breaker. It cannot be used across cavities or by using outdoor weather temperature. The measuring point number, installation direction, and record source version are locked under the same rule version.

[0049] The event type is fixed and inherited from the event type identifier of the previous step, including four types: tripping event, closing event, tripping event, and power restoration event. It is not re-determined in this step, and it is not allowed to be modified in reverse from the normalized result.

[0050] Before each record enters the normalization process, the event measurement dataset is first merged within the same window according to the unique equipment identifier, branch unique identifier, event type identifier, and reference time point. Then, alignment is completed according to sampling time and phase. Next, distorted segments caused by values ​​exceeding the upper limit of the range, continuous fixed values ​​exceeding the set duration, changes in adjacent sampling points exceeding the upper limit specified by the record source version, and phase misalignment are removed. Then, forward completion is performed on the fields that are allowed to be completed. Finally, the valid records within the same window required for normalization are formed. Among them, forward completion is only allowed for the original sampling records of operating current, ambient temperature, and near-contact area temperature, and the continuous missing sampling does not exceed 200 milliseconds. Automatic completion is only possible if the time interval is reached within seconds. Inter-terminal voltage drop, opening and closing action sequence, event type identifier, rated current value, and rule version number cannot be automatically completed. If any of the following six situations occur: missing event type identifier, missing rated current value, missing environmental measurement point number, multiple conflicting values ​​at the same sampling time, missing rule version, or missing record source version, the process will stop and the record will be transferred to pending verification or supplementary recording. When multiple records correspond to the same unique equipment identifier, branch unique identifier, event type identifier, reference time point, sampling time, or phase combination key, the earliest archived record with a complete version will be retained first. Other records will be marked with a duplicate tag and will not participate in the normalization process.

[0051] Operating condition normalization is performed in a fixed order: first, it is grouped and normalized according to load status; then, it is partitioned and normalized according to ambient temperature; finally, the corresponding field caliber is retained according to event type. Specifically, grouping and normalization means grouping event measurement data of the same event type according to light load, medium load, and heavy load conditions, and defining comparison boundaries. Secondary numerical conversions are no longer performed on inter-terminal voltage drop and near-contact temperature rise response. The normalized operating current is fixedly recorded as a percentage of the rated current, and the original ampere value is no longer retained as a normalized field. The normalized inter-terminal voltage drop is fixedly recorded as the original inter-terminal voltage drop with uniform load status boundaries, in millivolts, and is fixedly formed by the inter-terminal voltage drop. The "Ambient Temperature Zoning Unified Fixed" refers to further defining the comparison boundaries of records under the same load condition according to low temperature range, normal temperature range, and high temperature range, without rewriting the temperature rise response values ​​near the contact area. The unified near-contact area temperature rise response is taken from the original near-contact area temperature rise response record with the attached ambient temperature range identifier and unified load condition boundary, in degrees Celsius, and is fixedly formed by the near-contact area temperature rise response. The "Unified Opening and Closing Action Sequence" is taken from the original opening and closing action sequence record under the corresponding event type, retaining the original time difference direction and time difference unit, in milliseconds, and is fixedly formed by the opening and closing action sequence. In the power restoration event, this field is set to null and a reason code is appended.

[0052] The comparable event feature set generated after the operating condition normalization process includes at least the following: unique equipment identifier, unique branch identifier, branch type, event type identifier, reference time point, phase, load status, ambient temperature range identifier, normalized operating current, normalized inter-terminal voltage drop, normalized near-contact temperature rise response, normalized opening and closing action sequence, record source version number, rule version number, and cause code. Among these, the branch type is fixed to one of the following: motor branch, charging branch, lighting branch, heating branch, or standby branch. Once locked in the equipment ledger, it cannot be switched within the same continuous operating cycle. The comparable event feature set is stored in the feature record area of ​​the field control terminal in a structured text format and is simultaneously written to the feature sequence area of ​​the power distribution monitoring host. The next step only calls the records based on the unique equipment identifier, unique branch identifier, event type identifier, and reference time point, without overwriting the records already archived in this step.

[0053] The maximum delay for a single unified disk write can be set to 500 milliseconds, the number of concurrent unified writes can be set to 32 branches per terminal, the number of retries can be set to 3, the retry interval can be set to 200 milliseconds, old version records must not overwrite new archived records, and if the same device unique identifier, branch unique identifier, event type identifier, and reference time point combination key are repeatedly uploaded and the time deviation does not exceed 20 milliseconds, they are considered to be the same comparable event feature set, and only the communication status is updated.

[0054] When environmental measurement point failure, missing rated current parameter, missing event type identifier, insufficient valid samples within the window, or version conflict occurs, the original event measurement dataset already obtained should be retained. Reverse rewriting of the original record is prohibited. The source missing, parameter missing, insufficient sample, and version conflict reason codes should be added and written to the manual review mark. The minimum set of interfaces should include at least the device unique identifier, branch unique identifier, branch type, event type identifier, reference time point, phase, load status, ambient temperature range identifier, normalized operating current, normalized inter-terminal voltage drop, normalized near-contact temperature rise response, normalized opening and closing action sequence, record source version number, rule version number, and reason code. If the device unique identifier is missing, return device identifier missing; if the event type identifier is missing, return event type missing; if the rated current value is missing, return parameter missing.

[0055] During on-site inspection, a sample of no less than 300 event measurement windows is used as the basis for random inspection. The consistency of records before and after normalization, the comparability across working conditions after normalization, and the version consistency are checked by grouping the same model, the same branch type, and the same event type. The consistency can be set to no less than 99%, and the dispersion convergence amplitude of the same group after normalization can be set to no less than 20%. The dispersion convergence amplitude is fixed according to the reduction ratio of the difference between the maximum and minimum values ​​before normalization to the difference between the maximum and minimum values ​​after normalization in the same group of records.

[0056] Preferably, in the motor branch of a three-phase molded case circuit breaker with a rated current of 250 amps, the operating currents of phases A, B, and C in the measurement window of a certain closing event are 186 amps, 181 amps, and 184 amps, respectively. According to the proportion of the maximum rated current, it is classified as a heavy load state. The ambient temperature is 38 degrees Celsius and it is classified as a high temperature range. The event type is identified as a closing event. After the operating condition is normalized, the normalized operating currents recorded for phases A, B, and C, the normalized inter-terminal voltage drop with heavy load state boundary, the normalized near-contact temperature rise response with heavy load state boundary and high temperature range identification, and the normalized opening and closing action sequence are formed. Together, they constitute a comparable event feature set and are called by the next step.

[0057] Alternatively, when the ambient temperature cannot be obtained through a fixed temperature measurement point inside the cabinet, a common temperature measurement point record within the same installation cavity can be used to form an ambient temperature range identifier, but the measurement point location, number, version locking method, event type inheritance method, archiving method, and deduplication method remain unchanged.

[0058] S4. Construct an interference elimination matrix based on the comparable event feature set to filter out event features corresponding to terminal connection abnormalities, heat dissipation limitation abnormalities, and mechanism hysteresis abnormalities. The specific implementation is as follows: First, retrieve the comparable event feature set archived in the previous step, and then merge them into the same window according to the equipment unique identifier, branch unique identifier, event type identifier, reference time point, phase, normalized operating current, normalized inter-terminal voltage drop, normalized near-contact temperature rise response, normalized opening and closing action sequence, load status, and ambient temperature range identifier under the same event measurement window. Then, construct an interference elimination matrix based on this, and filter out event features corresponding to terminal connection abnormalities, heat dissipation limitation abnormalities, and mechanism hysteresis abnormalities within the same event measurement window. The interference elimination matrix is ​​fixed to a single correspondence record formed at the granularity of a single event measurement window. This record forms sub-records by phase and includes at least terminal connection abnormality markers, heat dissipation limitation abnormality markers, mechanism hysteresis abnormality markers, exclusion field identifiers, and retained event feature identifiers. It does not rewrite the original comparable event feature set, but only provides logical exclusion basis for downstream calls.

[0059] This action occurs at the operating site of molded case circuit breakers in low-voltage distribution cabinets, power distribution boxes, charging pile power distribution units, and industrial production line end power distribution branches. It is executed by the exclusion record program in the control terminal inside the cabinet. The action period is after the comparable event feature set is archived and before the target event feature enters the next stage. The upstream receives the locked version of the comparable event feature set, and the downstream only provides the retained event features after screening.

[0060] Terminal connection abnormality refers to an abnormal connection status at the incoming and outgoing connection points of the molded case circuit breaker body, excluding the upstream busbar remote connection point and the downstream cable remote connection point. In this step, it is identified by phase, without further subdivision by terminal. The determination is based on the normalized inter-terminal voltage drop, normalized operating current, and normalized near-contact temperature rise response corresponding to the phase within the same event measurement window. This is achieved by first comparing whether the normalized inter-terminal voltage drop of the phase is continuously higher than the upper boundary of the reference interval under the same model, branch type, event type, and load state boundary; then comparing whether the normalized near-contact temperature rise response of the phase has not reached the corresponding synchronous offset threshold during the same event measurement window, the same phase, and the period of continuous state holding threshold; and finally, combining this with whether the normalized opening and closing action sequence is within the allowable range. Heat dissipation limitation abnormality refers to a common issue within the same installation cavity. Temperature rise retention caused by thermal environment limitation is not identified as an independent electrothermal anomaly of a single branch. The determination is based on the temperature rise response of the target branch and adjacent branches within the same installation cavity, the ambient temperature range marker, and the unified inter-terminal voltage drop during the same event measurement window coverage period. First, compare whether the temperature rise response of the unified near-contact area of ​​no less than two adjacent branches increases in the same direction and continues to reach the state maintenance threshold. Then, compare whether the ambient temperature range marker is in the high temperature range. Finally, compare whether the unified inter-terminal voltage drop of the target branch has not reached the single branch concentrated offset threshold. Among them, the adjacent branch is fixed to the branch within the same installation cavity that shares the ambient temperature measurement boundary with the target branch and is marked as an adjacent installation position in the equipment ledger. The fixed installation cavity marker is taken from the cabinet configuration file or the locked value of the equipment ledger and cannot be switched within the same continuous operating cycle.

[0061] The mechanical hysteresis anomaly refers to the delay in the timing of the mechanical action chain. It is only used to identify opening, closing, and tripping events, and is not used in power restoration events. The determination is based on the unified opening and closing action sequence, unified inter-terminal voltage drop, and unified near-contact temperature rise response. It is formed by first comparing whether the unified opening and closing action sequence continuously exceeds the allowable range under the boundary of the same model, the same branch type, and the same event type, and then comparing whether the unified inter-terminal voltage drop and the unified near-contact temperature rise response have not reached the corresponding offset threshold within the same event measurement window.

[0062] The upper boundary of the reference interval, the allowable interval, the synchronization offset threshold, and the single-branch concentrated offset threshold are collectively referred to as the anomaly judgment boundary. All of the above anomaly judgment boundaries are preferentially taken from the preset values ​​of the rule version and locked separately according to the same model, branch type, and event type. During the first round of commissioning, the above preset values ​​of the rule version are formed from no less than 30 health event measurement window samples under the same model, branch type, and event type, and cannot be rewritten within the same continuous operating cycle after the version is locked. When the preset value of the rule version is missing, the determination of the corresponding interference source in this step is stopped and the process proceeds to pending verification.

[0063] Unsynchronized fixation refers to the situation where, within the same event measurement window and on the same phase, the target comparison field does not reach the corresponding offset threshold while continuously reaching the state maintenance threshold.

[0064] Before each record is constructed, it is first aligned according to the unique device identifier, branch identifier, event type identifier, reference time point, and phase. Then, distorted segments caused by inconsistent versions, multiple conflicting values ​​at the same sampling time, duplicate archiving in the same window, and missing phase are removed. Then, forward completion is performed on the fields that can be completed. Finally, valid records in the same window that can be used to construct the interference elimination matrix are formed. Among them, forward completion is only allowed for the original ambient temperature sampling records that depend on the ambient temperature range identifier and the original temperature sampling records that depend on the normalized near-contact area temperature rise response, when the continuous missing sampling does not exceed 200 milliseconds. Normalized inter-terminal voltage drop, normalized operating current, normalized opening and closing action sequence, event type identifier, load status, and rule version number shall not be automatically completed. When any of the following six situations occur, such as missing rule version, missing record source version, missing adjacent branch identifier, missing identifier of the same installation cavity, missing reference time point, or missing event type identifier, the further processing is stopped and the record is transferred to pending verification or pending supplementation.

[0065] When filtering out event features corresponding to terminal connection abnormalities, heat dissipation limitation abnormalities, and mechanism hysteresis abnormalities, the fixed meaning is that the corresponding field is marked as an exclusion field in the filtering record and is not included in the retained event features when called downstream, and the original comparable event feature set is not physically deleted; among them, terminal connection abnormalities only filter out the normalized inter-terminal voltage drop field corresponding to the abnormal phase, heat dissipation limitation abnormalities only filter out the normalized near-contact area temperature rise response field affected by the common thermal environment, and mechanism hysteresis abnormalities only filter out the normalized opening and closing action timing field. The fixed meaning of window-related event features is the same set of fields directly corresponding to the source of the abnormality, and is not extended to other untriggered fields.

[0066] If two or more types of interference sources are triggered simultaneously within the same event measurement window, they will be marked and eliminated in a fixed order: terminal connection abnormality, heat dissipation limitation abnormality, and mechanism hysteresis abnormality. The judgment of the latter type is still based on the original comparable event feature set. Fields that have been marked and eliminated in the former type will not be marked again. The reason code is fixed to record the first triggered interference source.

[0067] Upon completion, a screening record is generated. The screening record includes at least the unique device identifier, branch unique identifier, event type identifier, reference time point, phase, interference elimination matrix identifier, terminal connection anomaly marker, heat dissipation limitation anomaly marker, mechanism hysteresis anomaly marker, exclusion field identifier, retained event feature identifier, record source version number, rule version number, and reason code. It is stored in the exclusion record area of ​​the field control terminal in a structured text format and simultaneously written to the exclusion timing area of ​​the power distribution monitoring host. The next step only calls the retained event features according to the unique device identifier, branch unique identifier, event type identifier, and reference time point. The maximum delay for a single disk entry can be set to 500 milliseconds, the number of concurrent entries can be set to 32 branches per terminal, the number of retries can be set to 3, and the retry interval can be set to 200 milliseconds. Old version records must not overwrite new archived records. If the same unique device identifier, branch unique identifier, event type identifier, and reference time point are repeatedly uploaded and the time deviation does not exceed 20 milliseconds, it is considered as the same screening record, and only the communication status is updated.

[0068] When environmental measurement points fail, adjacent branch records are missing, installation cavity identifiers are missing, the timing sequence of unified opening and closing actions is empty and the event type does not match, or there are insufficient valid samples in the same window, the original comparable event feature set is retained, reverse rewriting of the original record is prohibited, and reason codes for missing source, incomplete boundary, insufficient sample, and version conflict are added and written to the manual review mark; the minimum set of interfaces must include at least the unique device identifier, unique branch identifier, event type identifier, reference time point, phase, interference exclusion matrix identifier, terminal connection abnormality mark, heat dissipation limitation abnormality mark, mechanism hysteresis abnormality mark, exclusion field identifier, retained event feature identifier, record source version number, rule version number, and reason code. If the unique device identifier is missing, return "device identifier missing"; if the event type identifier is missing, return "event type missing"; if the installation cavity identifier is missing, return "boundary incomplete".

[0069] During on-site inspection, a sample of no less than 300 event measurement windows is used as the basis for random inspection. The consistency rate, exclusion field matching rate, and retained event feature completeness rate of the interference exclusion matrix are verified by grouping according to the same model, the same branch type, and the same event type. The consistency rate can be set to no less than 99%, the exclusion field matching rate can be set to no less than 98%, and the retained event feature completeness rate can be set to no less than 99%.

[0070] Preferably, in the motor branch of a three-phase molded case circuit breaker with a rated current of 250 amps, in the comparable event feature set corresponding to a certain closing event, the voltage drop between the unified terminals of phase A is continuously higher than the upper boundary of the corresponding reference interval for 800 milliseconds, the temperature rise response of the unified near-contact area of ​​phase A does not reach the synchronous offset threshold within the same window, phases B and C do not show the same direction offset, and the timing of the unified opening and closing actions is within the allowable range. The interference elimination matrix constructed in this way marks the terminal connection abnormality as valid, the heat dissipation limitation abnormality and the mechanism hysteresis abnormality as invalid, and only the voltage drop between the unified terminals of phase A is screened out, forming one screening record, which is called by the next step.

[0071] Alternatively, when adjacent branch temperature rise records cannot be obtained, the heat dissipation limitation anomaly can be changed to the formation of records from the common environment temperature measurement point of the same installation cavity and the target branch for no less than 3 consecutive event measurement windows; when the common environment temperature measurement point is continuously in the high temperature range during the corresponding time period of the consecutive event measurement window, and the temperature rise response of the target branch near the contact area continuously increases in the same direction and continues to reach the state maintenance threshold, and at the same time the voltage drop between the unified terminals of the target branch does not reach the single branch concentrated offset threshold, the heat dissipation limitation anomaly is determined to be established; the installation cavity boundary, version locking method, screening order, archiving method, and deduplication method remain unchanged.

[0072] S5. Extract conductivity degradation features, thermal response hysteresis features, and post-action recovery features from the event features after filtering out interference, and combine them to form an ablation evidence cluster. The specific implementation is as follows: First, from the retained event features after filtering out interference in the previous step, retrieve the unified inter-terminal voltage drop, unified near-contact temperature rise response, unified opening and closing action sequence, unified operating current, load status, and ambient temperature range identifier under the same event measurement window according to equipment unique identifier, branch unique identifier, event type identifier, reference time point, and phase. Then, extract the conductivity degradation feature, thermal response hysteresis feature, and post-action recovery feature in sequence, and combine the three to form an ablation evidence cluster. Among them, the conductivity degradation feature is fixed to the feature record composed of phase identifier, unified inter-terminal voltage drop offset record, unified operating current boundary record, and duration record, which is applicable to closing events, tripping events, and power restoration events. The fixed thermal response hysteresis characteristic refers to the characteristic record consisting of the start time of temperature rise, the time of reaching the stable offset interval, the time of falling back to the baseline adjacent interval, and the corresponding duration. It is applicable to phases that retain the normalized near-contact temperature rise response in closing events, tripping events, and power restoration events. The fixed post-operation recovery characteristic refers to the characteristic record consisting of the time of operation completion, the time of normalized inter-terminal voltage drop entering the stable interval, the time of normalized near-contact temperature rise response entering the stable interval, and the recovery completion time. It is only applicable to opening events, closing events, and tripping events, and is not used for power restoration events.

[0073] This action occurs at the operating site of molded case circuit breakers in low-voltage distribution cabinets, power distribution boxes, charging pile distribution units, and industrial production line end distribution branches. It is executed by the feature extraction program in the control terminal inside the cabinet. The action period is after the screening record is archived and before the ablation evidence cluster is written. The upstream receives the retained event features, and the downstream provides the ablation evidence cluster. It does not rewrite the screening record of the previous step.

[0074] Conductivity degradation features are derived from the unfiltered normalized inter-terminal voltage drop and normalized operating current, and are compared between phases within the same load state and event type boundary. Specifically, the normalized inter-terminal voltage drop offset record refers to the offset of the phase's normalized inter-terminal voltage drop relative to the upper boundary of the reference interval under the same model, branch type, event type, and load state boundary. The normalized operating current boundary record refers to whether the phase's normalized operating current remains within the current load state boundary. The duration record refers to the duration for which the normalized inter-terminal voltage drop offset continuously reaches the state maintenance threshold. Conductivity degradation features are extracted when the normalized inter-terminal voltage drop is continuously higher than the upper boundary of the reference interval and the normalized operating current does not fall outside the current load state boundary.

[0075] The thermal response hysteresis feature is derived from the normalized near-contact temperature rise response that was not screened out, the ambient temperature range markers, and continuous temperature records before and after the reference time point. It is formed by first aligning the sampling times, and then identifying the temperature rise start time, the time when the temperature rise reaches the stable offset range, and the time when the temperature rise falls back to the baseline neighbor range. The temperature rise start time is fixed as the first sampling time when the normalized near-contact temperature rise response is continuously higher than the starting boundary corresponding to the average temperature of the previous baseline range and continues to reach the state maintenance threshold. The stable offset range is fixed as the range when the normalized near-contact temperature rise response continuously falls into the stable offset boundary locked by the rule version and continues to reach the state maintenance threshold. The baseline neighbor range is fixed as the range when the normalized near-contact temperature rise response falls back to the boundary of the average temperature of the previous baseline range and continues to reach the state maintenance threshold. The time when the temperature rise reaches the stable offset range and the time when the temperature rise falls back to the baseline neighbor range are both recorded as the first sampling time that meets the continuity condition in the corresponding range.

[0076] The post-action recovery features are derived from the unfiltered normalized opening and closing action sequence, normalized inter-terminal voltage drop, and normalized near-contact temperature rise response. This is achieved by first locating the action completion time, and then comparing the time corresponding to when the normalized inter-terminal voltage drop and normalized near-contact temperature rise response enter the stable interval after the action is completed. Specifically, the action completion time is fixed as the time when the auxiliary contact state flips to complete, corresponding to the normalized opening and closing action sequence. The stable interval is fixed as the interval where the normalized inter-terminal voltage drop or normalized near-contact temperature rise response enters the stable comparison boundary locked by the rule version and continues to reach the state maintenance threshold. The recovery completion time is fixed as the time between the action completion time and the later time when both the normalized inter-terminal voltage drop and normalized near-contact temperature rise response enter the stable interval, and no longer forms a separate stability field.

[0077] The initial boundary, stable offset boundary, adjacent boundary, and stable comparison boundary are all formation thresholds of the three types of features. They are preferentially taken from the rule version lock value and locked separately according to the same model, the same branch type, and the same event type. They cannot be switched within the same continuous operation cycle.

[0078] Before each record is extracted, it is first aligned according to the unique identifier of the equipment, the unique identifier of the branch, the event type identifier, the reference time point, and the phase. Then, it removes distorted segments caused by inconsistent versions, multiple conflicting values ​​at the same sampling time, duplicate archiving in the same window, and fields that have been filtered out. Then, it performs forward completion on fields that can be completed, and finally forms a valid record for feature extraction. Among them, forward completion is only allowed for the original temperature sampling record that depends on the temperature rise response of the unified near contact area, if the continuous missing sampling does not exceed 200 milliseconds. The unified inter-terminal voltage drop, unified opening and closing action sequence, unified operating current, event type identifier, load status, and ambient temperature range identifier must not be automatically completed. If any of the following five situations occur, such as missing rule version, missing record source version, missing reference time point, missing phase, or insufficient valid samples in the same window, the processing will stop and the record will be transferred to pending verification or pending supplementation.

[0079] When forming an ablation evidence cluster, the fixed granularity is a single event measurement window and a single phase. The successfully extracted conductivity degradation features, thermal response hysteresis features, and post-action recovery features under the same phase are grouped into one associated record according to the equipment unique identifier, branch unique identifier, event type identifier, reference time point, and phase. The minimum condition for the establishment of an ablation evidence cluster is that at least two of the three types of features are successfully extracted. If there are fewer than two types, an ablation evidence cluster is not formed, and only the reason for the missing feature is recorded. If any of the three types of features under the same phase is missing, the remaining extracted features are retained and the reason for the missing feature is marked in the ablation evidence cluster. It is not automatically reconstructed.

[0080] The missing reason codes include at least five categories: missing conductivity degradation features, missing thermal response hysteresis features, missing post-action recovery features, insufficient samples, and empty fields.

[0081] Upon completion, an ablation evidence cluster record is generated, which includes at least the equipment unique identifier, branch unique identifier, event type identifier, reference time point, phase, conductivity degradation characteristic identifier, thermal response hysteresis characteristic identifier, post-action recovery characteristic identifier, missing cause code, record source version number, and rule version number. It is stored in the evidence record area of ​​the field control terminal in a structured text format and simultaneously written to the evidence timing area of ​​the power distribution monitoring host. The next step only calls the record based on the equipment unique identifier, branch unique identifier, event type identifier, reference time point, and phase.

[0082] The maximum delay for a single disk write can be set to 500 milliseconds, the number of concurrent connections can be set to 32 branches per terminal, the number of retries can be set to 3, the retry interval can be set to 200 milliseconds, old version records must not overwrite new archived records, and if the same device unique identifier, branch unique identifier, event type identifier, reference time point, and phase combination key are repeatedly uploaded and the time deviation does not exceed 20 milliseconds, they are considered to be the same ablation evidence cluster, and only the communication status is updated.

[0083] During on-site inspection, no less than 300 event measurement window samples are used as the basis for random checks. The consistency rate of three types of feature extraction, the consistency rate of ablation evidence cluster merging, and the matching rate of missing cause codes are checked by grouping according to the same model, the same branch type, and the same event type. The consistency rate can be set to no less than 99%, and the matching rate can be set to no less than 98%.

[0084] Preferably, in the motor branch of a three-phase molded case circuit breaker with a rated current of 250 amps, in the event characteristics corresponding to a certain closing event, the voltage drop between the unified terminals of phase A is continuously higher than the upper boundary of the reference interval by 900 milliseconds, and the unified operating current remains within the heavy load state boundary. The temperature rise response of phase A near the contact area starts 120 milliseconds later than the reference boundary and falls back to the baseline adjacent interval after 1600 milliseconds. After the action is completed, the recovery completion time corresponding to the later moment when both the voltage drop between the unified terminals and the temperature rise response of the unified near the contact area enter the stable interval is 420 milliseconds. Thus, the conductivity degradation characteristics, thermal response hysteresis characteristics, and post-action recovery characteristics of phase A are extracted, and an ablation evidence cluster is recorded according to phase A.

[0085] Alternatively, when the post-action recovery characteristics cannot be directly obtained from the unified opening and closing action sequence, the reference time point is allowed as the action completion time only when the unified opening and closing action sequence is empty but the reference time point and event type identifier are complete and the continuous inter-terminal voltage drop records and continuous near-contact temperature rise response records of the same phase are complete. This is formed by the continuous inter-terminal voltage drop records and continuous near-contact temperature rise response records after the action is completed, and the action time records outside the event measurement window are not introduced. Other phase boundaries, version locking methods, archiving methods, and deduplication methods remain unchanged.

[0086] S6. Based on the evidence strength, consistency, and persistence of the ablation evidence cluster, generate the contact ablation grading result and output the corresponding maintenance judgment. The specific implementation is as follows: First, retrieve the conductivity degradation characteristic identifier, thermal response hysteresis characteristic identifier, post-action recovery characteristic identifier, and corresponding missing reason code from the previously archived ablation evidence cluster according to the equipment unique identifier, branch unique identifier, event type identifier, reference time point, and phase. Then, generate the contact ablation grading result based on the evidence strength, consistency, and persistence of the ablation evidence cluster, and output the corresponding maintenance judgment. This action occurs at the operating site of molded case circuit breakers in low-voltage distribution cabinets, power distribution boxes, charging pile distribution units, and industrial production line end distribution branches. It is executed by the grading recording program in the cabinet control terminal. The action period is after the ablation evidence cluster is archived and before the maintenance judgment is written. The upstream receives the locked version of the ablation evidence cluster, and the downstream only provides the contact ablation grading result and the corresponding maintenance judgment, without rewriting the original ablation evidence cluster.

[0087] The strength of evidence is fixed, referring to the record of the boundary attainment status of the three types of features in the same ablation evidence cluster relative to the corresponding rule version's grading boundary. It includes at least the record of the number of features reaching the grading boundary and the record of the highest boundary level reached. This is achieved by first comparing whether the conductivity degradation feature, thermal response hysteresis feature, and post-action recovery feature reach the mild, moderate, and severe boundaries respectively, and then forming the record based on the number of features reaching the boundary and the highest boundary level reached. The consistency of evidence is fixed, referring to the record of whether the landing points of at least two types of features among the successfully extracted features are at the same level or single-step adjacent levels, provided that the minimum conditions for the ablation evidence cluster are met. Mild and moderate are adjacent levels, moderate and severe are... Adjacent levels, mild and severe are not adjacent levels; evidence persistence refers to the continuous record of evidence in the same level direction or single-step adjacent level direction repeatedly appearing in the continuous event measurement window under the same equipment unique identifier, the same branch unique identifier, the same phase, and the same event type. It includes at least the record of the number of consecutive satisfactions and the persistence establishment mark. The continuous event measurement window is counted according to the adjacent archived records in the order of the reference time point, and the counting of continuous event measurement windows shall not be chained across event types. There can be no less than 3 continuous event measurement windows. When no more than 1 event measurement window is missing consecutively, it is only allowed to fill in the historical index records forward. The filling window does not generate a new evidence cluster separately, but is only used for continuous counting.

[0088] Before each record enters the classification process, it is first aligned according to the unique device identifier, branch unique identifier, event type identifier, reference time point, and phase. Then, distorted segments caused by version inconsistencies, duplicate archiving in the same window, missing reason code conflicts, and phase mismatches are removed. Next, forward completion is performed on fields that can be completed, and finally, a valid classification record is formed. Among them, conductivity degradation characteristic identifier, thermal response hysteresis characteristic identifier, post-action recovery characteristic identifier, missing reason code, and rule version number should not be automatically completed. When any of the following four situations occur, the processing is stopped and the record is transferred to pending verification or pending supplementation: when the continuous event measurement window is insufficient, a formal contact ablation classification result is not generated, but a temporary classification result for mild ablation is allowed to be generated based on the strength and consistency of evidence.

[0089] When generating contact ablation grading results, the process is performed at the single-event measurement window and single-phase level, following a fixed order of evidence strength, evidence consistency, and evidence persistence. Contact ablation grading results can be set as mild ablation, moderate ablation, and severe ablation, with a provisional grading result as a restricted type. The provisional grading result is a restricted result within the contact ablation grading system and is generated only when the formal grading conditions are not fully met. Mild ablation is generated when evidence strength reaches the mild boundary and evidence consistency is established; conversely, ablation is generated when evidence strength reaches the moderate boundary, evidence consistency is established, and evidence persistence reaches the persistent level. A moderate ablation grade is generated when the threshold is met. A severe ablation grade is generated when the evidence strength reaches the severe boundary, the evidence consistency is established, and the evidence persistence reaches the severe persistence threshold. When the evidence consistency is not established, no formal contact ablation grade result is generated; only the evidence inconsistency reason code is recorded. When the evidence strength reaches the moderate boundary but the evidence persistence is insufficient, only a temporary grade result of mild ablation is generated. When the evidence strength reaches the severe boundary but the severe persistence threshold is not met, severe ablation should not be generated; moderate ablation can be generated based on whether the moderate condition is met. If the moderate condition is not met, only a temporary grade result of mild ablation is generated.

[0090] The boundaries for the strength of evidence, the boundaries for the establishment of consistency of evidence, the persistence threshold, and the severe persistence threshold are all preferentially taken from the locked values ​​of the rule version, and locked separately according to the same model, the same branch type, and the same event type. During the first round of operation, the locked values ​​of the rule version are determined by the reference baseline record formed by no less than 30 health event measurement window samples under the same model, the same branch type, and the same event type, combined with the preset classification thresholds. Among them, the health event measurement window samples are used to form the reference baseline and the lower limit of the persistence threshold, and the preset classification thresholds are used to determine the mild, moderate, and severe boundaries. Switching is not allowed within the same continuous operation cycle. When the rule version is missing, the formal classification is stopped and the system is put into pending verification.

[0091] When outputting the corresponding maintenance judgment, the fixed judgment is to give one of the following based on the contact erosion classification result: continue operation judgment, planned maintenance judgment, or priority maintenance judgment, and simultaneously write the attached processing requirements field. Among them, mild erosion corresponds to continue operation judgment and writes subsequent retesting requirements; moderate erosion corresponds to planned maintenance judgment and writes processing requirements in the next maintenance window; severe erosion corresponds to priority maintenance judgment and writes requirements for immediate power outage maintenance; temporary classification results only correspond to continue operation judgment and write requirements for shortening the retesting cycle, and cannot directly correspond to priority maintenance judgment.

[0092] Upon completion, a hierarchical record is generated. The hierarchical record includes at least the equipment unique identifier, branch unique identifier, event type identifier, reference time point, phase, evidence strength identifier, evidence consistency identifier, evidence persistence identifier, contact erosion classification result, maintenance judgment, incidental handling requirements, cause code, record source version number, and rule version number. It is stored in the hierarchical record area of ​​the field control terminal in a structured text format and simultaneously written to the hierarchical time sequence area of ​​the power distribution monitoring host. The next step only calls the record according to the equipment unique identifier, branch unique identifier, event type identifier, reference time point, and phase.

[0093] The maximum latency for a single disk write can be set to 500 milliseconds, the number of concurrent connections can be set to 32 branches per terminal, the number of retries can be set to 3, the retry interval can be set to 200 milliseconds, old version records must not overwrite new archived records, and if the same device unique identifier, branch unique identifier, event type identifier, reference time point, and phase combination key are repeatedly uploaded and the time deviation does not exceed 20 milliseconds, they are considered as the same hierarchical record, and only the communication status is updated.

[0094] The minimum set of reason codes includes at least six categories: insufficient evidence, insufficient sample, version conflict, missing index, inconsistent evidence, and temporary classification.

[0095] During on-site inspection, no less than 300 event measurement window samples are used as the basis for random inspection. The consistency rate, maintenance judgment matching rate and the consistency rate of continuous event measurement windows are checked by grouping according to the same model, the same branch type and the same event type. The consistency rate can be set to no less than 99% and the matching rate can be set to no less than 98%.

[0096] Preferably, in the motor branch of a three-phase molded case circuit breaker with a rated current of 250 amps, in the evidence cluster of phase A ablation corresponding to a certain closing event, all three types of features reach the moderate boundary, the number of features reaching the boundary is 3, the highest boundary level is moderate, the three points of impact are in the same level direction, and the same level direction evidence appears repeatedly in 4 consecutive event measurement windows. Thus, the contact ablation classification result of moderate ablation of phase A is generated, and the planned maintenance judgment and the processing requirements in the next maintenance window are output.

[0097] Alternatively, when the continuous event measurement window is insufficient, a temporary classification result for mild ablation can be generated based solely on the strength and consistency of evidence, but moderate or severe ablation should not be generated directly. Other version locking methods, archiving methods, and deduplication methods remain unchanged.

[0098] In the operational scenario shown in this embodiment: taking a three-phase molded case circuit breaker with a rated current of 250 amps in the power distribution cabinet at the end of a new energy vehicle assembly workshop as the object, this molded case circuit breaker corresponds to a power branch of a welding conveyor line, the branch type is locked as a motor branch, the unique identifier of the equipment is MCCB-250-07, the unique identifier of the branch is LINE-MOTOR-03, the field acquisition device, current transformer, inter-terminal voltage drop sampling circuit, temperature measurement point near the contact area of ​​the casing, auxiliary contacts and mechanical position status sampling line have all been deployed according to the version of the record source, and the rule version has been locked after the first round of tuning.

[0099] After the branch line experienced one shutdown during continuous production, power was restored and accompanied by multiple normal start-stop operations. The control terminal inside the cabinet continuously received auxiliary contact status records, mechanical position status records, trip indication contact records, branch voltage existence records, branch current records, and manual operation marks.

[0100] Before a certain line start-up, the mechanical position status record flips from 0 to 1, the auxiliary contact status record then flips to the closed state, the branch voltage recovers to more than 90% of the rated voltage and remains continuously, the branch current reappears within 3 sampling cycles and the fluctuation range is within the allowable range, the event identification program completes the judgment in a fixed order of tripping event, opening event, closing event, and power restoration event, confirms that the action is a power restoration event, and uses the candidate state change point timestamp as the reference time point to establish a 500-millisecond pre-observation segment and a 1200-millisecond post-observation segment, forming the corresponding event measurement window and writing it into the event record area.

[0101] Subsequently, the measurement and recording program collected the operating currents of phases A, B, and C at a sampling rhythm of 20 milliseconds within the event measurement window, collected the inter-terminal voltage drops of phases A, B, and C at the same rhythm, and collected the temperature rise response of the near-contact area of ​​phases A, B, and C at a sampling rhythm of 200 milliseconds. It also recorded the time difference between the mechanical position state reversal time and the auxiliary contact state reversal time when the event was possible. Within this event measurement window, the operating current of phase A stabilized between 184 A and 188 A, the operating current of phase B stabilized between 179 A and 183 A, and the operating current of phase C stabilized between 181 A and 185 A. The inter-terminal voltage drop of phase A remained around 42 millivolts, the inter-terminal voltage drop of phase B was approximately 39 millivolts, and the inter-terminal voltage drop of phase C was approximately 40 millivolts. The temperature rise response of the near-contact area of ​​phase A gradually increased from the average temperature of the previous baseline interval and formed a continuous temperature rise trajectory. The event type identifier was fixed and inherited as a power restoration event, ultimately forming a complete event measurement dataset.

[0102] After reading the event measurement dataset, the unified recording program calculates the rated current percentage based on the rated current value in the equipment ledger. It then determines the unified load state of the event measurement window as the heavy load state based on the state corresponding to the largest rated current percentage among the three phases. Simultaneously, it confirms the ambient temperature as 37 degrees Celsius based on the temperature measurement point record of the common environment in the installation cavity, corresponding to the high temperature range. It then inherits the event type as a power restoration event, converts the original operating current into a unified operating current record, retains the inter-terminal voltage drop as a unified inter-terminal voltage drop record with the heavy load state boundary, and retains the near-contact area temperature rise response as a unified near-contact area temperature rise response record with the high temperature range identifier and the heavy load state boundary. Since this is a power restoration event, the unified opening and closing action sequence field is set to null and a cause code is appended, thus forming a comparable event feature set.

[0103] The exclusion record procedure continues to retrieve comparable event feature sets from adjacent branches within the same installation cavity in adjacent event measurement windows, constructs an interference exclusion matrix, and verifies whether terminal connection abnormalities, heat dissipation limitation abnormalities, and mechanism hysteresis abnormalities are established. In this event, although the voltage drop between the unified terminals of phase A is higher than the upper boundary of the reference range under the same model, branch type, event type, and load state boundary, the temperature rise response of the phase A unified near-contact area rises synchronously, and other adjacent branches within the same installation cavity do not show common temperature rise. In the power restoration event, mechanism hysteresis abnormality is not determined. Therefore, terminal connection abnormality, heat dissipation limitation abnormality, and mechanism hysteresis abnormality are not triggered in this event measurement window. The exclusion field identifier remains empty, and all fields are retained as event features to enter the next stage.

[0104] Subsequently, the feature extraction program extracted conductivity degradation features, thermal response hysteresis features, and post-action recovery features from the retained event features. Among them, the voltage drop between the unified terminals of phase A was continuously higher than the upper boundary of the reference interval, and the unified operating current was always kept within the heavy load state boundary, forming the conductivity degradation feature of phase A; the temperature rise response of phase A near the contact area lagged behind the average temperature of the previous baseline interval by about 120 milliseconds from the reference boundary, remained stable after reaching the offset interval, and fell back to the baseline adjacent interval after 1600 milliseconds, forming the thermal response hysteresis feature of phase A; since this was a power restoration event, the post-action recovery feature was not extracted and the missing reason code was recorded.

[0105] Based on the minimum conditions for the establishment of an ablation evidence cluster, two types of features, namely conductive degradation features and thermal response hysteresis features, have been successfully extracted from phase A. Therefore, a single ablation evidence cluster record for phase A is formed by merging the unique equipment identifier, branch unique identifier, event type identifier, reference time point, and phase.

[0106] The grading recording program then retrieves the A-phase ablation evidence cluster and the historical ablation evidence clusters corresponding to the first three consecutive event measurement windows under the same branch, phase, and event type to determine the strength, consistency, and persistence of the evidence. Among them, the conductivity degradation feature and thermal response hysteresis feature in this A-phase ablation evidence cluster both reach the moderate boundary, and the number of features reaching the grading boundary is 2, with the highest boundary level being moderate, constituting moderate evidence strength. The landing points of the two types of successfully extracted features are in the same grade direction, constituting the consistency of the evidence. At the same time, the evidence in the same grade direction appears repeatedly in four consecutive event measurement windows, constituting the persistence of the evidence. Therefore, the contact ablation grading result of moderate A-phase ablation is generated, and the planned maintenance judgment is output according to the pre-locked maintenance mapping relationship, while the processing requirements in the next maintenance window are written.

[0107] The hierarchical record is ultimately written synchronously in the hierarchical record area of ​​the field control terminal and the hierarchical time sequence area of ​​the power distribution monitoring host in a structured text format. This allows subsequent maintenance and management personnel to directly access the record without rewriting the original event records, event measurement datasets, comparable event feature sets, filtered records, and ablation evidence clusters. This completes the closed-loop operation scenario from event identification, form creation, data acquisition, operating condition normalization, interference elimination, evidence extraction to ablation classification and maintenance judgment.

[0108] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0109] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0110] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0111] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0112] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0113] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0114] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0115] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0116] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0117] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart assessment method for contact erosion of molded case circuit breakers, characterized in that, include: S1. Identify the opening, closing, tripping, and power restoration events of the molded case circuit breaker, and establish an event measurement window around each target operation event; S2. Collect operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, and event type identifier within the event measurement window to form an event measurement dataset; S3. Perform operating condition normalization processing on the event measurement dataset based on load status, ambient temperature, and event type to generate a comparable event feature set; S4. Construct an interference elimination matrix based on the comparable event feature set to screen out event features corresponding to terminal connection abnormalities, heat dissipation limitation abnormalities, and mechanism hysteresis abnormalities. S5. Extract conductive degradation features, thermal response hysteresis features, and post-action recovery features from the event features after filtering out interference, and combine them to form an ablation evidence cluster. S6. Generate contact ablation grading results based on the evidence strength, consistency, and persistence of the ablation evidence cluster, and output the corresponding maintenance judgment.

2. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 1, characterized in that, S1 includes: Event identification is performed in a fixed sequence for tripping events, tripping events, closing events, and power restoration events. The confirmed target runtime event reference time point is the timestamp of the candidate state change point; Establish an event measurement window around a reference time point; The event measurement window includes the pre-event baseline interval corresponding to the preceding observation segment and the post-event response interval corresponding to the following observation segment.

3. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 1, characterized in that, S2 include: Capture records within the window based on the event measurement window boundaries; Alignment is performed using a unified timestamp; Records are grouped into the same group within the window based on the device's unique identifier, branch's unique identifier, event type identifier, reference time point, sampling time, and phase. An event measurement dataset is formed by using the device unique identifier, branch unique identifier, event type identifier, reference time point, sampling time, phase, operating current, inter-terminal voltage drop, near-contact area temperature rise response, opening and closing action sequence, record source version number, rule version number, and cause code.

4. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 1, characterized in that, S3 include: First, group and normalize according to load status, then partition and normalize according to ambient temperature, and finally form corresponding normalized fields according to event type. The normalized operating current is recorded as a percentage of the rated current. The normalized inter-terminal voltage drop is the original inter-terminal voltage drop record with uniform load state boundary attached; The normalized near-contact temperature rise response is the original near-contact temperature rise response record with ambient temperature range markings and unified load state boundaries; The unified opening and closing action sequence is the original opening and closing action sequence record under the corresponding event type, which retains the original time difference direction and time difference unit.

5. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 1, characterized in that, S4 include: The data are grouped together according to the equipment unique identifier, branch unique identifier, event type identifier, reference time point, phase and measurement window under the same event, including the unified operating current, unified inter-terminal voltage drop, unified near-contact temperature rise response, unified opening and closing action sequence, load status, and ambient temperature range identifier. An interference exclusion matrix is ​​formed by using a single event measurement window as the granularity and by phase. The interference exclusion matrix includes terminal connection anomaly markers, heat dissipation limitation anomaly markers, mechanism hysteresis anomaly markers, exclusion field identifiers, and reserved event feature identifiers.

6. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 5, characterized in that, S4 also includes: Terminal connection anomalies are only filtered out from the normalized inter-terminal voltage drop field. The heat dissipation limitation anomaly only filters out the normalized near-contact area temperature rise response field; The mechanism lag anomaly only filters out the normalized opening and closing action timing field; In the filtered records, the corresponding fields are marked as exclusion fields and are not included in the retained event characteristics when called downstream.

7. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 1, characterized in that, S5 include: Extract conductivity degradation features, thermal response hysteresis features, and post-action recovery features based on the device's unique identifier, branch's unique identifier, event type identifier, reference time point, and phase. Using a single event measurement window and a single phase as the granularity, the successfully extracted conductivity degradation features, thermal response hysteresis features, and post-action recovery features are merged into ablation evidence clusters according to the unique equipment identifier, branch unique identifier, event type identifier, reference time point, and phase. The minimum condition for the ablation evidence cluster to be valid is that at least two of the three types of features are successfully extracted.

8. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 1, characterized in that, S6 include: Retrieve conductivity degradation characteristics, thermal response hysteresis characteristics, post-action recovery characteristics, and corresponding missing reason codes by device unique identifier, branch unique identifier, event type identifier, reference time point, and phase; Using a single event measurement window and a single phase as the granularity, contact ablation grading results are generated in a fixed order of evidence strength, evidence consistency, and evidence persistence.

9. The intelligent assessment method for contact erosion of a molded case circuit breaker according to claim 8, characterized in that, S6 also includes outputting corresponding maintenance judgments based on the contact erosion classification results, wherein: When the strength of evidence reaches the slight boundary and the consistency of evidence is established, a slight ablation is generated and a decision to continue running is output. When the strength of evidence reaches the moderate boundary, the consistency of evidence is established, and the persistence of evidence reaches the persistence threshold, a moderate ablation is generated and a planned maintenance judgment is output. When the strength of evidence reaches the severe boundary, the consistency of evidence is established, and the persistence of evidence reaches the severe persistence threshold, severe ablation is generated and a priority repair judgment is output.

Images