A dual-parameter condition assessment method for electrothermal disconnect switches in GIS
By acquiring and processing drive motor current, loop current, and temperature data in GIS disconnect switches, and using thermal parameter domain separation to separate the thermal response of the preceding and current contact states, the problem of thermal response identification in continuous switching operations is solved, and the test accuracy is improved.
Patent Information
- Application Number
- CN202611359993.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-03
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-03
AI Technical Summary
In continuous switching operations of GIS disconnect switches, existing technologies have difficulty effectively distinguishing between the attenuated thermal response of the previous contact state and the newly added thermal response of the current contact state, resulting in inaccurate electrical characteristic test results and affecting fault condition determination and health benchmark updates.
By acquiring the drive motor current, primary circuit current, GIS casing temperature, and ambient reference temperature during continuous switching, the operation stages are divided. Ambient temperature correction and electrical heating separation are performed using the calibrated thermal parameter domain to form an additional temperature rise sequence. Adjacent contact state versions are recursively derived using the same thermal parameter and current square excitation conservation. The previous and current temperature responses are separated, the equivalent incremental resistance is solved, and a versioned state assessment record is formed.
It enables accurate differentiation of the thermal response between the preceding and current contact states under continuous switching conditions, ensuring that the temperature response and contact thermal state are correlated with the corresponding contact state version, thereby improving the accuracy of electrical characteristic testing of GIS disconnect switches.
Smart Images

Figure CN122836557B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of electrical condition detection technology for high-voltage switchgear, and in particular to a dual-parameter electrothermal condition assessment method for GIS disconnect switches. Background Technology
[0002] GIS disconnect switches are used for substation busbar switching, operating mode adjustment, and electrical isolation of equipment. The contact state of their contacts and the state of their operating mechanism directly affect the operational safety of the conductive circuit. Current status detection methods typically collect the drive motor current during the opening and closing process, test the load changes of the operating mechanism based on the current response during the startup, transmission, and final contact stages, and collect the GIS casing temperature after closing to test the heating state of the contact conductive circuit based on the casing temperature rise. The drive motor current reflects the instantaneous load of the operating mechanism when overcoming mechanical resistance, while the casing temperature is affected by the contact resistance, primary circuit current, and the heat transfer process of the equipment. These two types of measurements can provide data from different time scales. Characterizing the operating status of a disconnector, under conditions where there is sufficient cooling time after a single operation, a relatively clear correspondence can usually be formed between the operation process and the subsequent temperature rise. However, during the continuous switching process of a double busbar, after the disconnector is closed for the first time and carries current for a short period of time, it may be opened again due to the adjustment of the switching sequence, and closed again before the shell temperature has recovered to a stable state. The heat generated during the first contact will still propagate to the shell along the conductor, shielding structure and gas medium after the contacts are separated. The second closing establishes a new contact state, and may be accompanied by a redistribution of the primary circuit current, so that the same temperature measurement window simultaneously bears the decayed thermal response of the previous contact state and the new thermal response of the current contact state.
[0003] Existing processing methods that use a single opening and closing operation as the basic test sample typically extract the temperature sequence based on the moment the operation is completed, and use the subsequent temperature rise as the test result of the current contact state. This processing method faces the problem of inconsistency between the contact state boundary and the thermal response boundary under continuous switching conditions. The measured contact state corresponding to the drive motor current has changed with the re-opening and closing, while the thermal response included in the shell temperature may still come from the previous contact state where the current carrying has ended. Even if the current signal and temperature signal are collected using a unified clock, and the temperature curve is directly segmented according to the re-closing moment, it is difficult to determine which part of the temperature change belongs to the free decay of the previous contact state and which part belongs to the forced response formed by the current contact state under the action of the new load current. When the two types of responses overlap in the same temperature measurement range, the delayed temperature rise caused by the previous abnormal contact may be attributed to the current contact state, and the temperature rise caused by the current contact may also be masked by the historical thermal state, thus affecting the object attribution, fault state determination, and health benchmark update of the GIS disconnector electrical characteristic test results. Therefore, it is necessary to solve the problem of distinguishing between the previous thermal response and the current contact thermal response under continuous switching conditions. Summary of the Invention
[0004] This application proposes a dual-parameter electrothermal condition assessment method for GIS disconnect switches to address the problems mentioned in the background art.
[0005] To achieve the above objectives, this application adopts the following technical solution: a method for evaluating the dual-parameter status of a GIS disconnector switch based on both electrothermal parameters, comprising:
[0006] During continuous switching, obtain the drive motor current, primary circuit current, GIS enclosure temperature and ambient reference temperature of the target electrical phase; divide the operation stage according to the drive motor current, and combine the primary circuit current boundary to form adjacent contact state versions, determine the closed current carrying interval, contact thermal excitation interval, total current square excitation determination interval and end crimping response interval;
[0007] The ambient temperature correction and normal electrical heating of the GIS shell temperature are separated by using the calibrated thermal parameter domain to form an additional temperature rise sequence; the additional thermal state at the end of the previous contact state version is continuously recursively deduced according to the same thermal parameter and the joint thermal excitation trajectory that satisfies the current square excitation conservation, thus forming a set of paired temperature responses.
[0008] Based on the total current square excitation, the interval change is determined to form a candidate assignment window. For the pairwise temperature response set corresponding to the candidate incremental resistance state with an incremental resistance difference not less than the calibrated incremental resistance interval, a separation test is performed to form the assignment window.
[0009] Within the attribution window, the paired temperature response sets are constrained by the additional temperature rise sequence. The joint feasible set of equivalent incremental resistance for adjacent contact state versions is solved. The conservative outer bound of the equivalent incremental resistance is formed by axial projection. The conservative outer bound of the equivalent incremental resistance is compared with the determination interval of the equivalent incremental resistance. The end-pressing response interval is compared with the allowable interval of the end-pressing response to form a versioned state evaluation record.
[0010] The beneficial effects of this invention are as follows:
[0011] This invention separates the decayed thermal response of the previous contact state from the newly added thermal response of the current contact state under the same thermal parameters and current square excitation conservation constraints. It also solves the equivalent incremental resistance only for adjacent contact state versions that have identifiable conditions. This is different from the existing processing method that directly extracts the temperature sequence according to the reclosing time and attributes the subsequent temperature rise to the current contact state. This ensures that the temperature response, contact thermal state and end crimping response under continuous switching conditions are all associated with the corresponding contact state version. Attached Figure Description
[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort:
[0013] Figure 1 This is a comparison chart of the stator current curves of the drive motor under normal and jammed conditions.
[0014] Figure 2 Temperature distribution diagrams of GIS contacts and housing under different equivalent incremental resistance conditions;
[0015] Figure 3 This is a graph showing the relationship between the maximum temperature rise of the casing and the maximum temperature rise of the contacts as a function of the current carrying capacity, the equivalent incremental resistance, and the ambient temperature. Detailed Implementation
[0016] 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.
[0017] like Figure 1 , Figure 2 and Figure 3 As shown, this invention discloses a method for evaluating the dual-parameter status of a GIS disconnector switch based on both electrothermal and thermal parameters, comprising the following specific steps:
[0018] This embodiment uses the equipment structure and field measurement channel of the GIS disconnector at the Yiwutang 220kV substation as the target object. The current records of the operating mechanism drive motor obtained by the Shandong Taikai test platform are only used to illustrate the engineering form of the drive motor current under normal, jammed, and stalled states, and are not directly used to form the action judgment limit, stage reference distribution, or end-point crimp response allowable range of the target equipment. All of the above objects are formed by the fixed health reference records of the target equipment. The temperature field, current-driven temperature rise records, and current-interrupted cooling records under different contact resistance conditions are used to form an initial thermal parameter domain that matches the structure of the target equipment. Before this parameter domain is used for online status assessment of the target equipment, at least three independent normal current-driven cooling records of the target equipment are used to verify whether its temperature trajectory is correct. If the temperature trajectory recorded by the target device cannot be covered by the parameter field, and the three records cover no less than two offline calibration current points, then S3 and S4 are not entered. The supplementary continuous switching numerical verification sequence is limited to the current, equivalent incremental resistance, ambient temperature, and thermal response duration actually covered by the verified parameter field. It is only used to verify whether the processing chain can obtain a consistent output under known input conditions. It is not used as proof of field measurement accuracy, nor is it used to provide a known equivalent incremental resistance label to the online processing. After the previous contact state is opened, the internal heat still propagates to the shell. After the current contact state is re-established and current is carried, the shell temperature is simultaneously affected by the previous residual heat and the current heating effect. The method of this invention is used to assign the two parts of the temperature response to the corresponding contact state version respectively.
[0019] The specific processing chain is as follows: A contact state version is generated based on the closing stage, opening stage, and continuous current-carrying boundary, and a current square excitation is generated for each version according to the temperature sampling period; the previous thermal state is retained in the same thermal state model, and the temperature response of the previous contact state and the current contact state are calculated separately; the current load change is used to check whether the two types of response sets can be separated; when the separation conditions are met, the joint feasible set of the previous and current equivalent incremental resistances is solved; the contact thermal state determination result and the end mechanical response determination result are associated with the corresponding contact state version to form a versioned state assessment record.
[0020] The drive motor current and supply voltage of the operating mechanism are collected by drive motor current sensor and voltage sensor. The target electrical phase current is collected by primary circuit current measurement channel. The temperature field of GIS shell and ambient reference temperature are collected by infrared thermal imaging device and ambient temperature sensor. Each acquisition channel records a unified absolute timestamp. The drive motor current is used for action phase and end-point crimping response analysis. The primary circuit current is integraled with square according to the temperature sampling period and used as thermal excitation.
[0021] In this embodiment, the target device is a three-phase box-type GIS disconnector in the Yiwutang 220kV substation. One target electrical phase is selected as the analysis object. The target temperature measurement area is determined according to the geometric projection of the contacts and shielding cover on the outer shell. Its boundary is the union of the geometric projection area and the position error band obtained by repeated image registration. Repeated image registration is performed under the conditions of stable equipment temperature, constant camera focal length and installation distance. The pixel displacement envelope after multiple resets is obtained with the fixed outer shell mark point as the reference to form the position error band. The reference temperature measurement area adopts a candidate mask with the same area and shape as the target temperature measurement area. The candidate mask is translated outside the target temperature measurement area according to the preset pixel step size during installation calibration. Candidate regions are generated and excluded that overlap with the position error zone, field of view edge, and projections of other known heat sources. For each remaining candidate region, the difference between the highest and lowest regional temperatures in the offline temperature fields of 0μΩ to 100μΩ, 1000A to 2500A, and -10℃ to 30℃ is calculated, and the candidate region with the smallest difference is selected. When there are ties, the region with the smaller number is selected according to the spatial number fixed during installation and calibration. The target temperature measurement region and the reference temperature measurement region are collectively referred to as the fixed temperature measurement region. Its boundary and pixel mask remain unchanged after installation and calibration and do not move according to the highest temperature position in the online thermal image. The environmental reference temperature is obtained from the location in the same GIS indoor space that is not directly heated by the radiation of the target conductor.
[0022] The sampling frequency of the drive motor current and supply voltage is preferably 2kHz to 10kHz, and 5kHz is used in this embodiment; the sampling frequency of the primary circuit current of the target electrical phase is preferably 1kHz to 5kHz, and 2kHz is used in this embodiment; the sampling period of the GIS enclosure temperature field and the ambient reference temperature is preferably 0.5s to 5s, and 1s is used in this embodiment. The aforementioned values ensure that there are at least 40 drive motor current sampling points in each 50Hz power frequency cycle, and that the temperature sampling period is significantly shorter than the enclosure thermal response change time. The maximum time scale error of the acquisition system is preferably 1ms to 20ms, and the maximum allowable time scale error is set to 10ms in this embodiment. This error adopts the upper limit of the allowable error given in the calibration certificate and is incorporated into the upper limit of the contact establishment interval, the contact termination interval, and the total current square excitation error, and is not used as an independent temperature correction quantity.
[0023] This embodiment establishes three types of offline fixed data during the offline phase. The first type is a fixed health reference set, with an optimal number of 3 to 10 effective records. In this embodiment, 6 healthy closing records are selected, all from the same device, in the same closing direction, with complete operation stages, qualified power supply voltage, and consistent measurement and installation conditions. The second type is normal electrical heating calibration records and temperature measurement verification records. The current carrying capacity is preferably covered from 1000A to 2500A, and in this embodiment, 1000A, 1500A, 2000A, and 2500A are selected. The ambient temperature is preferably covered from -10℃ to 30℃, with 20℃ used for supplementary numerical verification. The third type... The equivalent incremental resistance heating calibration record is used, preferably covering 0μΩ to 100μΩ. In this embodiment, 0μΩ, 20μΩ, 40μΩ, 60μΩ, 80μΩ, and 100μΩ are taken at 20μΩ intervals. 0μΩ indicates no additional contact resistance relative to the normal contact reference. For both normal electrical heating and equivalent incremental resistance heating, preferably 3 to 10 independent and complete temperature rise-cooling records are obtained at each calibration point. In this embodiment, supplementary numerical calibration forms 5 records at each calibration point. Each record is restarted after the model state or test equipment returns to the initial allowable temperature range. It includes a complete flow-through heating section and a subsequent flow-interruption cooling section under a defined operating condition; the effective duration of the heating-cooling record is preferably 900s to 3600s, and in this embodiment, the supplementary numerical calibration is set to 1800s to ensure that the longest thermal time constant has complete calibration time coverage; different current points or different equivalent incremental resistance points form independent records, and multiple calibration points are not spliced into a single heating-cooling record. The complete verification sequence for the equivalent incremental resistance determination interval is generated by these calibration records according to a fixed preceding state, current state, and load change program of 1000A to 1500A. Each target state preferably forms 3 to 10 sequences. In this embodiment, 5 sequences are used. Each calibration record and verification sequence saves the equipment or model version, measurement location, current flow, ambient temperature, and verification results of preset equivalent incremental resistance and loop resistance. The preset equivalent incremental resistance is only used for offline generation and result verification and is not used as input for online solution. The 10.5℃ normal record and 33.9℃ abnormal temperature record in the field test are only used to verify the magnitude of the temperature measurement channel. The two do not constitute the same continuous switching sequence and do not participate in the joint solution of the previous and current thermal responses. The above offline fixed data remain unchanged during online processing.
[0024] This embodiment performs voltage access based on the actual power supply voltage range covered by the health operation records of the same equipment. Specifically, a moving average root-mean-square voltage sequence for one power frequency cycle is calculated for each of the six fixed health reference records. The minimum value among all the minimum values of the sequence minus the voltage sensor's allowable error is used as the lower limit of access, and the maximum value among all the maximum values of the sequence plus the voltage sensor's allowable error is used as the upper limit of access. Access is granted when the moving average root-mean-square voltage of each complete power frequency cycle in the current operation is within this range. The preferred moving electrical energy integration window is 20ms to 100ms. The example uses 40ms, corresponding to two 50Hz power frequency cycles. The equivalent incremental resistance calculation domain is 0μΩ to 100μΩ, with a 20μΩ interval between adjacent calibration points as the specified resolution for identifiable verification. The normal and abnormal equivalent incremental resistance judgment intervals are formed by leave-one-back calculation of the complete verification sequence. Each verification sequence includes a known previous state, a known current state, a load change of 1000A to 1500A (the same as the supplementary numerical verification), and a complete temperature record. One complete verification sequence is used as the verification object in sequence, and other similar sequences are used to form... The thermal parameter domain is then used to back-calculate the target version in the excluded sequence according to the joint solution of S2, S3, and S4 and the axial projection process. Its known state label is not used during the back-calculation process. The minimum value of the lower boundary and the maximum value of the upper boundary of the entire back-calculation outer bounding interval are used to form the corresponding judgment interval. The 0μΩ target version is used to form the normal equivalent incremental resistance judgment interval; the 60μΩ target version, after independent fault pre-setting and loop resistance verification, is used to form the abnormal equivalent incremental resistance judgment interval; the 20μΩ, 40μΩ, 80μΩ, and 100μΩ records are used to constrain the thermal model, regardless of whether they are non-standard. Zero-point positioning automatically assigns anomaly labels. In the 5 sets of leave-one-back calculations for supplementary numerical verification, the normal equivalent incremental resistance judgment range corresponding to the 0μΩ target version is 0μΩ to 10μΩ. The abnormal equivalent incremental resistance judgment range corresponding to the 60μΩ target version after independent fault preset and loop resistance verification is 50μΩ to 70μΩ. S3 is used to check whether the candidate states with a difference of 20μΩ can be distinguished by temperature response. S4 only forms the contact thermal state judgment result when the conservative outer range width is less than 20μΩ and the range completely falls into an equivalent incremental resistance judgment range.
[0025] To illustrate the attribution process of heat across the contact state in this invention, this embodiment constructs a supplementary continuous switching numerical verification condition within the finite element temperature field model and current calibration range. The preferred duration of the preceding current flow is 120s to 300s, and in this embodiment, it is 180s. The preferred interruption interval after the preceding trip is 30s to 120s, and in this embodiment, it is 50s. The preceding contact state adopts an independently verified 60μΩ equivalent incremental resistance state. After 180s of current flow with a primary circuit current effective value of 1000A, the circuit is tripped. After 50s of interruption, the circuit is reclosed, and the current contact state adopts a 0μΩ equivalent incremental resistance state. After reclosing, the effective value of the primary circuit current is initially maintained at 1000A, then adjusted to 1500A, and continuously observed for 570s. 60μΩ and 0μΩ correspond to... Figure 2 The calibration status, 1000A and 1500A correspond to Figure 3 The current calibration conditions are used to verify whether the processing logic can restore the known input state, and are not used as field continuous switching test data.
[0026] A contact state version is defined by the closed state defined by the establishment of a contact and the corresponding contact termination, and its measurement records. The previous contact state version is denoted as V_(k-1), and the current contact state version formed by reclosing is denoted as V_k, where k represents the current version number. Each version saves the contact boundary, current square excitation, temperature record, end crimp response and state evaluation results respectively. The previous and current equivalent incremental resistances are solved together in subsequent steps.
[0027] The supplementary numerical verification condition takes the time of the first closing control command as 0s. The operation time period and corresponding data objects are listed in the table below. The time in the table is only used to limit the verification input of this embodiment. Subsequent contact status versions, thermal excitation, temperature observation and status results are all associated with the same target disconnecting switch and the same target electrical phase according to this time axis.
[0028] Table 1, Supplementary table of continuous switching numerical verification time axis and data object correspondence:
[0029] Verification period Supplementing the numerical verification process Main electrical quantities The data object formed 0s to 4.60s Pre-closing operation The drive motor current exhibits normal start-up, free drive, and end-pressing processes. Formation of the pre-closing operation window and contact establishment interval 4.10s to 184.10s Pre-contact state stable current flow The primary loop current is 1000A, and the preset equivalent incremental resistance is 60μΩ. A stable thermal excitation is formed at 180.00s; the measured current carrying capacity from the opening phase to 188.05s continues to be included in the previous version. 184.10s to 188.40s Pre-trip contact status before tripping termination The primary circuit current drops below the measurement lower limit. Formation of the preceding contact termination interval 188.05s to 237.98s Interruption interval between two contact states The primary circuit current is 0, and the temperature of the target temperature measurement area continues to change. When the preceding heat source stops being input, the preceding thermal state continues to propagate to the casing. 234.00s to 238.40s Reclosing operation The drive motor current once again showed a healthy end-crush response. This creates a current contact state version with a preset equivalent incremental resistance of 0 μΩ. From 238.06s to 270.00s Current contact state initial flow The effective value of the primary circuit current is 1000A. The initial forced thermal response that forms the current contact state 270.00s to 570.00s Continue to carry current after current load current adjustment The effective value of the primary loop current was adjusted from 1000A to 1500A and maintained until the analysis cutoff time. It generates a new electrical excitation that acts only on the current closed contact state and provides a candidate window of 46s to 300s.
[0030] The later boundary of the previous contact termination interval to the earlier boundary of the current contact establishment interval is 49.93 s. In the supplementary numerical verification, the previous 60 μΩ is a known preset state used when generating the verification temperature sequence. It is only used to verify in the offline stage that the previous heat is still higher than the temperature measurement resolution at the time of current contact establishment. The online processing does not read this preset value. The online judgment is based only on the allowable thermal state range formed by the interrupted cooling calibration and the measured additional temperature rise at the time of current contact establishment to establish a set of historical additional thermal states. If the set is empty, it is marked as outside the thermal model domain. If the set contains two types of interpretations, one below the temperature measurement resolution and one above the temperature measurement resolution, it is not confirmed in advance whether the previous residual heat is observable. Instead, all interpretations are passed on to S3 and S4.
[0031] The preceding contact state maintains stable current flow from 4.10s to 184.10s, with a stable current flow duration of 180.00s. After the tripping operation begins until the contact is confirmed to terminate at 188.05s, the primary circuit current is still included in the thermal excitation of the preceding version based on the measured value. Therefore, the actual current-carrying integral time of the preceding version is 183.95s, and the current interruption interval changes from 188.05s to 237.98s, with a calculated result of 49.93s. The current load change lasts from 270.00s to 570.00s, which provides the longest candidate assignment window of 300.00s.
[0032] S1 reads the drive motor current, drive motor supply voltage, primary circuit current, GIS enclosure temperature field, and ambient reference temperature under a unified absolute time scale, and uses the available closing arc signal as an auxiliary contact confirmation signal. When the supply voltage is within the aforementioned voltage access range throughout the entire operation period, this operation is used for action phase identification and end-point crimping response calculation. If any consecutive power frequency cycle exceeds the access range, only the original operation timing record is retained, and the valid stage identification flag and valid end-point crimping response flag are set to invalid. The contact state version will no longer be automatically generated based on this operation, nor will it enter the subsequent thermal state attribution processing.
[0033] Before each operation, a period of no operation is selected, consisting of 1 to 5 complete power frequency cycles and no operation command received. In this embodiment, 2 power frequency cycles are preferably selected. The root mean square value of the original drive motor current is calculated for one power frequency cycle. The maximum value of the root mean square value in the no-operation interval is added to the root mean square resolution of the current sensor to form the operation judgment limit. The duration of exceeding or falling below the operation judgment limit is preferably 1 to 3 power frequency cycles. In this embodiment, 1 power frequency cycle is selected: the start time of the operation is determined when the root mean square value is higher than the limit for 1 consecutive power frequency cycle, and the end time of the operation is determined when it is lower than the limit for 1 consecutive power frequency cycle. If the no-operation interval is less than 1 power frequency cycle, or if the change of the root mean square value in the interval exceeds the zero-point drift range given by the sensor calibration, the valid stage identification mark is set to invalid, and the automatic stage identification result and end crimping response are stopped using this operation.
[0034] Between the start and end times of the operation, a sliding window ranging from 20ms to 100ms is used, and in this embodiment, 40ms is used. Time integration is performed on the square of the original current of the drive motor to obtain the sliding electrical energy trajectory. The squaring process is used to retain the contribution of the drive motor current amplitude to the changes in the motor's electromagnetic energy and mechanical load, while eliminating the sign cancellation of the positive and negative half-cycles of AC. Subsequently, the time coordinate within the operation window is converted to dimensionless time from 0 to 1, and the sliding electrical energy is converted to dimensionless energy from 0 to 1 according to the minimum and maximum values within this operation window. If the maximum value is the same as the minimum value, it indicates that no identifiable energy change has been formed in this operation, the valid stage identification flag is set to invalid, and subsequent automatic stage identification is stopped.
[0035] Local curvature is calculated on the dimensionless energy trajectory. For equally spaced discrete sampling points, the first-order difference is obtained by dividing the difference between adjacent dimensionless energy points by the dimensionless time step. Then, the second-order difference is obtained by dividing the difference between adjacent first-order differences by the dimensionless time step, with one-sided differences used at the endpoints. Subsequently, the absolute value of the second-order difference is divided by the sum of 1 and the squares of the first-order differences to the power of 3 / 2. All sampling points within the operation window, excluding the first and last points, are used as initial boundary candidates. Internal boundary combinations are enumerated according to a fixed stage order, and combinations where the duration of any stage is not between the shortest and longest durations of the corresponding stage in the fixed healthy reference set are excluded. For each group of remaining candidate boundaries... Within each candidate stage, the absolute value of the difference between adjacent dimensionless energy values is calculated to obtain a non-negative stage change sequence. This sequence is then divided by the sum of the stage changes to form a stage morphology distribution with a total of 1. When the sum of the stage changes is 0, the candidate stage is excluded. When the number of stage sampling points differs in different records, the stage morphology distribution is first linearly interpolated on stage coordinates from 0 to 1. The number of sampling points is preferably between 51 and 201. In this embodiment, 101 points are used, including two endpoints. Then, the integral of the absolute value of the cumulative distribution difference between the stage morphology distribution and the reference distribution of the healthy stage in the same operating direction is calculated on stage coordinates from 0 to 1, which is used as the one-dimensional optimal transmission distance.
[0036] The phase identification adopts a two-level selection. For each operation direction and each phase, the health phase reference distribution selects a true distribution from the health records participating in the reference formation: calculate the one-dimensional optimal transmission distance between each distribution and the other distributions, and select the distribution with the smallest maximum distance; when there are ties, select the record with the earlier acquisition time. First, in the candidate boundary where the phase order is fixed and the duration of each phase is between the minimum and maximum duration of the fixed health reference set, retain the candidate solution with the smallest sum of phase morphological distances; then select the combination with the largest sum of internal boundary curvatures from the candidate solutions. The upper limit of the distance of the fixed health reference set is formed by leaving one out: take one health record as the verification record in turn, and re-form the health phase reference distribution with the remaining records, and take the maximum value of the corresponding distance of all verification records. In this embodiment, the fixed health reference set takes 6 records. After leaving one out each time, there are still 5 records used to form the phase reference. When there are multiple identical final solutions, select the combination with the earlier first internal boundary time. If the fixed health reference set has less than 3 valid records, or the phase distance of this operation exceeds the allowable upper limit, no automatic phase identification result is output, and no contact boundary is formed based on the result.
[0037] In the supplementary numerical verification, the two closing records were used to determine the start-up stage, free transmission stage, end-pressing stage, and stop stage based on the stage morphological distance and internal boundary curvature; the opening record of the termination contact state version was used to determine the start-up stage, contact release stage, free transmission stage, and stop stage according to the same rules. Figure 1The difference in stator current between normal and jammed states is shown, where jammed records with continuous oscillation and waveform distortion are not included in the fixed healthy reference set.
[0038] For the closing operation that forms the current contact state version, the end crimping stage is taken as the initial range of the contact establishment interval. The effective measurement lower limit of the primary circuit current channel is taken as the larger value between the minimum reliable measurement current given in the calibration certificate and the upper boundary of the background current for the two power frequency cycles before the operation. If a closing arc signal can be detected, the arc signal confirmation interval is formed by the arc signal extinction time and the arc signal channel calibration error, and intersected with the end crimping stage. If the primary circuit current is higher than the effective measurement lower limit for one consecutive power frequency cycle, the contact establishment interval is further narrowed by the starting boundary of the continuously measurable current-carrying interval. When the primary circuit current is lower than the effective measurement lower limit, the contact time of the contact is not estimated using this current. When the intersection of multiple confirmation intervals is empty, the effective identification of the contact boundary is set to invalid and the thermal state assignment of this version is terminated. When there is no arc signal and no measurable primary circuit current, the complete end crimping stage is retained as the contact establishment interval, and a single contact time is not forced.
[0039] For subsequent tripping operations that terminate this contact state version, the contact release phase is taken as the initial range of the contact termination interval, and this range is narrowed by using the available arc signal or the moment when the primary circuit current loses its continuous measurable state. The contact establishment interval and the contact termination interval together constitute the contact state version V_k. If no subsequent tripping has occurred in the current contact state, the contact termination interval remains empty. The current analysis cutoff time is only used as the temporary right endpoint for determining the closed current-carrying interval and the possible closed current-carrying interval, as well as the upper limit of the thermal excitation integral, and is not written into the contact termination interval.
[0040] The later boundary of the contact establishment interval to the earlier boundary of the contact termination interval is defined as the determined closed current-carrying interval, and the earlier boundary of the contact establishment interval to the later boundary of the contact termination interval is defined as the possible closed current-carrying interval. When the current contact state has not yet experienced subsequent tripping, both intervals use the current analysis cutoff time as the temporary right endpoint. The determined closed current-carrying interval corresponds to the time period under all permissible contact boundary interpretations that can be confirmed as closed. The possible closed current-carrying interval corresponds to the time period under at least one boundary interpretation that indicates the contact is closed. For a terminated contact state version, if the later boundary of the contact establishment interval is later than the earlier boundary of the contact termination interval, it means that it cannot be confirmed that this version was ever stably established, and this version will not enter the subsequent thermal state attribution processing.
[0041] Starting from the first absolute timestamp of the temperature record, the continuous sampling interval is divided according to the actual temperature sampling period. The temperature sampling period is preferably 0.5s to 5s, and in this embodiment, it is 1s. For each temperature sampling period, the integral of the square of the loop current on the intersection of the period and the determined closed current-carrying interval, and the integral on the intersection of the period and the possible closed current-carrying interval are calculated respectively. Then, the integral is divided by the temperature sampling period. The former result is used as the lower bound of the thermal excitation of the contact state version, and the latter result is used as the upper bound of the thermal excitation. Thus, the contact thermal excitation interval Q_k(n) of the nth temperature sampling period is formed, where n represents the temperature sampling period number, and the dimension of Q_k(n) is A².
[0042] Within the same temperature sampling period, the square of all primary loop currents is integrated and divided by the temperature sampling period to form the total current square excitation Q_T(n) not allocated according to the contact state version. Based on the primary loop current measurement error, integration discrepancy error, and maximum time-scale error, the total current square excitation determination interval Q_T^I(n) is formed. For each current sample value, the current value interval is formed with the sensor's allowable error. When this interval contains 0, the lower bound of the current square is 0; otherwise, the smaller of the two endpoint square values is taken. The upper bound of the current square is the larger of the two endpoint square values. Then, periodic integration is performed separately. The time-scale error is obtained by extending the left and right boundaries of the integration interval outwards by the maximum time-scale error. The integration discrepancy error is determined according to the maximum area difference that may be formed between adjacent current sampling points. Q_T(n) takes the midpoint of Q_T^I(n) and is only used to record the nominal excitation. Q_T^I(n) serves as the actual input for subsequent current conservation constraints. Q_T(n) and Q_T^I(n) are written as the same data item into the versioned electrothermal record. Subsequent steps only read this result and do not re-estimate the error.
[0043] Within a complete temperature sampling period far from the contact establishment and contact termination boundaries, the total current square excitation corresponding to 1000A and 1500A are 1000000A² and 2250000A², respectively. Within a temperature sampling period including the contact boundary, the lower bound of the contact thermal excitation only integrates to determine the closed current-carrying part, while the upper bound of the contact thermal excitation may integrate the closed current-carrying part. Therefore, the same period forms an interval rather than a single point value. The upper bound of the total current square excitation error is formed by the current channel calibration error, the integration discrete error, and the time scale error. The current temperature fitting result is not used for reverse adjustment.
[0044] During the final crimping phase of the closing operation, the square of the drive motor current is cumulatively integrated from the start of the phase to the current moment, and divided by the square integral of the current throughout the entire final crimping phase to obtain the cumulative electrical energy phase from 0 to 1. When the square integral of the current during the final crimping phase is 0, no cumulative electrical energy phase is formed, and the valid flag of the final crimping response is set to invalid. The two boundaries of the contact establishment interval are mapped to the phase axis respectively to obtain the contact phase interval.
[0045] The end-pressing weight is formed according to the contact phase interval: it is 0 before the contact phase interval, increases linearly from 0 to 1 within the contact phase interval, and remains 1 after the contact phase interval. The weight integral within the phase range of 0 to 1 is normalized. When the two boundaries of the contact phase interval coincide, a step weight with that boundary as the dividing line is adopted.
[0046] Select a real health response from the fixed health reference set as the fixed end-pressing reference response p^0. The selection rule is as follows: calculate the integral distance between each health response and the other health responses, and take the record with the smallest maximum distance; when there are parallel records, select the record with the earliest acquisition time. Each record in the fixed health reference set and the current record are calculated relative to the same p^0. Map the square of the drive motor current in the current end-pressing stage to the cumulative electrical energy phase axis, calculate the integral of its difference with p^0 under the end-pressing weight, and obtain the end-pressing response quantity. Current measurement error, acquisition time scale error and contact establishment interval boundary error enter the response quantity through interval propagation. It cannot be proven that the monotonic error variable is preferentially bisected according to the variable with the largest normalized interval width. When the current error interval width is not greater than the current sensor resolution, and the time and contact boundary error interval width is not greater than one drive motor current sampling cycle, stop subdivision, and summarize the minimum lower boundary and maximum upper boundary of the response quantity of all interval boxes to form the end-pressing response interval M_k, where the dimension of M_k is A².
[0047] The healthy closing records in the fixed healthy reference set use the same phase axis, the same fixed end-point crimping reference response p^0, and the same error propagation rule to form healthy reference response intervals. The minimum value of the lower boundary and the maximum value of the upper boundary of all healthy reference response intervals form the end-point crimping response allowable interval. The end-point crimping response allowable interval formed by the 6 healthy closing records in the supplementary numerical verification is -0.12A² to 0.14A². The closing record corresponding to the current contact state version forms M_k from -0.05A² to 0.07A². M_k is completely contained within this allowable interval, and the end-point mechanical response is judged to be without deviation.
[0048] S1 forms a versioned electrothermal record R_k, which includes the contact state version V_k, the contact thermal excitation interval sequence Q_k, the nominal sequence of total current square excitation Q_T, the determination interval sequence of total current square excitation Q_T^I, the end pressing response interval M_k, the temperature record T_k of the fixed temperature measurement area, and the item validity identifier ν_k. Q_k is composed of each Q_k(n) in the order of temperature sampling period. Q_T and Q_T^I adopt the same period order. ν_k includes the stage identification validity identifier, the contact boundary validity identifier, the primary loop current validity identifier, the time synchronization validity identifier, the temperature space validity identifier, and the end pressing response validity identifier. The preceding version record adopts the same structure and is denoted as R_(k-1). If the time synchronization error of any acquisition channel exceeds the maximum allowable value of 10ms set in this embodiment or cannot be verified by the unified time synchronization record, the time synchronization validity identifier is set to invalid and no cross-channel versioned electrothermal record is formed.
[0049] S2 first checks the target electrical phase, fixed temperature measurement area sequence, valid stage identification identifier, valid contact boundary identifier, valid primary loop current identifier, valid time synchronization identifier, and valid temperature space identifier in R_(k-1) and R_k. If the valid time synchronization identifier or valid temperature space identifier is invalid, a cross-version dynamic temperature model is not established. If the valid primary loop current identifier is invalid, only the temperature record is retained, and no contact thermal response is formed. The end-pressing response range is only transmitted with the versioned electrothermal record in S2 and does not participate in the temperature model fitting.
[0050] The last valid temperature sampling period before the formation of the preceding contact state is determined as the thermal analysis start period, and the temperature sampling period corresponding to the current analysis cutoff time is determined as the thermal analysis termination period. The right endpoint of the thermal analysis start period is no later than the earlier boundary of the preceding contact establishment interval, so that the model can accommodate the historical heat that existed before the formation of the preceding contact state. The current and temperature records within the thermal analysis period must be continuous. Data gaps that cannot be covered even if they exceed the maximum timescale error will not be interpolated, and an incomplete thermal analysis data flag will be set directly.
[0051] For the target temperature measurement area and the reference temperature measurement area, the area temperature is calculated using only the fixed pixel mask determined during installation and calibration. If there is a viewpoint drift, occlusion, or the fixed pixel mask cannot correspond to the same shell position at any sampling moment, the valid temperature space identifier is set to invalid. The ambient reference temperature at the same moment is subtracted from the temperature of the two areas respectively to form a 2D measured relative temperature difference vector, in which the component of the target temperature measurement area is arranged first and the component of the reference temperature measurement area is arranged second. The difference between the target temperature measurement area and the reference temperature measurement area is obtained only by subtracting the above two components. It is used to verify whether the local temperature rise changes together with the reference area. It is not used as a third independent model input to avoid the same temperature information being counted repeatedly.
[0052] After the pre-tripping of the supplementary numerical verification condition, the primary circuit current dropped to 0, and the temperature in the target temperature measurement area still changed along the current interruption propagation trajectory corresponding to the previous 60μΩ calibration state. No localized temperature rise of the same magnitude appeared in the reference temperature measurement area. This result is consistent with... Figure 2 The localized high-temperature distribution on the outer casing directly above the contact indicates that the previous thermal state must be retained after the circuit is opened.
[0053] Normal electrical heating calibration records use current data of 1000A, 1500A, 2000A, and 2500A to form the response range between the square of the primary circuit current and the temperature difference of the casing under normal contact conditions. Equivalent incremental resistance heating calibration records use states of 0μΩ, 20μΩ, 40μΩ, 60μΩ, 80μΩ, and 100μΩ to form the response range between the local additional heat source and the temperature difference of the casing. For the same GIS structure, the same set of thermal decay time constants is used to describe the natural heat dissipation process of the conductor, shielding structure, and casing within a given environmental range. Normal distributed heating and local incremental heating of contacts use independent input temperature gains. If there is no common thermal decay parameter domain that simultaneously covers the two types of calibration records, a common thermal mode model is not used, and the thermal model is marked as invalid.
[0054] The candidate thermal mode orders are preferably increased from low to high within the range of 1 to 4. For each candidate order, the lower bound of each time constant is set to 3 temperature sampling periods, and the upper bound is set to the effective duration of offline heating and interrupted cooling records, with the time constants constrained to increase sequentially. The temperature gain of normal electrical heating and equivalent incremental resistance heating are both limited to non-negative values. An interval branching method is used to search for the time constant and temperature gain: the temperature envelope generated by each parameter interval box for all calibration records is calculated, and any record that falls outside the envelope at any sampling time is excluded from the interval box. The retained interval boxes constitute the candidate thermal parameter domain for that order. In the supplementary numerical verification of this embodiment, the 3rd order model first passes the calibration and leave-one-out verification of normal electrical heating, equivalent incremental resistance heating, and interrupted cooling records simultaneously, so the 3rd order is preferred; the 1st and 2nd orders fail to cover the full range of temperature parameters. Throughout the entire thermal response phase, the 4th-order model is no longer considered a priority model. If the 4th-order model still cannot form a non-empty candidate thermal parameter domain, the thermal model is marked as invalid. The retention coefficient of the j-th thermal mode is obtained by dividing the temperature sampling period by the j-th time constant and taking the negative exponent, where j represents the thermal mode number. The thermal mode state of each temperature sampling period is obtained by multiplying the state of the previous period by the retention coefficient, plus 1 minus the product of the retention coefficient and the square excitation of the actual current in that period. After using this discrete recursion, the dimension of the thermal mode state corresponding to normal electrical heating is maintained as A², and the dimension of the temperature gain of normal electrical heating is ℃ / A². The dimension of the thermal mode state of unit equivalent incremental resistance is also A², and the dimension of the temperature gain of equivalent incremental resistance heating is ℃ / (μΩ·A²). The combination of each mode forms a unit equivalent incremental resistance temperature response sensitivity with the dimension of ℃ / μΩ.
[0055] The initial interval and branching rules for the thermal parameter domain search are further defined as follows: For the j-th thermal mode, the initial interval of the time constant is taken as 3 temperature sampling periods to the effective duration of the offline calibration record, and each time constant is strictly increased within each interval box; For the normal electrical heating temperature gain and equivalent incremental resistance heating temperature gain corresponding to the i-th fixed temperature measurement area, the lower bound is 0, and the upper bound is formed according to the offline calibration record: Under the given time constant interval box, the minimum positive lower bound of the corresponding mode state in all effective calibration sampling points is calculated, and the maximum allowable temperature rise upper bound of the temperature measurement area is divided by the minimum positive lower bound to obtain the conservative initial upper bound of the temperature gain; When the lower bound of the mode state contains 0 and cannot form a finite upper bound, the corresponding time constant interval is bisected first, and the temperature gain is not set as an unbounded interval. Each parameter interval box first checks the time constant order, the non-negativity of the temperature gain, and the calibration condition coverage. Range; if any condition is not met, it is directly excluded. Then, all heating and cooling records are propagated one by one. If the measured temperature difference and its offline calibration error interval at any sampling time do not intersect with the predicted temperature envelope, the interval box is excluded. For the interval boxes that are not excluded, one branch variable is temporarily shrunk to its midpoint. The reduction of the maximum predicted temperature envelope width of all calibration records relative to the previous shrinkage is calculated. The variable with the largest reduction is determined as the bisection variable for this time. If the reduction is the same, it is determined in ascending order of time constant, normal electrical heating temperature gain, equivalent incremental resistance heating temperature gain, and mode number. When the interval box can cover all calibration records and all leave-one verification records, and the contribution width of each variable interval to any predicted temperature envelope does not exceed the upper limit of the offline calibration error of the corresponding temperature measurement position, the branching stops. The union of all retained interval boxes constitutes the candidate thermal parameter domain of this order.
[0056] To perform repeatable forward and reverse calculations on all data, three nominal thermal time constants of 45 s, 180 s, and 900 s were used for supplementary numerical verification. Each time constant and each temperature gain participated in the joint solution within a 2% range above and below the nominal value. The three normal electrical heating temperature gains for the target temperature measurement region were 4.8 × 10^(-7), 3.6 × 10^(-7), and 2.8 × 10^(-7) °C / A², respectively, while those for the reference temperature measurement region were 4.3 × 10^(-7), 3.2 × 10^(-7), and 2.5 × 10^(-7). The three equivalent incremental resistance heating temperature gains of the target temperature measurement area are 5.0×10^(-8), 4.0×10^(-8), and 3.0×10^(-8)℃ / (μΩ·A²), respectively, while those of the reference temperature measurement area are 0.8×10^(-8), 0.7×10^(-8), and 0.5×10^(-8)℃ / (μΩ·A²), respectively. Under the conditions of an ambient temperature of 20℃ and a stable current of 2500A, the steady-state temperature of the target temperature measurement area in the 0μΩ state is approximately 27.0℃, and in the 100μΩ state it is approximately 102.0℃. Figure 2 and Figure 3 The values of approximately 27.4°C and approximately 103°C shown are of the same order of magnitude. This set of values is only used to supplement the numerical verification. The online calculation of the target device still uses the thermal parameter domain formed by its offline calibration.
[0057] The upper bound of the offline calibration error is formed by adding the upper bounds of the infrared temperature measurement allowable error, the ambient reference temperature measurement error, and the corresponding error of the fixed temperature measurement area. The corresponding error of the fixed temperature measurement area is determined by repeated image registration after 3 to 10 camera resets; in this embodiment, 5 times is used. Repeated registration is performed under stable conditions where the equipment temperature change does not exceed the allowable error of the ambient temperature sensor. Each time, the same outer shell mark point is used to complete the registration, and the maximum difference between the temperatures of the areas obtained by fixing the pixel mask after the same static temperature field is used as the corresponding error. The two different error calibrators of sensor reinstallation and software registration are not mixed. Normal electrical heating calibration and equivalent incremental resistance heating calibration are performed every Each calibration point preferably includes 3 to 10 independent records. In this embodiment, each calibration point has 5 records. Each record includes a complete heating segment and a flow interruption cooling segment. If any calibration point has fewer than 3 records or lacks a complete heating segment or flow interruption cooling segment, the calibration point will not be included in the parameter domain formation and cannot be replaced by interpolation from adjacent points. During leave-one verification, one complete record is excluded each time, and the remaining records from the same calibration point and other calibration points are used to re-form the candidate thermal parameter domain. It is required that the complete heating and cooling trajectories of the excluded records fall within the temperature envelope generated by the parameter domain. The candidate thermal parameter will only enter the thermal parameter domain when all calibration records and all leave-one verification records are covered by the error range.
[0058] The effective thermal mode order is the candidate order that first passes all calibration constraints and leave-one-out verification. The thermal parameter domain is jointly defined by the 1000A to 2500A current record, the 0μΩ to 100μΩ equivalent incremental resistance record, and the interrupted cooling record, and remains fixed during online processing. When the ambient temperature or current of the working condition to be analyzed exceeds -10℃ to 30℃ or 1000A to 2500A, or when the equivalent incremental resistance required for joint solution exceeds the 0μΩ to 100μΩ calculation domain, the current record is marked as outside the thermal model domain.
[0059] In the initial cycle of thermal analysis, the initial range of normal thermal modes is formed based on the offline calibration records. The difference between the initial ambient temperature and the current ambient reference temperature does not exceed the allowable error of the ambient temperature sensor. When the current before the start corresponds to the same calibration current point, the minimum and maximum states of each thermal mode recorded are used. When there are no completely matching records, the union of adjacent calibration record state intervals covering the current ambient temperature and current is used. Extrapolation is not performed on the operating conditions outside the calibration range. For each subsequent temperature sampling cycle, the thermal mode state of the previous cycle is multiplied by the corresponding retention coefficient, and then 1 is added minus the product of the retention coefficient and the actual total current square excitation candidate value in Q_T^I(n) of this cycle. The thermal mode state is combined with the normal electrical heating temperature gain of the target temperature measurement area and the reference temperature measurement area to form the normal electrical heating temperature difference of this cycle. The temperature difference generated by all allowable thermal parameters, initial state and actual total excitation candidate value constitutes the reachable set of normal electrical heating baseline.
[0060] For each fixed temperature measurement area, the minimum and maximum values of the normal electrical heating baseline set in that area are obtained, and the midpoint between the two is taken as the unique baseline center. Half of the difference between the two is taken as the baseline uncertainty half-width. The midpoint of this processing is uniquely determined by the upper and lower boundaries of the interval. The average value or frequency of occurrence of the current sample is not used. The measured relative temperature difference vector is subtracted from the baseline center to obtain the additional temperature rise sequence z(n). The dimensions of each component of z(n) are ℃. The additional temperature rise is not directly interpreted as contact anomaly. It still includes the residual heat of the previous contact, the current equivalent incremental resistance heating, the historical initial thermal state, the remaining environmental disturbances, and the measurement error.
[0061] The measurement and baseline error upper bound for each fixed temperature measurement area consists of the infrared thermometry allowable error, the ambient reference temperature measurement error, the corresponding error of the fixed temperature measurement area, and the baseline uncertainty half-width. The upper bound of the remaining temperature perturbation is taken as the maximum uncovered difference between the predicted temperature difference envelope of all excluded records in the leave-one-out verification and the measured temperature difference. Each time this difference is calculated, the excluded records do not participate in the formation of the corresponding candidate thermal parameter domain. Figure 3 The shell temperature inversion error of no more than 3.5℃ shown is only used to verify the error magnitude. The actual error boundary in this embodiment is formed by the verification record left by the target device each time, and 3.5℃ is not directly used as the online tolerance.
[0062] The supplementary numerical verification sets the half-width of the determination error for the target temperature measurement area and the reference temperature measurement area to 0.20℃ and 0.08℃, respectively. When generating the temperature sequence, determination perturbations with amplitudes not exceeding 0.03℃ and 0.015℃ are superimposed, respectively. The half-width of the error is greater than the amplitude of the superimposed perturbation, which is used to simultaneously accommodate parameter range, baseline error and numerical discrete error. This value is only used for this group of verifications and is not used as a fixed error threshold for the online processing of the target device.
[0063] For each temperature sampling cycle, the preceding contact thermal excitation interval, the current contact thermal excitation interval, and the total current square excitation determination interval Q_T^I(n) are read from S1. The actual total current square excitation q_T(n) for this cycle is set so that it is located in Q_T^I(n). The preceding actual thermal excitation q_(k-1)(n) is located in the preceding contact thermal excitation interval, and the current actual thermal excitation q_k(n) is located in the current contact thermal excitation interval, satisfying q_(k-1)(n)+q_k(n)=q_T(n). This forms the joint thermal excitation set for this cycle. This conservation relationship prevents the same primary loop current from being completely allocated to both the preceding and current contact state versions at the same time, and also avoids mistaking the nominal value of the total current square excitation as the error-free actual input.
[0064] A set of joint thermal excitation trajectories is formed by the set of joint thermal excitations from all temperature sampling periods. Each trajectory is jointly encoded by the same set of contact establishment boundaries, contact termination boundaries, and the actual total current square excitation q_T(n) and the previous thermal excitation q_(k-1)(n) of each temperature sampling period. The current thermal excitation is obtained by subtracting the previous thermal excitation from q_T(n). Therefore, each period only retains two constrained variables: the actual total excitation and the previous allocation. Each joint thermal excitation trajectory must originate from the same set of contact boundaries. It is not allowed to select favorable boundary combinations for splicing between different temperature sampling periods. If neither of the two contact state versions in a certain period can be closed and Q_T^I(n) does not contain 0, or if the previous and current thermal excitation intervals cannot cover any actual total excitation in Q_T^I(n) under the conservation conditions, then a current attribution conflict flag is set, and this period is not included in the thermal response sensitivity calculation.
[0065] Within the temperature sampling period when the preceding contact state has been determined to terminate and the current contact state is within the determined closed current-carrying interval, the preceding thermal excitation is fixed at 0, and the current thermal excitation is directly equal to the square excitation of the actual total current. Only the boundary period covered by the contact establishment interval or the contact termination interval retains the preceding and current assigned variables. This process reduces unnecessary free variables in the stable current-carrying period, while allowing the boundary uncertainty to continue to propagate backward.
[0066] There may be historical additional heat that has not yet decayed before the start of the thermal analysis cycle. S2 uses the additional thermal mode state consistent with the effective thermal mode order to represent this historical heat. The lower bound of each state is 0, and the upper bound is the maximum upper bound of the state that can be formed within the maximum calibration current duration of all equivalent incremental resistance heating calibration records. In this embodiment, it corresponds to 2500A, 100μΩ and the longest calibration current duration. This upper bound only covers the offline calibration range. If the running records that can be obtained before the start of the analysis show that the current or continuous current duration exceeds this range, it is directly marked as a thermal model. Outside the thermal model domain; the equivalent incremental resistance, which cannot be directly observed in actual operation, is not used as the entry input for this item. Its impact is verified by whether the initial additional temperature rise can be covered by the set of historical states. The dimension of the historical additional thermal mode state is μΩ·A², which is multiplied by the heating temperature gain of the equivalent incremental resistance to form the historical additional thermal response with the dimension of ℃. For each set of allowable thermal parameters, the initial temperature difference generated by the combination of historical additional thermal states should cover the initial additional temperature rise within the range of measurement and baseline error and remaining temperature disturbance. If all thermal parameters fail to meet the requirements, the record will be marked as outside the thermal model domain.
[0067] For the previous contact state version and the current contact state version, a unit equivalent incremental resistance thermal state consistent with the effective thermal mode order is established. In each temperature sampling period, the unit thermal state of the previous period is first multiplied by the corresponding holding coefficient, and then 1 is added and the product of the holding coefficient and the actual thermal excitation belonging to the contact state version in the same joint thermal excitation trajectory is subtracted. The unit thermal states are combined according to the equivalent incremental resistance heating temperature gain of the target temperature measurement area and the reference temperature measurement area to obtain the temperature response sensitivity of the unit equivalent incremental resistance of the version.
[0068] The preceding sensitivity and the current sensitivity must be generated by the same thermal parameters and the same joint thermal excitation trajectory, and stored as a pair. After the preceding contact state version is opened, its new thermal excitation becomes 0, but the unit thermal mode state accumulated before the opening continues to decay according to the holding coefficient. Therefore, the preceding sensitivity can still be non-zero after the current contact state is established. Before the current contact state is established, its unit thermal mode state has no current thermal excitation input. After entering the closed current carrying phase, the current sensitivity is gradually formed with the square excitation of the primary circuit current.
[0069] The residual thermal response set is formed by the natural decay of the historical additional thermal state, the residual temperature perturbation, and the measurement and baseline errors. The additional temperature rise in each temperature sampling period is composed of the residual thermal response, the product of the previous equivalent incremental resistance and the previous sensitivity, and the product of the current equivalent incremental resistance and the current sensitivity. S2 outputs this joint response relationship for S3 to perform identification verification.
[0070] S3 forms a candidate attribution window within the defined closed current-carrying interval of the current contact state version. The temperature sampling period must fall completely within this interval. If the valid temperature space identifier or the valid primary loop current identifier is invalid, or if the current record is marked as outside the thermal model domain, the temperature evidence is marked as undecided attribution.
[0071] For each defined current-carrying cycle, the total current square excitation defined interval Q_T^I(n) formed by S1 is directly read. If the Q_T^I intervals of two adjacent cycles do not intersect, it means that the square change of the load current has exceeded the range that can be explained by the measurement error, integration error and time scale error. The next cycle is then defined as the electrical excitation change cycle. The first defined current-carrying cycle after the current contact state is established is also defined as the electrical excitation change cycle as long as Q_T^I(n) does not contain 0. This rule does not recalculate the current error, nor does it set an additional empirical amplitude threshold.
[0072] Each electrical excitation change cycle is used as the response start cycle; if the previous temperature sampling cycle still belongs to the current determined current-carrying cycle set, it is used as the preceding reference cycle. The shortest duration of the candidate window is taken as the larger of two times: the first time is the moment when the lower boundary of the 20μΩ equivalent incremental resistance response increment of the current contact unit generated by the load change first exceeds the half-width of the determination error of the target temperature measurement area; the second time is the moment corresponding to the upper boundary of the shortest thermal time constant in the thermal parameter domain. For the first time, the current response when the current continues to run before the change and the current response when the change is applied are calculated respectively. The difference between the two forms the reachable set of response increments. If the response increment cannot exceed the half-width of the determination error within the longest allowable duration, no candidate window is formed, and the temperature evidence is marked as belonging to an unclaimed category. The longest duration of the candidate window is determined by taking the shortest value among the current determined closed current continuous period, the temperature continuous recording period, and the effective duration of offline calibration. Starting from the shortest duration, candidate windows are formed sequentially with a step size of 1 temperature sampling period. Each window contains at least 2 continuous temperature sampling periods. In this embodiment, the 20μΩ response increment exceeds the 0.20℃ half-width of the determination error after 7s. The upper boundary of the shortest thermal time constant is 45.9s, so the lower limit of the candidate window is rounded up to 46s. After the load change, the continuous recording lasts for 300s, so the candidate duration range is 46s to 300s. The additional temperature rise fitting results are not read during the window formation process. The supplementary numerical verification condition starts from the period when the effective value of the primary loop current is adjusted from 1000A to 1500A to form the candidate window.
[0073] For each candidate window, the preceding sensitivity and the current sensitivity corresponding to the same thermal parameter and the same joint thermal excitation trajectory are obtained from the paired sensitivities saved in S2. All sensitivity vectors in the window are vertically concatenated in the order of priority for temperature sampling period and second priority for fixed temperature measurement area. The preceding sensitivity is used as the first column of the paired thermal response matrix, and the current sensitivity is used as the second column. The column order remains fixed in subsequent solutions. All sampling periods in the window use the same thermal parameter and the same joint thermal excitation trajectory. Reselection is prohibited for each period.
[0074] The historical additional thermal state response, residual temperature perturbation, measurement and baseline error within the window are concatenated in the same order to form a set of residual thermal response trajectories within the window. Each trajectory maintains the same thermal parameters and the same initial additional thermal state source throughout the entire window. For each scalar position in the window vector, the minimum and maximum values of the residual thermal response are obtained respectively, and half of the difference between the two is used as the half-width of the error. If the half-width is less than the minimum temperature resolution of the corresponding temperature measurement position, the minimum temperature resolution is used. Then, a diagonal scaling matrix is formed by the reciprocal of the error half-width of each position, so that the temperature difference between different temperature measurement areas and different sampling times is converted into a dimensionless difference relative to its own determined error range. This scaling matrix does not represent statistical weights based on sample frequency.
[0075] Within the equivalent incremental resistance calculation domain from 0 μΩ to 100 μΩ, the preceding and current equivalent incremental resistance candidate values are combined into a two-dimensional candidate vector. For any candidate vector, the allowable thermal parameters, allowable joint thermal excitation trajectory, and allowable window residual thermal response are traversed to form the set of all window temperature responses that the candidate vector can generate. When at least one of the preceding or current components of two candidate vectors differs by 20 μΩ, they are regarded as a pair of candidate interpretations that need to be distinguished according to adjacent equivalent incremental resistance calibration points.
[0076] For each pair of candidate interpretations, allowable members are taken from the window temperature response sets of both, and the Euclidean distance between the two members is calculated after scaling by the determination error. Then, the minimum distance is taken for all candidate interpretations that need to be distinguished, all thermal parameters, all joint thermal excitation trajectories, and all residual thermal responses to obtain the conservative identifiable margin of the candidate window. This margin allows two competing interpretations to use different allowable thermal parameters and different allowable excitation trajectories. Therefore, the test object is whether the two complete temperature response sets intersect, rather than just whether the two columns of sensitivity are linearly independent under a fixed model.
[0077] The conservative identifiable margin is obtained using an interval branching method. Two sets of candidate incremental resistance vectors, two sets of thermal parameters, two sets of joint thermal excitation trajectories, and two sets of residual thermal responses are each represented as an initial interval box. A constraint is added that at least one incremental resistance component differs by no less than 20 μΩ. For each pair of interval boxes, the lower bound of the scaled Euclidean distance of the temperature difference is calculated. Interval boxes with a lower bound greater than 0 retain this positive lower bound; interval boxes with a lower bound containing 0 are bisected along the variable that causes the widest outer bound of the temperature difference. For interval boxes with a lower bound containing 0, the constraints of equal temperature responses, thermal excitation conservation, and residual thermal response membership are simultaneously applied. The Newton contraction of the row interval is performed. If the contraction result is empty, the interval box is excluded. After contraction, specific variable points are obtained and all sampling times are directly substituted to satisfy the equality of the two sets of temperature responses and the set membership constraints. When the two response sets intersect, the lower bound of the margin of the window is set to 0. When the lower bound of the distance of all remaining interval boxes is greater than 0, the smallest lower bound is taken as the provable lower bound of the margin. If separation or intersection cannot be proven after reaching the 5μΩ incremental resistance interval width and the response envelope accuracy determined by the temperature measurement resolution, the window is recorded as identifiable but unresolved and treated as a margin lower bound that cannot be proven to be greater than 0.
[0078] In this embodiment, the interval Newton contraction is implemented using a unified engineering approach. The midpoints of all variables within the interval to be contracted are used to form a midpoint vector. The temperature response equation, the prior and current thermal excitation conservation equations, and the residuals corresponding to the boundary of the set members are calculated periodically according to the complete candidate window. The temperature residuals are dimensionless according to the half-width of the determination error of the corresponding temperature measurement location, and the current square conservation residuals are dimensionless according to the half-width of the total current square excitation error. This avoids changes in the contraction and branching order due to different dimensions and numerical scales. The first-order partial derivative intervals of the residuals with respect to each variable are calculated on the current interval box to form an interval Jacobian matrix. The interval linear inclusion relationship formed by the midpoint residuals and the interval Jacobian matrix is solved. The resulting variable envelope is intersected with the original interval box. If the intersection result is empty, it proves that there are no members in the current interval box that satisfy the equality constraints and they are excluded. If the intersection result is a proper subset of the original interval box... The original interval box is replaced with the intersection and the contraction is repeated. If the interval Jacobian matrix contains an invertible case or the interval width does not decrease after contraction, it is not determined that there is no solution. Instead, the bisection continues according to the variable that contributes the most to the temperature envelope. If the interval Newton image is strictly contained inside the current interval box, the interval Jacobian matrix remains invertible within the interval box, and the thermal excitation boundary, residual response boundary, and variable value boundary are all satisfied, it is proven that there is at least one member in the interval box that satisfies the integrity constraint. The midpoint of the interval or the approximate point obtained by numerical optimization is only used for positioning and is not used as a proof of the non-emptiness of the set. The contraction and bisection continue until the equivalent incremental resistance interval reaches the predetermined stopping width, and the temperature response envelope width caused by the residual variable does not exceed the corresponding half width of the determination error. If the existence or exclusion cannot be proven even after reaching the stopping condition, the identifiable unresolved calculation or joint solution remains unresolved.
[0079] When the identifiable margin is provably lower bound that is strictly greater than 0, it means that the allowable temperature response sets corresponding to any two candidate incremental resistance vectors that differ by at least 20 μΩ do not intersect. The window has the state attribution condition at the resolution corresponding to the interval between adjacent equivalent incremental resistance calibration points. When the identifiable margin is 0 or cannot be proven to be greater than 0, the window does not meet the identifiable admission condition.
[0080] The identifiable margin of each candidate window is calculated sequentially. Windows with a provable lower bound greater than 0 are retained, and the window with the largest lower bound of the margin is selected first. When the lower bounds of the margins of multiple windows are the same or cannot be strictly ordered, the windows with shorter durations and earlier start periods are selected in sequence. S3 outputs the selected assigned window, the set of paired thermal response matrices, the set of remaining thermal response trajectories of the window, the determined error scaling matrix, the identifiable margin, the identifiable admission identifier μ_k, the specified resolution of 20μΩ, and the joint computational domain.
[0081] Supplementary numerical verification was performed on all candidate windows formed with a step size of 1 second within the range of 46s to 300s, and the aforementioned interval separation calculation was performed. The identifiable margin lower bound of the 60s window could not be proven to be greater than 0, while the 120s and 300s windows could both be proven to be greater than 0. Moreover, the identifiable lower bound of the 300s window was larger. Therefore, the 300s window was selected according to the established order of larger margin lower bound, shorter duration, and earlier start period. At the same time, the reference period of 1 second before the change was retained. This selection result is only applicable to the thermal parameter domain, error boundary, and load change conditions of this group, and should not be used as the basis for other equipment to use the 300s window.
[0082] When the identifiable margin lower bound is greater than 0, μ_k is set to 1 and the process proceeds to S4; otherwise, μ_k is set to 0 and the temperature evidence is marked as pending. μ_k indicates whether the current measurement conditions can support the joint solution at the specified resolution, and does not indicate the result of the contact thermal state determination.
[0083] When μ_k is 1, S4 follows the temperature sampling cycle order and fixed temperature measurement area order determined by S3, and vertically splices the additional temperature rise in the selected window into the measured additional temperature rise vector. If there is a data gap in the window, the fixed temperature measurement area order changes, or the valid temperature space identifier is set to invalid, the joint solution is terminated.
[0084] The joint solution variables include the candidate values of the preceding equivalent incremental resistance, the candidate values of the current equivalent incremental resistance, thermal parameters, the same set of contact boundaries, the actual total current square excitation for each temperature sampling period, the assigned value of the preceding thermal excitation, the historical additional thermal state for the initial period of thermal analysis, and the variables of residual temperature disturbance and measurement error. The actual total current square excitation is located at Q_T^I(n), and the current thermal excitation is obtained by subtracting the preceding thermal excitation from the actual total excitation. Both the preceding and current equivalent incremental resistances are limited to 0μΩ to 100μΩ. For any candidate interpretation, a pair of matrices generated by the same thermal parameter and the same joint thermal excitation trajectory are selected from the set of paired thermal response matrices. The historical additional thermal state uses the same initial state and propagates continuously with the same thermal parameter throughout the entire window. The residual temperature disturbance and measurement error are located at the boundaries formed by S2. The two columns of responses are multiplied by the candidate values of the preceding and current equivalent incremental resistance, respectively, and together with the continuously propagated historical additional thermal response, residual temperature disturbance, and measurement error, they jointly explain the measured additional temperature rise. The candidate interpretations that satisfy the constraints of all times and the two fixed temperature measurement areas constitute the joint feasible set.
[0085] This embodiment employs a constrained interval branching solution. First, the preceding and current equivalent incremental resistance calculation domains, thermal parameter domains, contact boundaries, the squared total current excitation of each cycle, the preceding thermal excitation allocation value, the historical additional thermal initial state, and the remaining perturbation variables are represented as initial interval boxes. The current thermal excitation is calculated using the current conservation relationship. Each interval box is propagated cycle-by-cycle according to the same thermal parameters and the same initial historical state, forming a complete window temperature response envelope. If the measured additional temperature rise falls outside the envelope at any scalar position, that interval box is excluded. For interval boxes that are not excluded, the aforementioned unified interval is applied. The interval Newton contraction rule handles temperature response equations, thermal excitation conservation, and the boundaries of each variable. Only when the interval Newton image is strictly contained within the current interval box and the boundaries of all set members are satisfied, is the corresponding interval box marked as having a proven member. This is used to prove that the joint feasible set is not empty. The periodic substitution results of the interval midpoint or approximate candidate point are only used for numerical verification and do not replace the interval existence proof. The interval box that has not been excluded continues to be bisected along the variable that contributes the most to the temperature outer wrap and is always preserved as a conservative outer wrap of all possible feasible members. It is not deleted because a single feasible member has been found.
[0086] The preferred stopping width of the equivalent incremental resistance interval for interval branching is 1μΩ to 10μΩ. In this embodiment, 5μΩ is used, which is 1 / 4 of the specified resolution of 20μΩ. 20μΩ is only used for the state resolution requirement of S3 and the final projection width admission of S4, and is not used as the stopping width of the interval branch. After the equivalent incremental resistance interval width is shrunk to 5μΩ, the thermal parameter and temperature gain intervals are shrunk further until the predicted temperature envelope width caused by them does not exceed the corresponding offline error boundary. If the measured additional temperature rise cannot be covered at any time or in any fixed temperature measurement area, the corresponding interval box is excluded. A single temperature peak value is not used as an independent judgment criterion.
[0087] When at least one specific candidate interpretation satisfying all constraints is verified, the joint feasible set is proven to be non-empty. When all interval boxes are excluded, they are marked as thermally incompatible. If, after reaching the equivalent incremental resistance interval stopping width of 5 μΩ and the corresponding temperature enclosing requirement, there are still interval boxes that cannot be proven to be unsolvable, the interval box is retained in the conservative enclosing of the joint feasible set. If no feasible member has been found, it is marked as joint solution unresolved. Interval boxes that have not been excluded but have not been verified by members cannot be used as feasible members on their own, but cannot be deleted from the conservative enclosing.
[0088] When the joint feasible set has been proven to be non-empty, all unexcluded shrinkage interval boxes are projected onto the preceding incremental resistance axis and the current incremental resistance axis, respectively. The minimum lower boundary and maximum upper boundary of the projection of each axis form the conservative outer bounding interval I_(k-1) of the preceding equivalent incremental resistance and the conservative outer bounding interval I_k of the current equivalent incremental resistance. The dimension of both intervals is μΩ. The conservative outer bounding interval represents the outer containment range of the projection of the true joint feasible set, which may include numerical intervals that have not yet been proven feasible but have not yet been excluded. The specific feasible members that have been verified are only used to prove that the set is non-empty and are not used to shrink the projection intervals separately. The joint feasible set and its interval box source are preserved. Any endpoint of the two projection intervals may not be regarded as the same candidate interpretation that can be freely combined.
[0089] Calculate the width of the previous and current conservative envelope intervals separately. If the width of any projected envelope reaches or exceeds 20 μΩ, it indicates that the existing solution envelope has not yet reached the state resolution capability specified in S3. The joint solution result is changed to unresolved, and contact thermal state confirmation is not performed. If both projected widths are less than 20 μΩ, the entire conservative envelope interval is compared with the offline equivalent incremental resistance judgment interval: if the entire conservative envelope interval is contained within the normal equivalent incremental resistance judgment interval, the contact thermal state of the corresponding version is confirmed to be normal; if the entire conservative envelope interval is contained within the abnormal equivalent incremental resistance judgment interval formed by independent fault preset and loop resistance verification, the contact thermal state is confirmed to be abnormal; if the conservative envelope interval only partially intersects with a certain judgment interval, intersects with two judgment intervals simultaneously, or is located between judgment intervals, it remains unresolved. The state determination only uses the complete interval inclusion relationship and does not rely on the midpoint of the interval or a single optimal point.
[0090] In the supplementary numerical verification, the parameter range of 2% above and below the nominal value was used for the three thermal time constants and six temperature gains. A joint solution was performed within a selected window of 300s. In the forward verification, the previous preset value was 60μΩ and the current preset value was 0μΩ. The conservative outer range of the previous equivalent incremental resistance was found to be 50.3μΩ to 69.1μΩ, with a range width of 18.8μΩ. The conservative outer range of the current equivalent incremental resistance was found to be 0μΩ to 1.2μΩ, with a range width of 1.2μΩ. The previous range was completely contained within the 50μΩ to 70μΩ abnormal equivalent incremental resistance judgment range, and the current range was completely contained within the 0μΩ to 10μΩ normal equivalent incremental resistance judgment range. Therefore, the previous version was judged to be in an abnormal contact thermal state, and the current version was judged to be in a normal contact thermal state. The 60μΩ and 0μΩ preset values were not read during the online solution process.
[0091] Read the current end-point crimping response interval M_k formed by S1. M_k only corresponds to the closing operation that forms the current contact state version V_k. The number of offline healthy closing records is preferably 3 to 10, and 6 are used in this embodiment. Each record should have the same equipment, closing direction, power supply voltage access conditions and measurement installation conditions as the current record. Each healthy record calculates the reference response interval relative to the same fixed end-point crimping reference response p^0, and forms the end-point crimping response allowable interval with the minimum value of the lower boundary and the maximum value of the upper boundary of all reference intervals. The fixed healthy reference set, p^0 and the end-point crimping response allowable interval remain unchanged during online processing, and the current M_k does not participate in its expansion.
[0092] When M_k is completely contained within the allowable range of end-crush response, the end-crush mechanical response is determined to be without deviation; when M_k does not intersect with the allowable range, it is determined to be a deviation; when the two partially intersect, it remains undecided. The supplementary numerical verification condition has M_k ranging from -0.05A² to 0.07A², which is completely contained within the allowable range of end-crush response from -0.12A² to 0.14A². Therefore, the end-crush mechanical response in the current version is determined to be without deviation.
[0093] The current equivalent incremental resistance conservative outer range and contact thermal state determination result, the current end crimping response range and end mechanical response determination result are jointly associated with the current contact state version V_k and written into the versioned state evaluation record of this version. The previous equivalent incremental resistance conservative outer range continues to be associated with V_(k-1). It should not be rewritten to V_k because the previous and current equivalent incremental resistances are solved in the same joint feasible set. The contact thermal state determination result and the end mechanical response determination result are saved separately and are not weighted averaged or canceled out.
[0094] The positive results of the supplementary numerical verification are as follows: the previous 60μΩ preset state yields a conservative outer bounding range of 50.3μΩ to 69.1μΩ, and the current 0μΩ preset state yields a conservative outer bounding range of 0μΩ to 1.2μΩ. The current end-press response range is included by the end-press response allowable range. The reverse verification maintains the same time boundary, load change, thermal parameter domain, and error boundary, only swapping the previous and current preset states to 0μΩ and 60μΩ, respectively. The joint solution yields the previous 0μΩ to 9.3μΩ and the current 57.4μΩ to 62.4μΩ. The previous state is included by the normal equivalent incremental resistance judgment range, and the current state is included by the abnormal equivalent incremental resistance judgment range. Neither the positive nor the reverse results swap the previous and current objects, and the width of each projection is less than the specified resolution of 20μΩ. The above preset values are only used for offline verification output and are not used as input for joint solution.
[0095] Counterexamples were performed for missing measurements, delays, conflicts, insufficient samples, and sensor anomalies. The results are shown in the table below. Each counterexample stopped subsequent contact thermal state confirmation after triggering the corresponding invalid, conflict, outside the thermal model domain, or unresolved conditions, and no contact thermal state confirmation result was generated.
[0096] Table 2, Verification Results of Forward / Reverse Attribution and Abnormal Operating Condition Blocking:
[0097] Verification conditions Input changes Actual calculation or comparison results Processing results Positive attribution The previous equivalent incremental resistance preset value was 60μΩ, and the current equivalent incremental resistance preset value is 0μΩ; the current A value is adjusted from 1000A to 1500A. Preceding current Ω from 50.3 μΩ to 69.1 μΩ; Current current Ω from 0 μΩ to 1.2 μΩ Previous contact thermal condition was abnormal, current contact thermal condition is normal Reverse attribution The previous equivalent incremental resistance was preset to 0 μΩ, and the current equivalent incremental resistance is preset to 60 μΩ; all other conditions remain unchanged. Preceding Ω 0 μΩ to 9.3 μΩ; Current Ω 57.4 μΩ to 62.4 μΩ Previous contact thermal condition was normal, current contact thermal condition is abnormal. Insufficient load change The primary circuit current was adjusted from 1000A to 1010A, with the allowable error being 1% of the reading. The two current squared intervals are approximately 980100A² to 1020100A² and 999800A² to 1040604A², respectively, and they overlap. If no electrical excitation cycle is formed, μ_k is set to 0. Time synchronization exceeded the limit The timescale error for any cross-channel is 15ms. Exceeding the maximum allowable value of 10ms in this embodiment No cross-channel versioned electrothermal records are generated. Temperature measurement missing One temperature sampling period is missing within the selected window and cannot be covered by timescale error. Window temperature vector discontinuity Terminate the joint solution and mark it as pending review. Contact Boundary Conflict The intersection of the end-pressing stage, the arc signal interval, and the continuous current-carrying interval is empty. Contact-established intervals cannot form non-empty sets. The contact boundary is invalid, and S2 is not entered. Insufficient sample size Only 2 fixed health reference records or only 2 necessary thermal marker records. Not meeting at least 3 of the admission requirements No stage reference or thermal parameter domain is formed Sensor or operating condition outside the domain The primary circuit current path is saturated, or the current carrying capacity is 2600A. The former cannot form Q_T^I, while the latter exceeds the calibration range of 1000A to 2500A. The primary loop current valid identifier is invalid or outside the thermal model domain.
[0098] When the identifiable admission identifier μ_k is 1, the joint feasible set has been proven to be non-empty, the current equivalent incremental resistance conservative outer range width is less than 20μΩ and the range is completely contained within an equivalent incremental resistance determination range, the current version record is added to the long-term thermal state trajectory. When the end-pressing response is valid and the end-mechanical response can be clearly determined, the current version record is added to the long-term mechanical response trajectory. The long-term trajectory record does not participate in the recalibration of the fixed health reference set, fixed end-pressing reference response, thermal parameter domain, joint calculation domain and equivalent incremental resistance determination range.
[0099] S4 outputs versioned status evaluation records and does not directly generate instructions for opening, closing, load adjustment, or switching sequence change.
[0100] When there are fewer than 3 fixed health reference records, only the original operation sequence is retained, and no automatic stage identification results, contact status versions, or end-point crimping responses are generated. When there are fewer than 3 equivalent incremental resistance heating calibration records at any necessary calibration point, no thermal parameter domain covering the calculation domain is generated. S3 and S4 do not perform temperature evidence attribution and equivalent incremental resistance solution. When the time synchronization error exceeds the allowable range, the time synchronization validity flag is set to invalid. When the temperature record is missing or the infrared viewpoint drifts, the temperature space validity flag is set to invalid. When the primary loop current channel is saturated or fails, the primary loop current validity flag is set to invalid.
[0101] If contact boundary confirmation information conflicts, the joint thermal excitation set is empty, the fixed temperature measurement area is mismatched, the candidate window cannot be formed, the identifiable calculation is unresolved, the joint solution is unresolved, or the operating conditions exceed the offline calibration range, the corresponding subsequent contact thermal state confirmation shall be stopped, and the record shall be associated with the corresponding contact state version as a record to be reviewed. Any record marked as invalid, conflicting, outside the thermal model domain, or unresolved shall not be added to the long-term state trajectory, nor shall it be written back to the fixed health reference set, thermal parameter domain, error boundary, or equivalent incremental resistance judgment interval.
[0102] In actual operation, versioned state evaluation records are only output when the input validity, identifiable access, proof of non-emptiness of the joint feasible set, and complete interval inclusion condition of S1 to S4 are all satisfied simultaneously. The 60μΩ and 0μΩ preset values used for supplementary numerical verification are not included in the online input, and the online results cannot be read in reverse or the preset values can be used.
[0103] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for evaluating the dual-parameter status of a GIS disconnector switch based on both electrothermal parameters, characterized in that: include: Acquire the drive motor current, primary circuit current, GIS enclosure temperature, and ambient reference temperature of the target electrical phase during continuous switching. The operation stages are divided according to the drive motor current. Adjacent contact state versions are formed by combining the primary circuit current boundary, and the closed current carrying interval, contact thermal excitation interval, total current square excitation determination interval and end crimping response interval are determined. Among them, the adjacent contact state version is the previous contact state version and the current contact state version; The specific operations for generating the contact state version, contact thermal excitation range, total current square excitation determination range, and end crimping response range include: Extract the final crimping stage of the closing operation and the contact release stage of the opening operation from the action stage, and form a continuously measurable range based on the primary circuit current. The contact establishment interval is formed by intersecting the end crimping stage with the continuously measurable interval. The contact release stage is formed by corresponding to the boundary where the primary circuit current exits the continuously measurable state. The contact state version is constituted by the contact establishment interval and the contact termination interval, and the closed current-carrying interval is defined. The square of the primary circuit current is integrated over the temperature sampling period to form the contact thermal excitation interval and the total current square excitation determination interval; a fixed health reference set is formed based on the health operation records of the same equipment, and a fixed end crimping reference response is determined from the fixed health reference set. The weighted integral of the square response of the drive motor current during the end crimping stage relative to the fixed end crimping reference response is calculated to form the end crimping response interval. The ambient temperature correction and normal electrical heating of the GIS shell temperature are separated by using a calibrated thermal parameter domain to form an additional temperature rise sequence; the additional thermal state at the end of the previous contact state version is continuously recursively deduced. Within each temperature sampling period, the actual total current square excitation is selected from the total current square excitation determination interval and allocated as the preceding thermal excitation and the current thermal excitation, such that the sum of the preceding thermal excitation and the current thermal excitation equals the actual total current square excitation; the thermal excitation allocation results of each temperature sampling period are combined into a joint thermal excitation trajectory according to the time sequence; the unit equivalent incremental resistance thermal response of adjacent contact state versions is recursively derived using the same thermal parameters and the same joint thermal excitation trajectory to form a set of paired temperature responses; The residual thermal response set is formed based on the additional thermal state decay response at the termination of the previous contact state version, the residual temperature perturbation, and the measurement and baseline error boundaries. The candidate assignment window is formed by determining the interval variation based on the total current square excitation. It is then spliced into the previous temperature response and the current temperature response in the temperature response set according to the temperature sampling cycle sequence and the fixed temperature measurement area sequence, and then scaled dimensionlessly according to the corresponding determination error half-width. The preceding equivalent incremental resistance and the current equivalent incremental resistance are combined into a candidate vector. Two candidate vectors with an incremental resistance difference of not less than the calibrated incremental resistance interval are combined into a candidate vector pair. The minimum scaling distance between the corresponding temperature response sets of the candidate vector pair is calculated using the paired temperature response sets to form an identifiable margin. The candidate vector pairs and the solution variables corresponding to the paired temperature response sets are represented as interval boxes. For interval boxes whose minimum scaling distance lower bound contains zero, interval Newton contraction and variable bisection are performed to exclude interval boxes that do not satisfy the temperature response constraints, current square excitation conservation constraints, and residual thermal response member constraints. Candidate assignment windows with identifiable margin lower bounds greater than zero are retained. Assignment windows are determined in the order of identifiable margin lower bound from high to low, window duration from short to long, and window start period from early to late. Within the home window, using the additional temperature rise sequence constraint paired temperature response sets, the solution variables corresponding to the preceding equivalent incremental resistance, the current equivalent incremental resistance, and the paired temperature response sets are represented as interval boxes. The temperature response is propagated periodically using the same thermal parameters and the same joint thermal excitation trajectory. Interval boxes whose temperature response envelope cannot cover the additional temperature rise sequence are excluded. For the retained interval boxes, interval Newton contraction and variable bisection are performed. The parameter combinations that satisfy the temperature response constraints within the home window constitute the joint feasible set of equivalent incremental resistances. The interval boxes that were not excluded are projected onto the preceding equivalent incremental resistance axis and the current equivalent incremental resistance axis, respectively, to form the corresponding equivalent incremental resistance conservative envelope intervals. The equivalent incremental resistance conservative envelope intervals are compared with the equivalent incremental resistance judgment intervals, and the end-pressing response intervals are compared with the end-pressing response allowable intervals to form a versioned state evaluation record.
2. The method for evaluating the dual-parameter status of a GIS disconnector switch based on electrothermal parameters according to claim 1, characterized in that, The specific steps for dividing an action into phases include: The action judgment boundary is formed based on the drive motor current in the non-operational interval before operation, and the operation window is extracted based on the sliding root mean square value of the drive motor current; the sliding integral of the square of the drive motor current in the operation window is performed to form the sliding electrical energy trajectory and it is dimensionless. A fixed health reference set is formed based on the health operation records of the same device. Boundary combinations that satisfy the constraints of stage sequence and stage duration are enumerated. The trajectory curvature and stage shape distance corresponding to the boundary combinations are calculated to determine the action stage.
3. The method for evaluating the dual-parameter electrothermal status of a GIS disconnector switch according to claim 1, characterized in that, The specific operations for forming the calibration thermal parameter domain and the additional temperature rise sequence include: Using offline calibration records covering both flow-through heating and flow-interruption cooling conditions, range identification and leave-one-out verification are performed on the thermal modal parameters to form a calibration thermal parameter domain; Based on the calibrated thermal parameter domain and the total current square excitation, the thermal mode state corresponding to the normal electrical heating is determined by interval recursion, forming the normal electrical heating baseline; the temperature of the fixed temperature measurement area is extracted from the GIS shell temperature, and the difference between the fixed temperature measurement area temperature and the ambient reference temperature is calculated to form the measured relative temperature difference. The normal electrical heating baseline is subtracted from the measured relative temperature difference to form the additional temperature rise sequence.
4. The method for evaluating the dual-parameter status of a GIS disconnector switch based on electrothermal parameters according to claim 3, characterized in that, The specific steps for forming a candidate attribution window include: The error half-width is determined by forming the corresponding temperature sampling period and fixed temperature measurement area based on the remaining thermal response set; the total current square excitation determination interval of adjacent temperature sampling periods is compared within the determined closed current carrying interval; when adjacent total current square excitation determination intervals do not intersect, the temperature sampling period with the later timestamp is set as the electrical excitation change period. The time required for the lower boundary of the temperature response increment corresponding to the calibrated incremental resistance interval to first exceed the half-width of the determined error is compared with the upper boundary of the shortest thermal time constant in the calibrated thermal parameter domain, and the larger value is used as the lower limit of the window duration. Starting from the electrical excitation change cycle, candidate assignment windows with a duration not less than the lower limit of the window duration are formed according to the temperature sampling cycle.
5. The method for evaluating the dual-parameter status of a GIS disconnector switch based on electrothermal parameters according to claim 4, characterized in that, The specific steps for creating a versioned status assessment record include: When the widths of the previous equivalent incremental resistance conservative outer envelope interval and the current equivalent incremental resistance conservative outer envelope interval are both less than the calibrated incremental resistance interval, the contact thermal state determination result is formed based on the complete inclusion relationship between each equivalent incremental resistance conservative outer envelope interval and the normal equivalent incremental resistance determination interval and the abnormal equivalent incremental resistance determination interval. The end-press response determination result is formed based on the complete inclusion, non-intersection, and partial intersection relationships between the end-press response range and the end-press response allowable range; The conservative outer range of the previous equivalent incremental resistance and the corresponding contact thermal state judgment result are associated with the previous contact state version. The conservative outer range of the current equivalent incremental resistance, the corresponding contact thermal state judgment result, the end crimping response range and the end mechanical response judgment result are associated with the current contact state version to form a versioned state assessment record.
Citation Information
Patent Citations
State identification method, device and equipment of isolating switch equipment and temperature measurement method of isolating switch equipment
CN112906576A
GIS isolation switch contact state identification method and system
CN118627261A
