A method for replacing a substation DC system without power interruption

Abstract

The application discloses a substation DC system non-stop power replacement method, and belongs to the technical field of smart grids, which comprises the following steps: configuring a non-stop power transformation platform and networking, and constructing a baseline state matrix; respectively using a jack type wire, a through mutual inductor and an insulating spacer to implement physical connection and isolation; injecting a perturbation pulse to solve a time constant to generate a current climbing instruction, and checking a terminal voltage difference to generate a conduction mark; dynamically updating a charging platform output voltage according to a feeder output current difference value, and solving a Pearson correlation coefficient; when parameters fall into a safety judgment interval of the baseline state matrix, disconnecting an original feeder output switch and removing isolation. The application adopts a closed-loop control strategy combining baseline state matrix construction, adaptive current injection and real-time correlation checking, can realize automatic and accurate disturbance-free transfer of a DC load, and guarantees safety of a replacement process and stability of power grid operation.

Description

Technical Field

[0001] This invention belongs to the field of smart grid technology, and in particular relates to a method for uninterrupted power supply replacement of DC systems in substations. Background Technology

[0002] As the core nerve center of the power system, the DC power supply system of a substation directly determines the safe and stable operation of the power grid. In actual operation and maintenance, when DC power supply equipment faces scenarios such as aging upgrades or outages due to faults, temporary DC power supply equipment is commonly used as a transitional solution to avoid interruption of DC load power supply within the substation and to ensure the continuous operation of key secondary equipment such as relay protection and automation control. Among related technologies, a typical solution is the mobile substation DC power supply panel disclosed in Chinese utility model patent CN201490760U. Its core design is to split the DC power supply system into an independent charging panel and a feeder panel. Both are equipped with locking casters for mobility. The charging panel integrates an AC power input terminal and a high-frequency rectifier module, while the feeder panel is equipped with a DC load output switch to meet the basic functional requirements of temporary power supply.

[0003] However, existing technologies have significant limitations and are difficult to adapt to the needs of efficient on-site operation and maintenance. On the one hand, the modular structure requires the separate handling and placement of two independent panels during on-site operations, and complex on-site electrical connections and debugging are necessary before they can be put into use. This not only significantly increases the workload of wiring and deployment time, but also easily leads to safety hazards due to on-site wiring errors. On the other hand, the dispersed placement of multiple independent devices occupies a large amount of on-site operation and maintenance space. Especially in time-sensitive scenarios such as substation emergency repairs, the cumbersome process of equipment handling, deployment, and debugging will seriously restrict the efficiency of emergency repair operations, and may even affect the safe and stable operation of the power grid due to untimely power transition. Summary of the Invention

[0004] To address the aforementioned problems, the present invention aims to provide a method for uninterrupted replacement of DC loads in substations. This method employs a closed-loop control strategy that combines baseline state matrix construction, adaptive current injection, and real-time correlation verification. This approach enables automated, precise, and disturbance-free transfer of DC loads, ensuring the safety of the replacement process and the stability of the power grid operation.

[0005] The above objectives can be achieved through the following approach:

[0006] A method for uninterrupted power supply replacement of a substation DC system, comprising the following steps:

[0007] Configure an uninterrupted power supply retrofit platform, including a mobile DC charging platform, a mobile DC feeder platform, a mobile battery pack, and a wireless transmission monitoring platform. The platform is networked through electrical connection components. The wireless transmission monitoring platform is controlled to collect the bus voltage sequence and branch current sequence of the original DC system, and calculate the components and quantiles to construct a baseline state matrix.

[0008] Identify the power supply branch and signal branch of the original DC system. For the power supply branch, establish an electrical connection to the mobile DC feeder platform using plug-in wires and configure a through-core current transformer to map the branch current signal in real time. For the signal branch, connect the plug-in wires through the terminal extension method and implement physical isolation and anti-detachment constraints using insulating spacers and wire clamps.

[0009] The mobile DC feeder platform is controlled to inject micro-perturbation pulse signals into the power supply branch, acquire transient response waveforms and calculate load decay time constant to generate current ramp command sequence, apply micro-amplitude test current and acquire terminal voltage difference, and generate branch conduction flag when the baseline state matrix determines that it falls into the voltage sub-position signal interval.

[0010] Responding to the branch conduction flag and executing the current ramp-up command sequence, the target branch current, feeder output current and non-target branch voltage sequence of the original DC system are collected synchronously. The output voltage of the mobile DC charging platform is dynamically updated based on the difference of the feeder output current between adjacent samples, and the Pearson correlation coefficient is calculated for the target branch current and non-target branch voltage.

[0011] When the Pearson correlation coefficient is within the safe correlation range of the baseline state matrix, and the feeder output current converges to the zero value judgment neighborhood of the baseline state matrix, disconnect the feeder output switch of the original DC system and remove the insulating partition, and send the takeover completion message through the wireless transmission monitoring platform.

[0012] Preferably, the calculation of components and quantiles to construct the baseline state matrix includes:

[0013] By using a one-click connection combination, a DC power transmission channel is established between the mobile DC charging platform, the mobile DC feeder platform, and the mobile battery pack, and a wireless transmission monitoring platform is connected via a communication bus.

[0014] The analog signals from the original DC system are received through the wireless transmission monitoring platform, and signal filtering and discretization are performed to generate bus voltage sequence and branch current sequence.

[0015] Steady-state components are extracted using bus voltage sequences and branch current sequences. Variance is calculated using bus voltage sequences and branch current sequences. Numerical boundaries are defined based on the variance. Baseline state matrix is ​​constructed by combining steady-state components and numerical boundaries.

[0016] Preferably, the baseline state matrix is ​​constructed by combining steady-state components and numerical boundaries, including:

[0017] The bus voltage sequence and branch current sequence are subjected to an arithmetic average operation over the entire time period to generate the expected voltage and current values ​​as steady-state components.

[0018] The standard deviations of the bus voltage series and branch current series are calculated. With the steady-state component as the center, three times the standard deviation are superimposed to construct the upper bound of voltage fluctuation, the lower bound of voltage fluctuation, the upper bound of current fluctuation, and the lower bound of current fluctuation, which serve as numerical boundaries and are written into the baseline state matrix.

[0019] Preferably, the use of insulating spacers and wire clamps to implement physical isolation and anti-detachment constraints includes:

[0020] For the power supply branch, select a plug-in type wire with a protective spring, insert the pin end of the plug-in type wire into the test hole of the current terminal block of the power supply branch, and pass the plug-in type wire through the center hole of the through-core transformer to connect to the mobile DC feeder platform to establish an electrical connection.

[0021] For signal branches, insert insulating spacers on the front and back sides of the wiring position in the remote signaling terminal of the signal branch. Connect the plug-in wire to the remote signaling terminal while retaining the original circuit wiring state, and use the wire clamp to lock the plug-in wire to the switch cabinet door or workbench side of the original DC system.

[0022] Preferably, the generation of branch conduction flags includes:

[0023] The mobile DC feeder platform is driven to output short-time voltage pulses and synchronously acquire feedback voltage sequences. The time required for the feedback voltage sequence to decay from the peak to half the peak value is calculated and used as the load decay time constant.

[0024] Based on the load decay time constant, the current ramp-up slope is indexed from the preset strategy table. When the load decay time constant is greater than the capacitive characteristic threshold of the strategy table, an S-shaped ramp-up curve is established; otherwise, a linear ramp-up curve is established as the current ramp-up command sequence.

[0025] The mobile DC feeder platform is controlled to output a constant current test current, and the terminal voltage difference between the plug-in wire and the original DC system is collected. When the absolute value of the terminal voltage difference is less than the difference between the upper limit of the voltage fluctuation and the steady-state component in the baseline state matrix, a branch conduction flag is generated.

[0026] Preferably, the calculation of the time required for the feedback voltage sequence to decay from its peak to half its peak value, as the load decay time constant, includes:

[0027] Traverse the feedback voltage sequence, lock the sampling point with the largest value as the peak amplitude, and extract the timestamp of the sampling point as the peak time;

[0028] In the feedback voltage sequence, the sampling point located after the peak moment is retrieved, and the sampling point with a value of half the peak amplitude is locked as the half-peak moment. The time difference between the half-peak moment and the peak moment is calculated as the load attenuation time constant.

[0029] Preferably, the calculation of the Pearson correlation coefficient between the target branch current and the non-target branch voltage includes:

[0030] The mobile DC feeder platform is driven to gradually increase the injected current according to the current ramp-up command sequence, while the target branch current, feeder output current and non-target branch voltage sequence of the original DC system are recorded synchronously through the wireless transmission monitoring platform.

[0031] The difference between the current and previous values ​​is calculated based on the feeder output current. This difference is then added to the current output voltage of the mobile DC charging platform to perform dynamic voltage updates.

[0032] Calculate the covariance of the target branch current and the non-target branch voltage sequence, calculate the product of the standard deviations of the target branch current and the non-target branch voltage sequence, and divide the covariance by the standard deviation product to obtain the Pearson correlation coefficient.

[0033] Preferably, the sequence of target branch current, feeder output current, and non-target branch voltage of the original DC system includes:

[0034] The current setpoint in the current ramp-up command sequence is analyzed, and the mobile DC feeder platform is adjusted to control the output current to rise in a step-like manner following the current setpoint.

[0035] By using a wireless transmission monitoring platform to issue a synchronous sampling trigger command, the acquisition channel is controlled to simultaneously read analog signals and perform analog-to-digital conversion, directly acquiring the target branch current, feeder output current and non-target branch voltage sequences that are consistent in time.

[0036] Preferably, the takeover completion message sent via the wireless transmission monitoring platform includes:

[0037] The difference between the upper bound of the current fluctuation and the expected value of the current is calculated based on the baseline state matrix and used as the zero value determination neighborhood. At the same time, the difference between the upper bound of the voltage fluctuation and the expected value of the voltage is calculated based on the baseline state matrix and divided by the expected value of the voltage to obtain the safety association interval.

[0038] When the absolute value of the feeder output current is less than the zero value of the neighborhood and the absolute value of the Pearson correlation coefficient is less than the boundary value of the safety association interval, the connection conditions are confirmed to be met.

[0039] The original DC system feeder output switch is tripped to cut off the power supply path of the original DC system to the power supply branch, and the insulating spacer is pulled out from the signal branch to restore the physical connection.

[0040] The wireless transmission monitoring platform generates structured messages and sends them to the dispatch center via wireless communication.

[0041] The present invention has the following advantages:

[0042] This invention quantifies and models the normal operating state of a DC system by constructing a baseline state matrix, providing a data benchmark and judgment basis for the entire replacement process. This changes the previous mode of relying on human experience for judgment, making operational decisions more scientific and improving the safety and controllability of replacement operations.

[0043] This invention employs a strategy of perturbation injection detection and load characteristic adaptation. By calculating the load decay time constant, it personalizes the current ramp-up curve, achieving flexible and smooth intervention for loads of different characteristics. It suppresses the inrush current and voltage drop that may occur during load transfer, ensuring the power supply quality and operational stability of the downstream secondary equipment and control system.

[0044] This invention introduces Pearson correlation coefficient for real-time correlation verification. By continuously monitoring the correlation between the operation of the target branch and the status of non-target branches, a dynamic safety monitoring mechanism is established. This mechanism can promptly detect and avoid chain reactions caused by potential electrical coupling or grounding faults, ensuring the non-disruptive nature of replacement operations and guaranteeing the safe isolation of the entire substation system. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating a method for uninterrupted power replacement of a DC system in a substation according to the present invention.

[0046] Figure 2 This is a schematic diagram of the statistical distribution and numerical boundary construction of the baseline state matrix in Embodiment 1 of the present invention;

[0047] Figure 3 This is a schematic diagram of the load feature identification and adaptive current ramp-up strategy in Embodiment 1 of the present invention;

[0048] Figure 4 This is a schematic diagram of the structure of a DC system replacement system for a substation that is not powered on. Detailed Implementation

[0049] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0050] Example 1: As Figure 1 As shown, a method for uninterrupted power supply replacement of a substation DC system includes the following steps:

[0051] Configure a live-line transformation platform, which includes a mobile DC charging platform, a mobile DC feeder platform, a mobile battery pack, and a wireless transmission monitoring platform. The platform is networked through electrical connection components. The wireless transmission monitoring platform is controlled to collect the bus voltage sequence and branch current sequence of the original DC system, and calculate the components and quantiles to construct a baseline state matrix.

[0052] Identify the power supply branch and signal branch of the original DC system. For the power supply branch, establish an electrical connection to the mobile DC feeder platform using plug-in wires and configure a through-core current transformer to map the branch current signal in real time. For the signal branch, connect the plug-in wires through the terminal extension method and implement physical isolation and anti-detachment constraints using insulating spacers and wire clamps.

[0053] The mobile DC feeder platform is controlled to inject micro-perturbation pulse signals into the power supply branch, acquire transient response waveforms and calculate load decay time constant to generate current ramp command sequence, apply micro-amplitude test current and acquire terminal voltage difference, and generate branch conduction flag when the baseline state matrix determines that it falls into the voltage sub-position signal interval.

[0054] Responding to the branch conduction flag and executing the current ramp-up command sequence, the target branch current, feeder output current and non-target branch voltage sequence of the original DC system are collected synchronously. The output voltage of the mobile DC charging platform is dynamically updated based on the difference of the feeder output current between adjacent samples, and the Pearson correlation coefficient is calculated for the target branch current and non-target branch voltage.

[0055] When the Pearson correlation coefficient is within the safe correlation range of the baseline state matrix, and the feeder output current converges to the zero value judgment neighborhood of the baseline state matrix, disconnect the feeder output switch of the original DC system and remove the insulating partition, and send the takeover completion message through the wireless transmission monitoring platform.

[0056] Solving for components and quantiles to construct the baseline state matrix includes:

[0057] By using a one-click connection combination, a DC power transmission channel is established between the mobile DC charging platform, the mobile DC feeder platform, and the mobile battery pack, and a wireless transmission monitoring platform is connected via a communication bus.

[0058] To address the cumbersome and error-prone wiring issues of traditional emergency power supplies, a rapid physical network is established to build the hardware foundation. Operators use a one-button connection connector to physically interface the mobile DC charging platform, mobile DC feeder platform, and mobile battery pack. This one-button connection connector is a composite connector integrating high-power DC conductors and low-voltage communication cables. Internally, it includes low-impedance copper busbars capable of carrying DC loads of tens to hundreds of amperes, and also integrates shielded twisted-pair cables. When the connector is closed, a DC power transmission channel is established, allowing bidirectional power flow between modules. Simultaneously, the wireless transmission monitoring platform connects to the network via an industrial-grade communication bus, automatically registering and handshaking nodes on each platform, establishing a full-duplex data interaction link to ensure millisecond-level transmission of control commands and status data.

[0059] The analog signals from the original DC system are received through the wireless transmission monitoring platform, and signal filtering and discretization are performed to generate bus voltage sequence and branch current sequence.

[0060] The wireless transmission monitoring platform connects to the positive and negative buses of the original DC system via high-impedance voltage probes and uses open-type sensors mounted on the target branches to receive analog electrical signals representing bus voltage and branch current, typically 0-5V or 4-20mA standard signals. Subsequently, the platform's internal digital signal processor (DSP) performs low-pass filtering on the analog signals, using a digital filter to remove power frequency interference and high-frequency electromagnetic noise. Next, an analog-to-digital converter (ADC) discretizes the filtered signals at a fixed sampling frequency. Each sample generates a numerical point with a high-precision timestamp; consecutive sampling points are arranged chronologically to form the bus voltage sequence and the branch current sequence, respectively.

[0061] For example, assuming the rated voltage of the original DC system is 220V, the wireless transmission monitoring platform is set to a sampling frequency of 1000Hz and a continuous acquisition time of 10 seconds. The platform converts the acquired analog voltage signal into a digital quantity using an ADC, obtaining a bus voltage sequence containing 10,000 discrete points. Similarly, for a power supply branch with a load of approximately 10A, the platform synchronously generates a branch current sequence containing 10,000 discrete points. .

[0062] Steady-state components are extracted using bus voltage sequences and branch current sequences. Variance is calculated using bus voltage sequences and branch current sequences. Numerical boundaries are defined based on the variance. Baseline state matrix is ​​constructed by combining steady-state components and numerical boundaries.

[0063] The wireless transmission monitoring platform reads the bus voltage sequence and branch current sequence from memory. First, it calculates their arithmetic mean, defining it as the steady-state component. The physical meaning of the steady-state component is to characterize the DC operating point under the current operating conditions. Then, the platform calculates the variance or standard deviation of the sequence. This indicator measures the dispersion of the signal relative to the steady-state component, i.e., background noise or ripple amplitude. Finally, based on the variance, numerical boundaries are defined, using the statistical principle of three times the standard deviation (3σ) to cover a probability interval of 99.7% as the limit range of normal fluctuations. Finally, the steady-state component, upper fluctuation bound, and lower fluctuation bound are encapsulated in a predetermined format to construct a baseline state matrix. This matrix is ​​the sole quantitative benchmark for judging whether the contact is good and whether the current has returned to zero. The calculation logic can be quantitatively described by the following formula: First, calculate the steady-state component of the voltage or current. Taking the bus voltage sequence as an example, the steady-state component... The calculation is as follows:

[0064] ;

[0065] in, The extracted steady-state component represents the expected voltage value for a voltage sequence and the expected current value for a current sequence; N represents the total number of sampling points in the sequence, which is determined by the sampling frequency and sampling duration. This represents the value at the i-th sampling time in the sequence; The summation sign is used. The physical principle of the formula is to eliminate random white noise through integral averaging and extract the true value of the DC component. Next, the standard deviation of the sequence is calculated. Used to define the fluctuation range:

[0066] ;

[0067] in, Represents standard deviation; This represents the squared deviation of each sampling point relative to the steady-state component; N-1 is used to obtain an unbiased estimate through correction. The physical meaning of the formula represents the combined intensity of ripple and measurement noise. Finally, a baseline state matrix is ​​constructed based on the standard deviation. :

[0068] ;

[0069] in, and These are the steady-state components of voltage and current, respectively; and , respectively, are the standard deviations of the voltage and current sequences; k is the confidence coefficient, the value of which is based on the statistical law of normal distribution, and is usually taken as k=3, at which point the numerical boundary... It covers 99.73% of normal fluctuation samples, with Base representing the steady-state component. If the electromagnetic environment is extremely harsh, k can be adjusted to 4 or 5 to relax the boundaries. The elements in the first row of the matrix represent the expected voltage value, the upper bound of voltage fluctuation, and the lower bound of voltage fluctuation, respectively; the elements in the second row represent the expected current value, the upper bound of current fluctuation, and the lower bound of current fluctuation, respectively.

[0070] For example, the wireless transmission monitoring platform calculates a bus voltage sequence containing 10,000 points. Assume the steady-state voltage components are calculated. The voltage standard deviation was calculated. Assume that the steady-state component of the branch current is calculated. The standard deviation of the current was calculated. Set the confidence level coefficient k=3. Then the calculated upper bound of the voltage fluctuation is: The lower bound is The upper limit of current fluctuation is The lower bound is The final constructed baseline state matrix for:

[0071] ;

[0072] The matrix is ​​then stored in the dynamic memory of the wireless transmission monitoring platform for real-time verification.

[0073] The baseline state matrix is ​​constructed by combining steady-state components and numerical boundaries, including:

[0074] The bus voltage sequence and branch current sequence are subjected to an arithmetic average operation over the entire time period to generate the expected voltage and current values ​​as steady-state components.

[0075] To obtain the steady-state components representing a long-term operating baseline, the wireless transmission monitoring platform invokes the steady-state component calculation logic, but limits the data range for the operation to "all time periods." This means the processor does not segment or sample the data, but instead iterates through all discrete points within the entire sampling period. The platform uses an accumulator to sum the values ​​in the bus voltage sequence and branch current sequence, and then divides them by the total number of sampling points. This operation is physically equivalent to performing a deep integration of the signal, and its engineering purpose is to completely smooth out short-term load fluctuations or random pulse interference that may occur during the sampling period, thereby extracting the purest DC operating point, i.e., the expected voltage and current values.

[0076] For example, suppose a wireless transmission monitoring platform collects voltage data from a DC bus over a 5-minute period. During this time, due to the start-up and shutdown of nearby devices, the bus voltage experiences frequent, minor fluctuations around 220V. The platform performs an arithmetic average calculation over the entire period, summing the values ​​from the 300,000 sampling points collected over these 5 minutes and averaging them to obtain an expected voltage value of 220.5V. This value is then locked as the voltage reference center in the subsequent verification process, unaffected by individual instantaneous voltage jumps.

[0077] The standard deviations of the bus voltage series and branch current series are calculated. With the steady-state component as the center, three times the standard deviation are superimposed to construct the upper bound of voltage fluctuation, the lower bound of voltage fluctuation, the upper bound of current fluctuation, and the lower bound of current fluctuation, which serve as numerical boundaries and are written into the baseline state matrix.

[0078] After obtaining the steady-state components, a reasonable "safety envelope" needs to be determined. The platform first quantifies the dispersion of bus voltage and branch current based on standard deviation calculation logic. Based on this, it limits the boundary construction to 3σ. The processor reads the calculated voltage and current standard deviations and multiplies them by a coefficient of 3. Then, centered on the expected voltage and current values, this 3σ value is added to and subtracted from each. The sum is used as the upper bound of the fluctuation, and the difference as the lower bound. For typical substation DC ripple noise, its amplitude distribution approximately follows a normal distribution. The interval of 3 times the standard deviation can cover approximately 99.7% of normal operating samples, meaning that only a very small probability of anomalies will fall outside this interval, thus reducing the false positive rate while ensuring high sensitivity. Figure 2 As shown, the scatter distribution in the central region reflects the real-time operating status, and the dashed rectangle defines the confidence interval for normal fluctuations, which is composed of the steady-state components superimposed with three times the standard deviation.

[0079] For example, given a known expected voltage value of 220.5V, the platform calculates the standard deviation of this data segment to be 0.2V. To construct numerical boundaries, the platform performs a multiplication operation to obtain three times the standard deviation, which is 0.6V. Then, it performs addition and subtraction operations: upper limit of voltage fluctuation = 220.5V + 0.6V = 221.1V; lower limit of voltage fluctuation = 220.5V - 0.6V = 219.9V. Similarly, if the expected current value is 50A and the standard deviation is 0.1A, then the upper limit of current fluctuation is 50.3A, and the lower limit is 49.7A. The wireless transmission monitoring platform writes these four calculated boundary values, along with the expected value, into the baseline state matrix according to a specific data structure, completing the transformation from raw data to a quantized benchmark.

[0080] Physical isolation and anti-detachment constraints implemented using insulating spacers and wire clamps include:

[0081] For the power supply branch, select a plug-in type wire with a protective spring, insert the pin end of the plug-in type wire into the test hole of the current terminal block of the power supply branch, and pass the plug-in type wire through the center hole of the through-core transformer to connect to the mobile DC feeder platform to establish an electrical connection.

[0082] For power supply branches carrying the main load current, the aim is to establish a temporary bypass power supply channel with low contact resistance and accurate measurement. Operators select a plug-in type conductor with a protective spring. The pin at the front end of this conductor has an external spring-loaded structure made of beryllium copper or phosphor bronze, such as a lantern-type or drum-spring type, which provides continuous radial tension during insertion, ensuring tight contact with the inner wall of the terminal hole. This design effectively increases the contact area, significantly reducing contact resistance and thus avoiding significant temperature rise or poor contact sparks when large currents continuously flow. During operation, the pin end of this plug-in type conductor with the protective spring is firmly inserted into the current terminal block test hole of the corresponding air switch terminal in the power supply branch. These test holes are standard interfaces reserved on the terminal block, specifically for electrical testing or temporary power access without loosening the main wiring. After insertion, the main body of the plug-in type conductor must first pass through the center hole of the current transformer, and then its other end is connected to the power output terminal of the mobile DC feeder platform. The current transformer configured here is a non-invasive current sensor that uses the principle of electromagnetic induction to map the magnitude of the current flowing through the plug-in conductor into a standard secondary side small current or voltage signal in real time and linearly, providing key feedback metering data for current ramp-up control.

[0083] For example, assume the rated current of the power supply branch to be replaced is 20 amps. The operator selects a 32-amp rated current-carrying plug-in conductor with a protective spring, featuring a 4mm diameter pin and gold-plated beryllium copper spring. The operator forcefully pushes the pin into the test hole of the current terminal block until the spring fully springs open and locks in place. The conductor is then passed through the center of a 50A / 5V feedthrough transformer. At this point, the mobile DC feeder platform can accurately determine the current value injected into the power supply branch by reading the voltage signal output from the feedthrough transformer. For example, when 10 amps of current flow through the conductor, the feedthrough transformer outputs a 1-volt voltage signal to the control system, thereby achieving closed-loop control.

[0084] For signal branches, insert insulating spacers on the front and back sides of the wiring position in the remote signaling terminal of the signal branch. Connect the plug-in wire to the remote signaling terminal while retaining the original circuit wiring state, and use the wire clamp to lock the plug-in wire to the switch cabinet door or workbench side of the original DC system.

[0085] For signal branches transmitting status information or control logic, the engineering objective is to temporarily disconnect and switch the signal path without interrupting the physical integrity of the original circuit. During operation, the operator inserts thin insulating spacers into the gaps on both sides of the contact surface between the screw and the contact piece at the remote signaling terminal of the target signal branch. These insulating spacers are typically made of a high-insulation-strength material, with a thickness between 0.5 mm and 1 mm. Their function is to physically block the electrical path between the screw, the original wire contact piece, and the busbar, creating a "sandwich" isolation structure. After confirming that the original circuit has achieved electrical isolation, the operator connects the terminal of another plug-in wire to the lower layer of the screw at the remote signaling terminal and tightens the screw. At this point, the physical connection of the original circuit is maintained, but the electrical connection is broken, and the signal path is led out to the new platform through this temporarily connected plug-in wire. To prevent temporary wiring from loosening or falling off due to personnel movement, vibration, or accidental dragging, which could lead to signal loss or system malfunction, the cable portion of the plug-in wire must be securely locked to the inside of the original DC system switch cabinet door or a nearby fixed workbench using wire clamps. This provides effective anti-fall-off restraint and ensures connection reliability throughout the entire modification process.

[0086] For example, for a remote signaling terminal transmitting a "closed position" signal, the operating voltage is 24 volts. The operator uses two red, 0.8 mm thick insulating spacers, inserting them respectively under the screw head and the contact plate of the original terminal block. After confirming that the original circuit resistance is infinite (measured with a multimeter), the plug-in wire connected to the wireless transmission monitoring platform's acquisition end is connected to the screw and tightened. Then, approximately 15 cm from the terminal block, a strong magnetic wire clip is attached to the inside of the switch cabinet door panel, and the plug-in wire is inserted into the clip slot and locked. This way, even if a slight external pulling force is applied to the wire, the force will be absorbed by the clip and not transmitted to the fragile terminal connection point, thus ensuring the stability of signal acquisition.

[0087] The generation of branch continuity flags includes:

[0088] The mobile DC feeder platform is driven to output short-time voltage pulses and synchronously acquire feedback voltage sequences. The time required for the feedback voltage sequence to decay from the peak to half the peak value is calculated and used as the load decay time constant.

[0089] To ensure the reliability of the physical connection and probe the electrical characteristics of the target branch before formally taking over the load, a perturbation injection process is performed. The power module inside the mobile DC feeder platform outputs a short-duration voltage pulse, typically a square wave or step signal with an amplitude of 5% to 10% of the rated voltage and a duration of 20 to 50 milliseconds. Because the pulse has low energy and extremely short duration, it will not impact the operating load equipment. Simultaneously with the pulse injection, the wireless transmission monitoring platform or the feeder platform's built-in high-speed acquisition module synchronously acquires the feedback voltage sequence at a high sampling rate, recording the transient response waveform of the voltage at the pulse injection point over time. The processor then analyzes this feedback voltage sequence; the core algorithm is to find the "half-peak" decay time of the waveform. The processor first traverses the sequence to find the point of maximum voltage amplitude and records its time. and amplitude Next, the processor... Searching the subsequent sequence, we found that the voltage amplitude dropped to... The moment is recorded as Finally, calculate the difference between these two moments, i.e. This difference is defined as the load decay time constant. This time constant physically reflects the size of the equivalent capacitance of the load circuit: the larger the capacitance, the slower the release of stored charge, the smoother the voltage decay, and the longer the time constant.

[0090] For example, suppose a mobile DC feeder platform injects a short pulse of 10 volts into the power supply branch. The acquired feedback voltage sequence shows that the voltage peaks at 10 volts at 50 milliseconds. The voltage then begins to decay exponentially, reaching 5 volts at 150 milliseconds, which is half of its peak value. The processor performs a subtraction operation: 150 - 50 = 100 milliseconds. Therefore, the load decay time constant for this branch is determined to be 100 milliseconds.

[0091] Based on the load decay time constant, the current ramp-up slope is indexed from the preset strategy table. When the load decay time constant is greater than the capacitive characteristic threshold of the strategy table, an S-shaped ramp-up curve is established; otherwise, a linear ramp-up curve is established as the current ramp-up command sequence.

[0092] After obtaining the load decay time constant, the engineering objective shifts to generating the optimal current ramp-up command sequence to achieve "soft start." Based on the calculated constant value, a matching control strategy is retrieved from a strategy table. This strategy table is a lookup table pre-built and stored in memory based on historical experimental data and different types of DC load models, such as resistive, inductive, and capacitive loads. The key decision logic lies in identifying "large capacitive loads." When the measured load decay time constant is greater than the capacitive characteristic threshold in the strategy table, which is set between 100 and 200 milliseconds, it indicates that the load exhibits significant capacitive characteristics. In this case, a linear ramp-up would easily generate excessive charging inrush current. Therefore, an S-shaped ramp-up curve is chosen. The S-shaped curve has the characteristics of "slow at both ends and fast in the middle." The current change rate is extremely small in the initial stage, giving the capacitor sufficient pre-charging time to avoid inrush; the middle stage rises rapidly to improve efficiency; and the final stage flattens out again for seamless connection. Conversely, if the time constant is less than this threshold, the load is determined to be predominantly resistive, and a simple linear ramp-up curve is established to achieve faster power take-off. Figure 3 (a) shows the transient response of the perturbation voltage and the calculation process of the time constant under different load characteristics. Figure 3 (b) shows the sequence of linear or S-shaped current ramp-up commands generated based on the calculated time constant.

[0093] For example, the preset strategy table specifies a capacitive characteristic threshold of 100 milliseconds. Scenario A: The time constant of a certain branch is measured to be 150 milliseconds. Because 150 > 100, the load is determined to be a strongly capacitive load. An S-shaped function algorithm is called to generate a current sequence containing 100 set points, which, when connected, form an "S" shape. Scenario B: The time constant of another branch is measured to be 20 milliseconds. Because 20 < 100, the load is determined to be a resistive load. A linear function is called to generate a uniformly increasing current sequence.

[0094] The mobile DC feeder platform is controlled to output a constant current test current, and the terminal voltage difference between the plug-in wire and the original DC system is collected. When the absolute value of the terminal voltage difference is less than the difference between the upper limit of the voltage fluctuation and the steady-state component in the baseline state matrix, a branch conduction flag is generated.

[0095] Before implementing high-current connection, physical connection quality verification must be performed to prevent overheating or arcing at the joints due to poor contact. The mobile DC feeder platform switches to constant current output mode and applies a small test current to the branch. This current is typically set to 50 mA to 200 mA, which is sufficient to generate a measurable voltage drop without causing operational disturbance. Simultaneously, a high-precision voltage acquisition module measures the terminal voltage difference between the plug-in conductor and the original DC system terminal block. This voltage difference is mainly generated by the contact resistance of the temporary connection. To determine if this voltage difference is acceptable, a baseline state matrix is ​​introduced as a criterion. Logically, if the voltage drop introduced by the temporary connection is smaller than the original system's normal background noise, then the connection is absolutely safe. The difference between the "upper bound of voltage fluctuation" and the "steady-state component of voltage" in the baseline state matrix is ​​calculated; physically, this difference is equal to three times the standard deviation and is taken as the maximum permissible deviation. When the absolute value of the measured voltage difference is less than this difference, the contact is considered acceptable, and a branch continuity indicator is generated. This judgment process can be quantified using the following formula:

[0096] ;

[0097] in, This represents the absolute value of the collected terminal voltage difference, in volts. This represents the upper bound of voltage fluctuations stored in the baseline state matrix; This represents the steady-state voltage component stored in the baseline state matrix, i.e., the expected voltage value. (Right side of the equation) The physical meaning represents the inherent background noise amplitude of the original DC system, i.e., statistically significant. The principle behind this inequality is that as long as the contact voltage drop is submerged in its own background noise, the electrical effects of the connection can be considered negligible.

[0098] For example, the mobile DC feeder platform outputs a test current of 0.1 amps. The voltage acquisition module measures the voltage difference between the plug-in conductor and the original terminal. The voltage is 0.05 volts. The steady-state voltage component was retrieved from the baseline state matrix in memory. Volt, upper limit of voltage fluctuation Volts. Perform calculations: Allowable deviation. Volts. Compare: Since the inequality holds, it is determined that the physical connection is very tight, and a "branch continuity flag" is generated, triggering the formal current ramp-up process.

[0099] The time required for the feedback voltage sequence to decay from its peak to half its peak value is calculated and used as the load decay time constant, including:

[0100] Traverse the feedback voltage sequence, lock the sampling point with the largest value as the peak amplitude, and extract the timestamp of the sampling point as the peak time;

[0101] To accurately calculate the load attenuation time constant, the built-in processor of the wireless transmission monitoring platform or mobile DC feeder platform needs to perform digital signal processing on the acquired feedback voltage sequence. This feedback voltage sequence is a set of discrete voltage values ​​ordered by time, acquired at high frequency by an analog-to-digital converter after the injection of a perturbation pulse signal. The purpose is to accurately locate the highest energy point at the moment of pulse injection from the transient response waveform, i.e., the starting point of the discharge process. The processor first initializes a maximum value variable and a corresponding timestamp variable. Then, it performs a complete traversal scan of the feedback voltage sequence stored in memory. During the scan, the processor compares the voltage value of the current sampling point with the maximum value variable one by one. If the current sampling point value is greater than the maximum value variable, the maximum value variable is updated to the current value, and the acquisition time corresponding to that sampling point is recorded simultaneously. After the traversal is complete, the final locked maximum value is the peak amplitude, and the corresponding time is the peak moment. This process effectively eliminates background noise interference before pulse injection, ensuring that the reference point for subsequent attenuation calculations is accurate.

[0102] For example, suppose the feedback voltage sequence collected by the wireless transmission monitoring platform contains 500 data points with a sampling interval of 0.1 milliseconds. The processor iterates from the first point to the 500th point. After comparison, it finds that the voltage value of the 50th sampling point is 12.0 volts, and this value is greater than the values ​​of all other points in the sequence. The processor then locks 12.0 volts as the peak amplitude. At the same time, it reads the product of the index of this point and the sampling interval to determine that the 5.0 millisecond is the peak moment.

[0103] In the feedback voltage sequence, the sampling point located after the peak moment is retrieved, and the sampling point with a value of half the peak amplitude is locked as the half-peak moment. The time difference between the half-peak moment and the peak moment is calculated as the load attenuation time constant.

[0104] After determining the discharge initiation point, a directional search is initiated to find the critical node where the voltage energy decays to half and to quantify the decay rate. First, the half-peak threshold is calculated, which is the peak amplitude multiplied by 0.5. Then, in the feedback voltage sequence, a linear search is performed in the positive time axis, starting from the index of the next sampling point immediately following the peak moment. During the search, the voltage value of the current sampling point is continuously compared with the calculated half-peak threshold. When a sampling point's voltage value is found to be equal to or less than the half-peak threshold for the first time, the search immediately stops. The timestamp corresponding to this sampling point is recorded as the half-peak moment. Finally, a subtraction operation is performed, subtracting the peak moment from the half-peak moment; the resulting time difference is defined as the load decay time constant. This calculation logic is based on a simplified model of the circuit discharge principle, utilizing the "half-life" characteristic to characterize the equivalent capacitance of the load. The calculation logic can be quantified using the following formula: First, determine the half-peak moment. Conditions to be met:

[0105] ;

[0106] and .in, The peak amplitude is locked; The voltage value of the first sampling point retrieved in the sequence that is less than or equal to half the peak amplitude; This represents the peak value. Next, the load decay time constant is calculated. :

[0107] ;

[0108] in, This is the required load decay time constant. Physically, it means that under the same discharge path impedance, The larger the value, the stronger the load side's ability to store charge, and the slower the voltage drop; conversely, the smaller the value, the more resistive the load is, and the faster the charge dissipates.

[0109] For example, the peak amplitude is known to be 12.0 volts, and the peak time is 5.0 milliseconds. The half-peak threshold is calculated to be 6.0 volts. The data is then retrieved starting from the 5.1-millisecond data point. Due to the large parallel filter capacitor on the load side, the voltage drops slowly. The voltage at the sampling point first drops to 5.98 volts at 105.0 milliseconds. 105.0 milliseconds is locked as the half-peak time. The load decay time constant = 105.0 - 5.0 = 100.0 milliseconds. Since 100 milliseconds is a large value, this branch is determined to be a capacitive load.

[0110] For example, suppose that in the test of another branch, the peak amplitude is also 12.0 volts, and the peak time is 5.0 milliseconds. However, since this branch is mainly a purely resistive load, the voltage drops rapidly after removal. Upon investigation, it was found that the voltage had already dropped to 5.5 volts at 5.5 milliseconds, less than 6.0 volts. 5.5 milliseconds is identified as the half-peak time. The load decay time constant = 5.5 - 5.0 = 0.5 milliseconds. Since 0.5 milliseconds is a very small value, this branch will be determined to be a resistive load, suitable for a fast linear ramp-up strategy.

[0111] The calculation of the Pearson correlation coefficient between the target branch current and the non-target branch voltage includes:

[0112] The mobile DC feeder platform is driven to gradually increase the injected current according to the current ramp-up command sequence, while the target branch current, feeder output current and non-target branch voltage sequence of the original DC system are recorded synchronously through the wireless transmission monitoring platform.

[0113] To achieve a smooth load transfer from the original DC system to the uninterruptible power supply (UPS) platform, a current ramp-up procedure is initiated upon receiving a branch continuity flag. The engineering objective is to control the rate of current injection growth based on a generated "S-shaped" or "linear" current ramp-up command sequence, preventing bus voltage drops or relay protection malfunctions caused by sudden current changes. The digital controller of the mobile DC feeder platform analyzes the discrete setpoints in the command sequence and, by adjusting the duty cycle of the power devices, drives their power modules to gradually increase the current injected into the power supply branch until the current reaches the steady-state load current value of the target branch. During this period, the wireless transmission monitoring platform uses its multi-channel data acquisition card to record three key time-series data points at high frequency using a synchronous triggering method: Target branch current sequence: read from sensors built into the original DC system or temporarily installed clamp meters, reflecting the current reduction process on the original power supply path. Feeder output current sequence: measured in real-time by a feedthrough transformer, reflecting the current increase process on the new power supply path. Non-target branch voltage sequence: Select several branches that are physically close but electrically unrelated, and collect their voltage to ground or to the negative bus as interference monitoring targets.

[0114] For example, assume the target branch's load current is 10A. The mobile DC feeder platform, following instructions, linearly increases the output current from 0A to 10A within 5 seconds. During this process, the wireless transmission monitoring platform records data every 10 milliseconds. Ultimately, three sequences containing 500 data points are formed in memory: the feeder output current sequence increasing from 0 to 10, the target branch current sequence decreasing from 10 to 0, and the non-target branch voltage sequence.

[0115] The difference between the current and previous values ​​is calculated based on the feeder output current. This difference is then added to the current output voltage of the mobile DC charging platform to perform dynamic voltage updates.

[0116] To maintain absolute bus voltage stability and prevent voltage fluctuations caused by changes in connection cable voltage drop during load transfer, a dynamic voltage feedforward compensation strategy is implemented. The aim is to compensate for voltage drops caused by factors such as feeder impedance and connection resistance in real time. In each control cycle, the control system reads the current feeder output current value. and the current value at the previous moment Calculate the difference between the two. This difference reflects the increase in load current. Multiply this current difference by a preset compensation factor. In engineering practice, this coefficient is taken as the equivalent resistance of the connecting wire, for example, 0.05 ohms, resulting in a small voltage compensation. Then, this compensation is added to the current output voltage setting of the mobile DC charging platform, thereby dynamically fine-tuning the voltage of the power supply bus to offset the increased line voltage drop caused by the increase in current, ensuring that the voltage at the load end is always stable within the normal range defined by the baseline state matrix.

[0117] Calculate the covariance of the target branch current and the non-target branch voltage sequence, calculate the product of the standard deviations of the target branch current and the non-target branch voltage sequence, and divide the covariance by the standard deviation product to obtain the Pearson correlation coefficient.

[0118] Throughout the current ramp-up process, a critical safety check is performed in parallel. Its engineering purpose is to assess the electrical impact of the connection operation on other parts in real time, particularly to prevent the risks of "cross-connection" or "common impedance coupling." The Pearson correlation coefficient is calculated on the synchronously acquired target branch current sequence X and non-target branch voltage sequence Y. The Pearson correlation coefficient is a dimensionless statistical indicator that measures the degree of linear correlation between two variables, ranging from -1 to +1. The closer the absolute value is to 1, the stronger the linear correlation; the closer the absolute value is to 0, the less correlated they are. By calculating this coefficient, it is possible to quantitatively determine whether the increase in injected current has caused abnormal fluctuations in the non-target branch voltage. If the absolute value of the coefficient exceeds the safety threshold, it indicates electrical coupling, and the operation must be stopped immediately. The calculation process of this coefficient is based on the following standard statistical formula: First, calculate the target branch current sequence... Non-target branch voltage sequence covariance :

[0119] ;

[0120] Where N is the length of the sequence involved in the calculation (the total number of sampling points in the sequence), which is usually the length of the nearest sliding window, such as 100 points; and These are the values ​​at the i-th sampling point (sampling time); and These are the arithmetic means of the two series within the current window. Covariance reflects the consistency of the overall trend of change of the two variables. Then, the standard deviations of the two series are calculated. and Finally, the Pearson correlation coefficient was calculated. :

[0121] ;

[0122] This value is updated at the end of each sampling period and compared with the safe association interval defined in the baseline state matrix.

[0123] For example, suppose that during a takeover process, the target branch current... The voltage is dropping rapidly as it is taken over by the new power source. At this time, the wireless transmission monitoring platform detects a nearby non-target branch voltage. A synchronous and significant voltage drop occurred. Calculations based on data from the most recent second show that when… For every 1A decrease, The voltage drops by 0.5V. Calculate the covariance. Given a large positive value, calculate the product of standard deviations. It is a positive value. Dividing the two gives... Since 0.95 is much greater than the safety threshold, a serious "common impedance coupling" fault was immediately identified, which may be due to two branches sharing the same loosely connected neutral wire. An alarm was then triggered to suspend the takeover, thus preventing the potential accident from escalating.

[0124] The records of the target branch current, feeder output current, and non-target branch voltage sequences of the original DC system include:

[0125] The current setpoint in the current ramp-up command sequence is analyzed, and the mobile DC feeder platform is adjusted to control the output current to rise in a step-like manner following the current setpoint.

[0126] To ensure high temporal consistency of data acquired during current ramp-up, thereby guaranteeing the accuracy of correlation analysis, a current control strategy was employed. The engineering objective was to generate a controllable current injection process, discretizing the continuous dynamic process into a series of steady-state segments. The microprocessor of the mobile DC feeder platform first reads the current ramp-up instruction sequence from memory, which is essentially an array of current setpoints with time indices. A high-precision closed-loop feedback adjustment mechanism is used, with the current setpoint at the current moment serving as the target reference. This reference value is compared with the feedback value of the actual output current, the error signal is calculated, and the duty cycle or drive voltage of the switching transistors in the power conversion unit is dynamically adjusted accordingly. Through this adjustment, the power output unit of the mobile DC feeder platform is driven, ensuring that its output current strictly follows these setpoint changes. This results in a stepped upward trend in the actual feeder output current, maintaining a brief "dwell time" at each setpoint. The design of this dwell time is crucial; it allows transient oscillations in the circuit to decay completely, ensuring that sampling is performed in a current-stable state, thus eliminating noise interference during dynamic adjustment.

[0127] For example, suppose a current ramp-up command sequence requires the current to rise from 0 amps to 5 amps within 500 milliseconds, in steps of 1 amp. The mobile DC feeder platform parses this command, first setting the output target to 1 amp. The PID controller adjusts the power transistor duty cycle, causing the output current to rise rapidly within 2 milliseconds and stabilize at 1.0 amp. The platform maintains this state for 98 milliseconds. Subsequently, the target is updated to 2 amps, and the controller adjusts again, causing the current to jump and stabilize at 2.0 amp, again maintaining this state for 98 milliseconds. This cycle repeats until the current reaches 5 amps. This step-like ramp-up provides five clear and stable sampling windows for subsequent data acquisition.

[0128] By using a wireless transmission monitoring platform to issue a synchronous sampling trigger command, the acquisition channel is controlled to simultaneously read analog signals and perform analog-to-digital conversion, directly acquiring the target branch current, feeder output current and non-target branch voltage sequences that are consistent in time.

[0129] Synchronous data acquisition is performed during the dwell time of the current at each step. The purpose is to eliminate data phase deviation caused by asynchronous sampling clocks between different measurement channels. The wireless transmission monitoring platform, acting as the central hub of the entire data acquisition process, is internally configured with a high-precision global clock source or hardware trigger circuit. At the arrival of each sampling cycle, the platform broadcasts a hard real-time synchronous sampling trigger command to all relevant data acquisition channels via a high-speed bus or dedicated signal line. This trigger command directly acts on the "conversion hold" pin of the analog-to-digital converter (ADC) on each acquisition card. Upon receiving this command, all ADC chips responsible for measuring the target branch current of the original DC system, the output current of the mobile platform feeder, and the voltage of non-target branches simultaneously "freeze" the current analog signal voltage value and immediately initiate the analog-to-digital conversion process. Since the transmission delay of the trigger signal is typically in the nanosecond range, it is negligible compared to millisecond-level signal changes; therefore, it can be assumed that all channels complete data acquisition at the same physical moment. The acquired digital quantities are then timestamped and stored in a buffer. This hardware-level synchronization mechanism enables the direct acquisition of three sets of data sequences that are perfectly aligned on the time axis, eliminating time deviations and ensuring that the calculated covariance and Pearson correlation coefficient truly reflect the instantaneous coupling relationship between physical quantities.

[0130] For example, in a takeover operation, the wireless transmission monitoring platform is configured with a sampling rate of 1kHz. When the clock reaches T=10.050 seconds, the platform sends a high-level trigger signal. Upon receiving the signal, channel A, connected to the current transformer, locks in the analog value of the feeder output current as 5.00A; channel B, connected to the original branch sensor, locks in the analog value of the target branch current as 5.02A at the same instant; channel C, connected to the non-target branch, locks in the analog value of the non-target branch voltage as 220.1V at the same instant. These three values ​​are recorded as a single line of data in the database. Without this synchronization mechanism, channel C might only sample at T=10.055 seconds; if voltage fluctuations occur at this time, the calculated correlation would be incorrect. This ensures... , and The time t in the equation is strictly consistent. This represents the time-varying sequence of the target branch current. This represents the sequence of feeder output current changes over time. This represents the time-varying sequence of non-target branch voltages.

[0131] The takeover completion message sent via the wireless transmission monitoring platform includes:

[0132] The difference between the upper bound of the current fluctuation and the expected value of the current is calculated based on the baseline state matrix and used as the zero value determination neighborhood. At the same time, the difference between the upper bound of the voltage fluctuation and the expected value of the voltage is calculated based on the baseline state matrix and divided by the expected value of the voltage to obtain the safety association interval.

[0133] To achieve seamless takeover and ensure absolute operational safety, a rigorous threshold definition process must be implemented. The engineering objective is to quantify abstract stability characteristics into concrete, real-time comparable numerical boundaries based on the established baseline state matrix. The wireless transmission monitoring platform reads the baseline state matrix from memory and first calculates the zero-value judgment neighborhood. The platform extracts the upper bound of current fluctuation and the expected current value from the matrix and calculates the difference between them. Physically, this difference equals three times the current standard deviation, representing the background noise amplitude of the branch under normal steady-state conditions. Defining this difference as the zero-value judgment neighborhood means that as long as the measured current is less than this background noise, statistically, the current can be considered "zero," i.e., submerged in noise and without effective load. Secondly, the judgment boundary value of the safety correlation interval is calculated. The platform extracts the upper bound of voltage fluctuation and the expected voltage value from the matrix, calculates the difference, and then divides it by the expected voltage value. Physically, this is a dimensionless ratio representing the voltage's "noise-to-signal ratio." This ratio is defined as the upper limit of the Pearson correlation coefficient, i.e., the decision boundary value. The principle is that if the correlation coefficient between the target current and the non-target voltage is less than the original noise proportion of the voltage, it indicates that the correlation is extremely weak, a random phenomenon, and there is no risk of electrical coupling. The threshold calculation is quantified using the following formula:

[0134] ;

[0135] ;

[0136] in: The boundary value representing the neighborhood of the zero-value determination; This is the upper bound of the current fluctuation in the baseline state matrix; This represents the expected value of the current. The boundary value representing the determination of the security association interval; This is the upper bound of voltage fluctuations in the baseline state matrix; This represents the expected voltage value.

[0137] For example, suppose the baseline state matrix data read by the wireless transmission monitoring platform is as follows: Expected current value A, Upper bound of current fluctuation A; Expected voltage value V, upper bound of voltage fluctuation V. Perform calculation: Determine the neighborhood of the zero value. A. Security Association Determination Boundary Values This means that as long as the residual current is less than 0.05A and the absolute value of the Pearson coefficient is less than 0.001, the condition is considered met.

[0138] When the absolute value of the feeder output current is less than the zero value of the neighborhood and the absolute value of the Pearson correlation coefficient is less than the boundary value of the safety association interval, the connection conditions are confirmed to be met.

[0139] At the end of the current ramp-up process, a real-time confirmation phase for takeover conditions is initiated. The purpose is to double-confirm that "the old path is broken" and "the new path is uninterrupted." The wireless transmission monitoring platform continuously monitors two real-time indicators: the original DC system feeder output current (i.e., the residual current flowing through the original switch) and the Pearson correlation coefficient (calculated in real-time). The logic comparison unit compares the absolute value of the original DC system feeder output current with zero to determine the neighborhood. A comparison was performed. Simultaneously, the absolute value of the Pearson correlation coefficient was compared with the boundary value of the safe association interval. A comparison is performed. Only when both "less than" conditions are met simultaneously is the current state determined to be "safe takeover state". This indicates that the original branch current has been substantially reduced to zero, and the operation has not caused statistically significant electrical interference to adjacent branches.

[0140] The original DC system feeder output switch is tripped to cut off the power supply path of the original DC system to the power supply branch, and the insulating spacer is pulled out from the signal branch to restore the physical connection.

[0141] Once the takeover conditions are confirmed by the logic unit, the operator is immediately notified or the physical disconnection action is automatically executed. For power supply branches, since the current flowing through the original feeder output switch is now less than the background noise, falling within the microamp or milliamp range, disconnecting the switch will not generate an arc, achieving "arc-free tripping." The operator or automatic actuator moves the switch to the "open" position, completely severing the physical connection between the original DC system and the branch. For signal branches, the operator pulls out the previously inserted insulating spacer from the gap in the remote signaling terminal. As the insulating spacer is removed, the crimping tab below the terminal screw re-contacts the original conductor under the action of elasticity, thus instantly restoring the physical continuity of the original signal circuit without disconnecting the wire.

[0142] The wireless transmission monitoring platform generates structured messages and sends them to the dispatch center via wireless communication.

[0143] The application layer software of the wireless transmission monitoring platform automatically collects key metadata for this operation, including: operation task ID, baseline state matrix parameters, final Pearson coefficient value, and timestamp of the switch opening time. The platform encapsulates this data into a standard structured message format, such as JSON or XML. Subsequently, it drives the onboard wireless communication module to send the message to the remote dispatch center master station system via an encrypted channel on the power grid or public network. After receiving and parsing the message, the dispatch center automatically updates the power grid topology map, marking the branch as "maintenance / takeover" status, signifying the successful completion of this uninterrupted power replacement process.

[0144] Example 2: A substation DC system uninterrupted power replacement system, used to implement the method in Example 1, such as... Figure 4 As shown, the system includes:

[0145] The platform networking and baseline construction module is used to configure the uninterrupted power supply transformation platform. The uninterrupted power supply transformation platform includes a mobile DC charging platform, a mobile DC feeder platform, a mobile battery pack, and a wireless transmission monitoring platform. The platform networking is completed through electrical connection components. The wireless transmission monitoring platform is controlled to collect the bus voltage sequence and branch current sequence of the original DC system, and calculate the components and quantiles to construct the baseline state matrix.

[0146] The branch access and isolation fixing module is used to identify the power supply branch and signal branch of the original DC system. For the power supply branch, the plug-in wire is used to establish an electrical connection with the mobile DC feeder platform, and a through-core current transformer is configured to map the branch current signal in real time. For the signal branch, the plug-in wire is connected through the terminal extension method, and physical isolation and anti-detachment constraints are implemented using insulating spacers and wire clamps.

[0147] The perturbation injection and conduction determination module is used to control the mobile DC feeder platform to inject perturbation pulse signals into the power supply branch, collect transient response waveforms and calculate the load decay time constant to generate current ramp-up command sequence, apply a small-amplitude test current and collect terminal voltage difference. When the baseline state matrix determines that it falls into the voltage distribution signal interval, a branch conduction flag is generated.

[0148] The current ramp-up and voltage update module is used to respond to the branch conduction flag and execute the current ramp-up command sequence. It synchronously collects the target branch current, feeder output current and non-target branch voltage sequence of the original DC system. It dynamically updates the output voltage of the mobile DC charging platform based on the difference of the feeder output current of adjacent samples and calculates the Pearson correlation coefficient between the target branch current and the non-target branch voltage.

[0149] The correlation verification and cut-out uploading module is used to disconnect the feeder output switch of the original DC system and remove the insulating partition when the Pearson correlation coefficient is within the safe correlation range of the baseline state matrix and the feeder output current converges to the zero value judgment neighborhood of the baseline state matrix. The module then sends the completed message to the control unit via the wireless transmission monitoring platform.

Claims

1. A method for uninterrupted power supply replacement of a DC system in a substation, characterized by the following steps: include: Configure an uninterrupted power supply retrofit platform, including a mobile DC charging platform, a mobile DC feeder platform, a mobile battery pack, and a wireless transmission monitoring platform. The platform is networked through electrical connection components. The wireless transmission monitoring platform is controlled to collect the bus voltage sequence and branch current sequence of the original DC system, and calculate the components and quantiles to construct a baseline state matrix. Identify the power supply branch and signal branch of the original DC system. For the power supply branch, establish an electrical connection to the mobile DC feeder platform using plug-in wires and configure a through-core current transformer to map the branch current signal in real time. For the signal branch, connect the plug-in wires through the terminal extension method and implement physical isolation and anti-detachment constraints using insulating spacers and wire clamps. The mobile DC feeder platform is controlled to inject micro-perturbation pulse signals into the power supply branch, acquire transient response waveforms and calculate load decay time constant to generate current ramp command sequence, apply micro-amplitude test current and acquire terminal voltage difference, and generate branch conduction flag when the baseline state matrix determines that it falls into the voltage sub-position signal interval. Responding to the branch conduction flag and executing the current ramp-up command sequence, the target branch current, feeder output current and non-target branch voltage sequence of the original DC system are collected synchronously. The output voltage of the mobile DC charging platform is dynamically updated based on the difference of the feeder output current between adjacent samples, and the Pearson correlation coefficient is calculated for the target branch current and non-target branch voltage. When the Pearson correlation coefficient is within the safe correlation range of the baseline state matrix, and the feeder output current converges to the zero value judgment neighborhood of the baseline state matrix, disconnect the feeder output switch of the original DC system and remove the insulating partition, and send the takeover completion message through the wireless transmission monitoring platform.

2. The method for uninterrupted power supply replacement of a substation DC system according to claim 1, characterized in that, Solving for components and quantiles to construct the baseline state matrix includes: By using a one-click connection combination, a DC power transmission channel is established between the mobile DC charging platform, the mobile DC feeder platform, and the mobile battery pack, and a wireless transmission monitoring platform is connected via a communication bus. The analog signals from the original DC system are received through the wireless transmission monitoring platform, and signal filtering and discretization are performed to generate bus voltage sequence and branch current sequence. Steady-state components are extracted using bus voltage sequences and branch current sequences. Variance is calculated using bus voltage sequences and branch current sequences. Numerical boundaries are defined based on the variance. Baseline state matrix is ​​constructed by combining steady-state components and numerical boundaries.

3. The method for uninterrupted power supply replacement of a substation DC system according to claim 2, characterized in that, The baseline state matrix is ​​constructed by combining steady-state components and numerical boundaries, including: The bus voltage sequence and branch current sequence are subjected to an arithmetic average operation over the entire time period to generate the expected voltage and current values ​​as steady-state components. The standard deviations of the bus voltage series and branch current series are calculated. With the steady-state component as the center, three times the standard deviation are superimposed to construct the upper bound of voltage fluctuation, the lower bound of voltage fluctuation, the upper bound of current fluctuation, and the lower bound of current fluctuation, which serve as numerical boundaries and are written into the baseline state matrix.

4. The method for uninterrupted power supply replacement of a substation DC system according to claim 1, characterized in that, Physical isolation and anti-detachment constraints implemented using insulating spacers and wire clamps include: For the power supply branch, select a plug-in type wire with a protective spring, insert the pin end of the plug-in type wire into the test hole of the current terminal block of the power supply branch, and pass the plug-in type wire through the center hole of the through-core transformer to connect to the mobile DC feeder platform to establish an electrical connection. For signal branches, insert insulating spacers on the front and back sides of the wiring position in the remote signaling terminal of the signal branch. Connect the plug-in wire to the remote signaling terminal while retaining the original circuit wiring state, and use the wire clamp to lock the plug-in wire to the switch cabinet door or workbench side of the original DC system.

5. The method for uninterrupted power supply replacement of a substation DC system according to claim 3, characterized in that, The generation of branch continuity flags includes: The mobile DC feeder platform is driven to output short-time voltage pulses and synchronously acquire feedback voltage sequences. The time required for the feedback voltage sequence to decay from the peak to half the peak value is calculated and used as the load decay time constant. Based on the load decay time constant, the current ramp-up slope is indexed from the preset strategy table. When the load decay time constant is greater than the capacitive characteristic threshold of the strategy table, an S-shaped ramp-up curve is established; otherwise, a linear ramp-up curve is established as the current ramp-up command sequence. The mobile DC feeder platform is controlled to output a constant current test current, and the terminal voltage difference between the plug-in wire and the original DC system is collected. When the absolute value of the terminal voltage difference is less than the difference between the upper limit of the voltage fluctuation and the steady-state component in the baseline state matrix, a branch conduction flag is generated.

6. The method for uninterrupted power supply replacement of a substation DC system according to claim 5, characterized in that, The time required for the feedback voltage sequence to decay from its peak to half its peak value is calculated and used as the load decay time constant, including: Traverse the feedback voltage sequence, lock the sampling point with the largest value as the peak amplitude, and extract the timestamp of the sampling point as the peak time; In the feedback voltage sequence, the sampling point located after the peak moment is retrieved, and the sampling point with a value of half the peak amplitude is locked as the half-peak moment. The time difference between the half-peak moment and the peak moment is calculated as the load attenuation time constant.

7. The method for uninterrupted power supply replacement of a substation DC system according to claim 1, characterized in that, The calculation of the Pearson correlation coefficient between the target branch current and the non-target branch voltage includes: The mobile DC feeder platform is driven to gradually increase the injected current according to the current ramp-up command sequence, while the target branch current, feeder output current and non-target branch voltage sequence of the original DC system are recorded synchronously through the wireless transmission monitoring platform. The difference between the current and previous values ​​is calculated based on the feeder output current. This difference is then added to the current output voltage of the mobile DC charging platform to perform dynamic voltage updates. Calculate the covariance of the target branch current and the non-target branch voltage sequence, calculate the product of the standard deviations of the target branch current and the non-target branch voltage sequence, and divide the covariance by the standard deviation product to obtain the Pearson correlation coefficient.

8. The method for uninterrupted power supply replacement of a substation DC system according to claim 7, characterized in that, The records of the target branch current, feeder output current, and non-target branch voltage sequences of the original DC system include: The current setpoint in the current ramp-up command sequence is analyzed, and the mobile DC feeder platform is adjusted to control the output current to rise in a step-like manner following the current setpoint. By using a wireless transmission monitoring platform to issue a synchronous sampling trigger command, the acquisition channel is controlled to simultaneously read analog signals and perform analog-to-digital conversion, directly acquiring the target branch current, feeder output current and non-target branch voltage sequences that are consistent in time.

9. A method for uninterrupted power supply replacement of a substation DC system according to claim 3, characterized in that, The takeover completion message sent via the wireless transmission monitoring platform includes: The difference between the upper bound of the current fluctuation and the expected value of the current is calculated based on the baseline state matrix and used as the zero value determination neighborhood. At the same time, the difference between the upper bound of the voltage fluctuation and the expected value of the voltage is calculated based on the baseline state matrix and divided by the expected value of the voltage to obtain the safety association interval. When the absolute value of the feeder output current is less than the zero value of the neighborhood and the absolute value of the Pearson correlation coefficient is less than the boundary value of the safety association interval, the connection conditions are confirmed to be met. The original DC system feeder output switch is tripped to cut off the power supply path of the original DC system to the power supply branch, and the insulating spacer is pulled out from the signal branch to restore the physical connection. The wireless transmission monitoring platform generates structured messages and sends them to the dispatch center via wireless communication.

Images