A local energy-saving dispatching device and method based on power fingerprinting and electricity price collaborative decision-making.

By employing a collaborative decision-making method combining a power sampling unit and a mechanical DIP switch in a socket-level scenario, local real-time coordination of load identification and scheduling priority is achieved. This solves the problems of high hardware cost and network dependence in existing technologies, and realizes energy-saving scheduling with high reliability and convenience.

CN123121692APending Publication Date: 2026-11-17NANTONG INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611395053.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-09
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies struggle to achieve local real-time coordination in load identification, peak-valley pricing time period determination, and scheduling priority adjudication in residential peak-valley electricity pricing scenarios. This results in high hardware costs, high computing power requirements, and reliance on network communication, making it difficult to apply in socket-level scenarios.

Method used

It employs a power sampling unit, a load feature extraction logic unit, a local decision parameter storage unit, and a scheduling execution unit. It performs load identification and scheduling priority determination in the absence of external network connection. It generates scheduling instructions by combining mechanical DIP switches and local electricity price information, and drives relays to perform on/off control.

Benefits of technology

It achieves a fully hardware closed loop for load identification and scheduling decision-making under the constraint of low-cost MCU, with high system reliability, convenient user operation, avoids scheduling failure caused by network interruption or cloud failure, and supports socket-level energy-saving scheduling.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123121692A_ABST
    Figure CN123121692A_ABST
Patent Text Reader

Abstract

This invention discloses a local energy-saving dispatching device and method based on power fingerprinting and electricity price collaborative decision-making, relating to the field of electrical energy control technology. The method includes: after power-on, the dispatching execution unit defaults to closed-loop mains power output; it queries the corresponding default dispatching priority based on the device identity; and, combined with the user settings of the priority setting subunit and the current time-period electricity price information in the local configuration storage subunit, independently generates dispatching instructions and drives the dispatching execution unit to perform on / off control in the absence of an external network connection. This invention, through three-dimensional input of power sampling output, pre-stored electricity price rules, and hardware-set priority via DIP switches, can automatically query the device identity, determine the final priority, and generate dispatching instructions based on the current electricity price period, executing device on / off control. This forms a complete hardware closed loop locally, from perception to identification decision-making to execution, exhibiting good system reliability and user convenience.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of electrical energy control technology, and in particular to a local energy-saving dispatching device and method based on power fingerprint and electricity price collaborative decision-making. Background Technology

[0002] With the full implementation of peak-valley time-of-use pricing policies for residential electricity, using electricity price signals to guide users to proactively adjust their electricity consumption has become an important means of reducing household electricity costs and alleviating grid supply and demand pressures. Against this backdrop, intelligent electricity management technology based on power load identification (power fingerprinting) has attracted widespread attention. Its core idea is to automatically identify appliance types by collecting and analyzing the electrical characteristics of connected loads, and then implement on / off control of their operating status according to electricity price periods, thereby achieving the automatic transfer of high-energy-consuming, delayable, and interruptible loads to off-peak periods.

[0003] Existing energy-saving control solutions for residential peak-valley electricity pricing scenarios mainly fall into the following technical categories: First, high-performance processors combined with cloud-based methods are used for power load identification and scheduling decisions. This approach has high hardware costs and relies on network communication; a network outage results in the loss of closed-loop control capability. Second, local load identification is achieved at the meter level using frequency domain transformation and neural network models. However, this requires high computing power, and the identification results are only used for passive metering, failing to form a control loop with actuators such as relays. Third, non-intrusive algorithms are used at the inlet bus level to decompose multiple loads. The priority allocation is a post-processing algorithm strategy, not a scheduling execution priority based on relay on / off states, and it does not involve electricity price storage or actuators. However, these technologies are limited by high-cost hardware, high computing power constraints, and network communication limitations, making them difficult to apply in socket-level scenarios. Therefore, a technical solution is urgently needed that can achieve local real-time coordination of load identification, peak-valley electricity price period determination, and scheduling priority decision-making to address these issues. Summary of the Invention

[0004] The purpose of this invention is to solve the problems in the prior art by proposing a local energy-saving dispatching device and method based on power fingerprint and electricity price collaborative decision-making.

[0005] A local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making includes: The power sampling unit is configured to collect voltage, current and power information of the connected load at fixed intervals; The local controller has a built-in load feature extraction logic unit, which is configured to extract the peak start-up current, start-up duration, steady-state power value and steady-state power factor of the connected load based on the output of the power sampling unit when there is no external network connection, and compare them with the locally pre-stored power fingerprint database to determine the device identity. The local decision parameter storage unit includes a local configuration storage subunit and a priority setting subunit; the local configuration storage subunit non-volatilely stores the peak-valley electricity price time period division rules and the mapping relationship between device identity and default scheduling priority; the priority setting subunit is a mechanical DIP switch used by the user to set the scheduling priority. The scheduling execution unit is electrically connected to the local controller and is used to perform on / off control according to the scheduling instructions generated by the local controller. The local interaction unit, electrically connected to the local controller, includes a status indication subunit and a time display and calibration subunit; the status indication subunit is used to output visual indications of the device's operating status and scheduling execution status; the time display and calibration subunit is used to display the current time and provide users with local time calibration when the real-time clock loses synchronization. The local controller is configured such that: after the device is powered on, the scheduling execution unit defaults to closing the mains power output; according to the device identity, it queries the corresponding default scheduling priority, combines the user settings of the priority setting subunit and the current time period electricity price information in the local configuration storage subunit, and independently generates scheduling instructions and drives the scheduling execution unit to perform on / off control in the absence of external network connection.

[0006] Preferably, the local controller is configured to: read the status of the priority setting sub-unit at a fixed period, and when the type of load currently connected is identified, call the device identity default scheduling priority mapping table pre-stored in the local configuration storage sub-unit to obtain the corresponding default scheduling priority.

[0007] Preferably, the priority setting subunit is a mechanical DIP switch, whose position is physically maintained by a mechanical structure and does not require power. The DIP switch includes at least a fully OFF position and three individually ON positions. The fully OFF position of the DIP switch indicates that the user has not set a scheduling priority. The local controller reads the state of the DIP switch via GPIO and converts it into a priority level parameter.

[0008] Preferably, the local controller is configured to: initiate a preset confirmation window period and control the status indication subunit to issue a first prompt only when an edge event of the DIP switch changing from a non-full OFF position to a full OFF position is detected; if a confirmation action of the DIP switch being moved from full OFF to any valid position and then back to full OFF is detected within the confirmation window period, then the new load fingerprint enrollment mode is entered.

[0009] Preferably, the local controller is configured in the new load fingerprint entry mode as follows: the power sampling unit collects the peak value of the starting current, the starting duration, the steady-state power value and the steady-state power factor of the newly connected load, generates a new dimensionless fingerprint vector after normalization, and writes it into the fingerprint library of the local configuration storage subunit. The new load fingerprint entry mode will automatically exit and reset if no effective load is detected within a preset timeout period, while the original fingerprint database remains unchanged.

[0010] Preferably, the dimensionless fingerprint vector pre-stored in the local decision parameter storage unit is a four-dimensional vector, which is obtained by dividing the peak value of the starting current, the starting duration, the steady-state power value, and the steady-state power factor by their respective preset benchmark values.

[0011] This invention also proposes a local energy-saving dispatching method based on power fingerprinting and electricity price collaborative decision-making, which, based on the aforementioned local energy-saving dispatching device, includes the following steps: Voltage, current, and power information of the connected load are collected at fixed intervals; In the absence of external network connection, the peak starting current, starting duration, steady-state power value and steady-state power factor of the load are extracted, a dimensionless fingerprint vector is generated and compared with the locally stored power fingerprint database to determine the device identity; After the device is powered on and initialized, the control relay is closed by default to output mains power; Based on the device identity, the locally stored default scheduling priority is queried, and combined with the user's hardware priority settings, the final priority is determined; Dispatch instructions are generated based on the final priority and the current electricity price information, and on / off control is executed.

[0012] Preferably, the final priority of the ruling includes: When the user hardware priority setting is consistent with the default scheduling priority, that priority shall be used as the final priority; If the user hardware priority setting is inconsistent with the default scheduling priority, the user hardware priority setting shall be the final priority, and a user-set priority prompt shall be issued.

[0013] Preferably, the step of generating dispatch instructions based on the final priority and the current time period electricity price information includes: Ultimate priority is the necessary level: regardless of peak or valley periods, the relay remains closed; The final priority is the delay-level: the relay is disconnected after a preset time during peak hours and power is automatically restored during off-peak hours; The final priority is interruptible: the relay is disconnected after a preset time during peak hours and power is automatically restored during off-peak hours; When the final priority is interruptible and the peak operating power is greater than the preset overload power, the relay is immediately disconnected without any delay limit.

[0014] Compared with existing technologies, the advantages of this invention are: 1. This invention uses three-dimensional inputs—power sampling output, pre-stored electricity price rules, and DIP switch hardware priority settings—to automatically query device identity, determine final priority, and generate scheduling instructions based on the current electricity price period to execute device on / off control. This forms a complete hardware closed loop locally, from perception to identification and decision-making to execution, resulting in good system reliability and ease of use for users.

[0015] 2. Under the constraint of low-cost MCU, this invention can complete load identification and scheduling decision-making in a short period of time by comparing four-dimensional time-domain power fingerprint features and normalized Euclidean distance. Moreover, all identification and decision-making logic is completed independently by the local controller without the need to connect to the cloud. This overcomes the dependence on high-performance processors and cloud support in the prior art, eliminates the risk of scheduling failure caused by network interruption or cloud failure, and realizes a commercially viable energy-saving scheduling solution at the socket level. Attached Figure Description

[0016] Figure 1 This is a block diagram of the overall system structure of the present invention.

[0017] Figure 2 This is a schematic diagram of the power sampling unit circuit in this invention.

[0018] Figure 3 This is a schematic diagram of the relay driving and execution unit circuit in this invention.

[0019] Figure 4 This is a flowchart of the scheduling and control process in this invention.

[0020] Figure 5 This invention describes the principle of a non-isolated AC220V to +3V3 power supply. Detailed Implementation

[0021] To facilitate understanding of this application and to make the above-mentioned objectives, features and advantages of this application more apparent, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0022] Reference Figure 1-5 As shown, a local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making includes: Power sampling unit: configured to collect the voltage, current, active power and power factor of the connected load at a fixed period of 250ms, and output electrical signals to the local controller.

[0023] The local controller is electrically connected to the power sampling unit, the local decision parameter storage unit, and the scheduling execution unit. It has a built-in load feature extraction logic unit configured to independently complete identification and scheduling decisions without an external network connection. This logic unit extracts four-dimensional features of the load: peak starting current, starting duration, steady-state power value, and steady-state power factor. These features are compared with a locally pre-stored power fingerprint database to identify the load type. The local controller is further configured as follows: after power-on initialization, the relay is closed by default to output mains power; the relay remains closed until load identification is complete; after identification, on / off control is executed based on the three-element collaborative judgment result; the local controller queries the locally pre-stored default scheduling priority (necessary / delayable / interruptible) based on the identified device identity, combines the user hardware settings of the priority setting subunit with the current time-period electricity price information from the locally configured storage subunit, and independently generates scheduling instructions without an external network connection; and updates the load's voltage, current, active power, and power factor with a 250ms sampling period via a timer interrupt.

[0024] In this context, power fingerprint refers to the identifiable power curve features exhibited by different household appliances during startup and steady-state operation due to differences in their internal electrical structures. It includes four dimensions: peak startup current Ip, startup duration Δt1, steady-state power value P, and steady-state power factor PF. The dimensionless fingerprint vector is a dimensionless four-dimensional vector obtained by dividing the above four original features by their respective preset benchmark values ​​(Ip / 30A, Δt1 / 500ms, P / 2500W, PF / 1.0), with each component ranging from 0 to 1. This four-dimensional time-domain feature selection and normalization method is a specific simplification implemented under the constraint of low-cost MCU computing power to complete the entire process of feature extraction, comparison, and decision-making.

[0025] Local decision parameter storage unit: electrically connected to the local controller, including a local configuration storage subunit and a priority setting subunit; the local configuration storage subunit is a serial EEPROM, which non-volatilely stores the peak-valley electricity price time period division rules and the default scheduling priority mapping table of device identity in the form of electrical signals; the priority setting subunit is a three-position mechanical DIP switch, whose state of each position is physically maintained by the mechanical structure and does not require power maintenance. After the local controller is powered on, it reads the mechanical state through GPIO and converts it into priority level parameters; the local controller is configured to read the DIP switch state in every 250ms sampling period; the local configuration storage subunit and the priority setting subunit are parallel decision input terminals, which together with the power sampling unit constitute the three-dimensional input of the local controller.

[0026] Scheduling execution unit: electrically connected to the local controller, used to execute on / off control according to the scheduling instructions independently generated by the local controller.

[0027] The local interaction unit includes a status indicator subunit and a time display and calibration subunit. The status indicator subunit includes a green LED (fingerprint enrollment completion prompt), a yellow LED (delayed execution / fingerprint enrollment stage prompt / inconsistency between user settings and default priority prompt), and a red LED (interruptible execution / RTC out-of-sync alarm prompt), each connected to the local controller GPIO through a current-limiting resistor. The time display and calibration subunit includes an LED digital tube module (CLK pin, DIO pin) and two touch buttons ("Mode" button, "Adjust" button), used to display the current time and the local time calibration after RTC out-of-sync, and does not involve any network communication functions.

[0028] Among them, the output of the power sampling unit (load identity), the time period rules of the local configuration storage subunit (electricity price time period), and the priority parameters of the priority setting subunit (scheduling priority) are comprehensively determined by the local controller in a three-element collaborative manner to generate scheduling instructions that directly drive the scheduling execution unit.

[0029] Specifically, the ternary synergy determination is as follows: ① The final priority is the necessary level: the relay remains closed regardless of peak or valley periods; ② The final priority is the delayable level: the relay is disconnected after a 3-5 second delay during peak hours + the yellow LED lights up (pending execution); power is automatically restored during off-peak hours; ③ The final priority is interruptible: the relay is disconnected after a 3-5 second delay during peak hours + the red LED lights up (pending execution); power is automatically restored during off-peak hours; When the final priority is interruptible and the peak operating power is greater than the preset overload power, the relay is immediately disconnected without any delay limit.

[0030] In the pending state, users can temporarily switch to the necessary level via DIP switch to restore power and solve urgent power needs.

[0031] Under normal operating conditions, when the load is shut down, the current scheduling ends, the status flag is cleared, the relay remains closed, and a new round of scheduling begins when a new load arrives.

[0032] The default scheduling priority mapping table binds the scheduling policy to the device identity determined by the identification result, rather than to a specific socket location, enabling device following and plug-and-play functionality.

[0033] Unified logic for the DIP switch's fully OFF state and fingerprint enrollment mode: In this invention, the fully OFF state of the three-position DIP switch indicates that the user has not set the hardware priority, and the device automatically uses the factory default mapping table for scheduling. This meaning remains consistent throughout all operating stages of the device.

[0034] The local controller monitors the DIP switch status in real time. Only when it detects an edge event of the DIP switch transitioning from a non-fully OFF position to a fully OFF position, it initiates a preset confirmation window (e.g., 5 seconds), during which a yellow LED flashes slowly at a frequency of 1Hz to alert the user. If the user moves the DIP switch from OFF to any valid position within the confirmation window and then quickly moves it back to OFF (completing one "dip-back" confirmation action), the device determines that the user has intentionally triggered the switch and formally enters the new load fingerprint enrollment mode, at which point the yellow LED becomes constantly lit for confirmation.

[0035] After entering the fingerprinting mode, the user connects a new load and starts operation. The local controller controls the power sampling unit to continuously collect the four-dimensional characteristics of the load—peak starting current, starting duration, steady-state power value, and steady-state power factor—at a period of 250ms. After normalization processing, a new dimensionless fingerprint vector is generated and automatically written into the fingerprint database in the EEPROM. After the fingerprint is written, a green LED flashes rapidly at a frequency of 2Hz to indicate successful entry to the user. The device automatically performs a soft reset and restarts, entering normal operation mode. If the device does not detect a jump in load power from 0 to >5W within 60 seconds after entering the fingerprinting mode, it automatically exits the fingerprinting mode and resets to normal operation mode, while the original fingerprint database remains unchanged.

[0036] Example 1 (I) Hardware composition and connection relationship Power sampling unit: like Figure 1 and Figure 2 The power sampling unit circuit diagram is shown. The power sampling unit uses the BL0937 power metering chip, operating at 3.3V, and is connected in series with a manganese-copper shunt in the mains live wire circuit. The BL0937's CF pin outputs a pulse signal with a frequency proportional to the active power; the CF1 pin outputs a pulse signal with a frequency proportional to either the effective current value (SEL=0) or the effective voltage value (SEL=1). The local controller controls the SEL pin level via GPIO pins, switching between the effective current value and the effective voltage value output in a time-division multiplexing manner. The local controller captures the pulse frequencies of the CF and CF1 pins at regular intervals, converts them using calibration coefficients to obtain the active power P, the effective voltage value Urms, and the effective current value Irms, and then calculates the power factor PF = P / (Urms × Irms) using the built-in load feature extraction logic unit.

[0037] The CF / CF1 / SEL signals of the BL0937 are electrically isolated from the local controller's GPIO via an optocoupler EL357N. The overall sampling accuracy of the power sampling unit is 1.0 class, with a refresh cycle of 250ms. The CF / CF1 / SEL signals are connected to the local controller's GPIO after optocoupler isolation. The input side of optocouplers U1 / U2 is powered by +3V3 / L_IN, and the output side is powered by +3.3V / VSS. The input side of optocoupler U3 is powered by +3.3V / VSS, and the output side is powered by +3V3 / L_IN, achieving secure signal transmission between the non-isolated sampling domain and the isolated control domain.

[0038] Local controller: The local controller uses an STM32F103C8T6, and the complete pin connection of the local controller is shown in the table below: Table 1 Local Controller Pin Connection Relationship

[0039] Real-time Clock (RTC) Module and Backup Mechanism: This embodiment uses the built-in RTC module of the STM32F103C8T6 chip, with the clock source provided by an external 32.768kHz passive crystal oscillator, and the VBAT pin is connected to a CR2032 button battery.

[0040] When the device powers on and initializes, the local controller reads the RTC second count value and compares it with the factory-set reference timestamp preset in the internal firmware. If the read RTC time is earlier than the factory reference timestamp (i.e., the RTC second counter resets due to VBAT depletion), it is determined that the RTC time has lost synchronization. At this time, the local controller forcibly classifies the current period as a peak period (the most conservative strategy), and the red LED flashes rapidly at a frequency of 3Hz to remind the user to replace the battery.

[0041] RTC local time calibration: When the RTC loses synchronization and triggers a red LED flashing alarm, the user can manually calibrate the RTC time through the time display and calibration subunit after disconnecting the input power and replacing the battery.

[0042] Local decision parameter storage unit: Local configuration storage sub-unit: AT24C02N serial EEPROM, non-volatile storage peak and valley time period division rules (peak time period 8:00-21:00, valley time period 21:00-next day 8:00) and device identity default scheduling priority mapping table.

[0043] Priority setting sub-unit: Three individual ON valid levels (pull-up 10kΩ to 3.3V), level definition: first ON = necessary level, second ON = delayed level, third ON = interruptible level.

[0044] Scheduling Execution Unit: The GPIO output of the local controller is connected to the transistor driver circuit after being isolated by an EL357N optocoupler. The relay used is HF115F / 005-1ZS3, and the maximum continuous load current of a single channel is designed not to exceed 10A.

[0045] Local Interaction Unit: The status indicator subunit consists of a green LED (PA8), a yellow LED (PA9), and a red LED (PA10) for indicating the device's operating status. The time display and calibration subunit uses a TM1637 4-digit LED digital tube module (CLK pin connected to PA13, DIO pin connected to PA14, VCC / GND connected to 3.3V / VSS power supply) and two touch buttons ("Mode" button connected to PA11, "Adjust" button connected to PA12).

[0046] This device is equipped with two independent power supplies: (1) Non-isolated AC220V to +3V3 power supply unit, such as Figure 5 As shown, the BP2525B chip is used, with an output of +3V3 (the negative terminal of the power supply is not isolated from the mains live wire L_IN and shares a common ground), which only supplies power to the BL0937 in the power sampling unit and its peripheral circuits; (2) General AC-DC dual output switching power supply module (isolation type), input AC220V, output +3.3V / VSS (for local controller, local decision parameter storage unit and local interaction unit) and +5V / GND (for optocoupler output side and relay coil of scheduling execution unit), its output ground VSS / GND is electrically isolated from the mains power side.

[0047] The key component parameters in this embodiment are shown in the table below: Table 2 Key Component Parameters

[0048] (II) Fingerprint Database Establishment and Identification Process The original features and normalized vectors of the power fingerprints for the four types of loads are shown in the table below: Table 3 Power fingerprints and dimensionless fingerprint vectors for four types of loads

[0049] The typical dimensionless fingerprint vectors of the above four loads are stored in the serial EEPROM pre-stored fingerprint library of the local decision parameter storage unit.

[0050] Recognition process: When the power surge exceeds 5W, if a load (>5W) is detected operating before the power surge, resulting in a single-circuit multi-load situation, it should be treated as an unknown load to ensure power supply and avoid accidental power outages.

[0051] If no load (≤5W) is detected before the power surge, then start identification: After detecting the power surge, start recording the characteristics of the start-up phase. ① Record Ip and Δt1 during the start-up phase (within a 500ms window); ② After the power change rate is <2% for 1s, record the steady-state P and PF.

[0052] The original four-dimensional features are normalized according to the baseline values: Ip / 30A, Δt1 / 500ms, P / 2500W, PF / 1.0, which is the dimensionless fingerprint vector {Ip′, Δt1′, P′, PF′}. Euclidean distance calculations are then performed between this vector and the four dimensionless fingerprint vectors in the fingerprint database.

[0053] The formula for calculating Euclidean distance (in four-dimensional space) is as follows:

[0054] The minimum Euclidean distance min(d) is taken. If min(d) < the preset judgment threshold (0.35 in this embodiment), it is identified as the corresponding load type. Based on the identified load type, the default scheduling priority mapping table of device identity is queried to obtain the default priority. The final priority is obtained according to the priority decision rules. All DIP switches are OFF (not set by the user), using the default priority; If the DIP switch setting is set to the default priority, then that priority will be used. DIP switch settings ≠ default priority; user-defined priority is used (yellow light flashes 3 times quickly). If min(d) ≥ threshold, it is judged as an unknown load and processed as necessary.

[0055] The judgment threshold is set based on the minimum Euclidean distance among common different types of loads. First, the Euclidean distance between different types of load components needs to be calculated. The Euclidean distances (experimentally calibrated) between common different types of loads are shown in the table below: Table 4. Euclidean distances between different load types

[0056] The decision threshold is set at 65% to 70% of the minimum distance between categories (0.50), striking a balance between avoiding misjudgment of similar loads and confusion between different load categories. In this embodiment, the decision threshold is set to 0.50 × 70% = 0.35.

[0057] New fingerprint enrollment mode: The local controller reads the DIP switch status every 250ms sampling period. Only when it detects an edge event of the DIP switch changing from a non-full OFF position to full OFF, it initiates a 5-second confirmation window, during which the yellow LED flashes slowly at a frequency of 1Hz. If the user moves the DIP switch from full OFF to any valid position and then quickly moves it back to full OFF within 5 seconds (completing the confirmation action), the device officially enters the data entry mode.

[0058] After entering the input mode, the local controller collects the four-dimensional features of the load, generates a new dimensionless fingerprint vector after normalization, and writes it to the EEPROM. If no effective load starting current is detected within 60 seconds, the input mode is automatically exited and the system is reset.

[0059] (III) Scheduling and Control Logic Device power-on initialization: Reads the electricity price rules and default mapping table from the local configuration storage sub-unit, reads the DIP switch priority, initializes the RTC, and the relay is closed by default to output mains power.

[0060] Power is continuously sampled at 250ms intervals (the DIP switch status is read each cycle). When the load starts and the power jumps from zero to >5W, the identification timing is started. Power is continuously supplied during the 500ms startup window and steady-state establishment period. The decision is then made after the identification is completed.

[0061] After load identification is completed, a ternary collaborative determination is performed: ① The final priority is the necessary level: the relay remains closed regardless of peak or valley periods; ② The final priority is the delayable level: the relay is disconnected after a 3-5 second delay during peak hours + the yellow LED lights up (pending execution); power is automatically restored during off-peak hours; ③ The final priority is interruptible: the relay is disconnected after a 3-5 second delay during peak hours + the red LED lights up (pending execution); power is automatically restored during off-peak hours; When the final priority is interruptible (overload protection): if the total peak operating power is >3000W, the relay will be disconnected immediately without any delay limit.

[0062] In the pending state, users can temporarily switch to the necessary level via DIP switch to restore power (solve urgent power needs).

[0063] Under normal operating conditions, the load is turned off (power <5W for 5 seconds), the status marker is cleared, and the relay remains closed and ready to operate.

[0064] Multi-load handling: This device is suitable for single-circuit, single-load scenarios. It is not recommended to connect multiple loads via a power strip. If a load is already running when a power surge is detected (power surge from a non-zero value), it is considered a multi-load connection. The device identifies it as an unknown load, processes it according to the necessary level, keeps the relay closed, ensures power supply, and avoids accidental disconnection.

[0065] After the relay is activated, a debounce delay of 100ms to 500ms is initiated.

[0066] (iv) Experimental verification plan 1. The test conditions and experimental platform are shown in the table below: Table 4 Test conditions and experimental platform

[0067] 2. Core Functionality Verification Verification method: The host computer can read the startup phase Ip, Δt1 and steady state P and PF from the USART debug interface. It can also read the identified load type and the corresponding default priority and final priority. If it is consistent with the actual situation, the identification is accurate. The same applies below.

[0068] Test 1: Delayed access + peak time access (2100W storage electric water heater, default priority delayed access), used to demonstrate the core scheduling logic.

[0069]

[0070] Test 2: Known load but default priority conflicts with DIP switch settings (1100W household air conditioner, default priority is interruptible level), used to demonstrate the user priority decision mechanism.

[0071]

[0072] Test 3: Multi-load access scenario (12W LED lighting, default interruptible priority and 320W drum washing machine, default delayed priority) to demonstrate multi-load processing logic.

[0073]

[0074] Test 4: Fingerprint enrollment mode (new load, microwave oven 1500W, default priority interruptible level), confirmation trigger and automatic exit, used to demonstrate the interaction of the DIP switch.

[0075]

[0076] 3. Stability test Experimental objective: To verify the stability and scheduling correctness of the device when connected to a storage-type electric water heater (2100W, default priority with delay level) for 72 hours of continuous operation.

[0077] Test method: The device was connected to AC power input, with the DIP switches set to all OFF (using default priority), and ran continuously for 72 hours. During this period, scenarios such as load start-up, load shutdown, and time period switching were performed, and an RTC step loss scenario was simulated (recovery occurred after disconnecting the VBAT battery for 5 minutes). The test results are shown in the table below: Table 5 Stability test results

[0078] Experimental conclusions: The device operated independently and stably for 72 hours without relying on the network, with no system crashes or resets, and an overall recognition accuracy of ≥96.7%. All scheduling logic was executed locally autonomously and was not affected by network conditions.

[0079] 4. Recognition accuracy experiment under voltage fluctuation conditions Experimental objective: To verify that the recognition accuracy can still meet engineering requirements under three voltage conditions: 198V / 220V / 242V.

[0080] Test method: Connect the device input to the output terminal of the autotransformer. Set the input voltage to 198V, 220V, and 242V respectively. Perform 10 identification tests for each of the three standard loads at each voltage (90 tests in total) according to the above method. The number of accurate identifications is shown in the table below.

[0081]

[0082] Experimental conclusion: Within a voltage fluctuation range of ±10%, the overall identification accuracy is ≥96.7%, and the device is robust to grid voltage fluctuations.

[0083] Example 2: Multi-loop expansion device In the three-output socket configuration, each channel is independently configured with the aforementioned power sampling unit, priority setting sub-unit, and relay drive circuit, while sharing the same local controller (STM32F103C8T6) and the same local decision parameter storage unit (AT24C02N EEPROM).

[0084] The local controller polls and samples each channel sequentially at a fixed period of 250ms. The power sampling data of each channel is independently buffered, and power fingerprint identification and scheduling determination are performed independently without interference. After any relay in the scheduling determination is activated, that channel independently starts a debouncing delay of 100ms to 500ms. During the delay, only the repeated switching of this circuit is prohibited, without affecting the normal sampling and determination of other circuits.

[0085] Multi-circuit fuse configuration: Each output terminal is independently equipped with a 10A slow-blow fuse, and the total input terminal is equipped with a 16A slow-blow fuse.

[0086] In summary, this invention employs a three-dimensional collaborative decision-making input comprised of load identification via power sampling output, peak-valley electricity price rules pre-stored in EEPROM, and scheduling priorities set by the DIP switch hardware. The local controller automatically queries the default priority corresponding to the device's identity and adjudicates it against user settings. Combined with the current electricity price period, it generates a final scheduling command and drives a relay to perform on / off control. Simultaneously, the physical holding characteristic of the mechanical DIP switch eliminates the need for power to maintain the set state, balancing user-defined control with low system power consumption requirements. The entire process does not rely on any network connection, cloud computing power, or predictive algorithms, enabling independent real-time decision-making on a low-cost, low-computing-power MCU. This overcomes the technical shortcomings of existing technologies, such as the disconnect between identification and scheduling, and failure upon network outage.

[0087] As is known from common technical knowledge, this invention can be implemented through other embodiments that do not depart from its spirit or essential characteristics. Therefore, the disclosed embodiments described above are merely illustrative in all respects and are not the only ones. All modifications within the scope of this invention or its equivalents are included in this invention.

Claims

1. A local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making, characterized in that, include: The power sampling unit is configured to collect voltage, current and power information of the connected load at fixed intervals; The local controller has a built-in load feature extraction logic unit, which is configured to extract the peak start-up current, start-up duration, steady-state power value and steady-state power factor of the connected load based on the output of the power sampling unit when there is no external network connection, and compare them with the locally pre-stored power fingerprint database to determine the device identity. The local decision parameter storage unit includes a local configuration storage subunit and a priority setting subunit; the local configuration storage subunit non-volatilely stores the peak-valley electricity price time period division rules and the mapping relationship between device identity and default scheduling priority; the priority setting subunit is a mechanical DIP switch used by the user to set the scheduling priority. The scheduling execution unit is electrically connected to the local controller and is used to perform on / off control according to the scheduling instructions generated by the local controller. The local interaction unit, electrically connected to the local controller, includes a status indication subunit and a time display and calibration subunit; the status indication subunit is used to output visual indications of the device's operating status and scheduling execution status; the time display and calibration subunit is used to display the current time and provide users with local time calibration when the real-time clock loses synchronization. The local controller is configured such that the scheduling execution unit defaults to closing the mains power supply after the device is powered on. Based on the device identity, the corresponding default scheduling priority is queried. Combined with the user settings of the priority setting subunit and the current time period electricity price information in the local configuration storage subunit, scheduling instructions are independently generated and the scheduling execution unit is driven to perform on / off control in the absence of external network connection.

2. The local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making as described in claim 1, characterized in that: The local controller is configured to: read the status of the priority setting sub-unit at a fixed period, and when the type of load currently connected is identified, call the device identity default scheduling priority mapping table pre-stored in the local configuration storage sub-unit to obtain the corresponding default scheduling priority; When the user settings differ from the default scheduling priority, the user settings take precedence.

3. A local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making as described in claim 1, characterized in that: The priority setting subunit is a mechanical DIP switch, whose position is physically maintained by a mechanical structure and does not require power. The DIP switch includes at least a fully OFF position and three individually ON positions. The fully OFF position of the DIP switch indicates that the user has not set a scheduling priority.

4. A local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making as described in claim 3, characterized in that: The local controller is configured to: initiate a preset confirmation window period and control the status indicator subunit to issue a first prompt only when an edge event of the DIP switch changing from a non-full OFF position to a full OFF position is detected; if a confirmation action of the DIP switch being moved from full OFF to any valid position and then back to full OFF is detected within the confirmation window period, then the new load fingerprint enrollment mode is entered.

5. A local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making according to claim 5, characterized in that: The local controller is configured in the new load fingerprint entry mode as follows: the power sampling unit collects the peak value of the starting current, the starting duration, the steady-state power value and the steady-state power factor of the newly connected load, generates a new dimensionless fingerprint vector after normalization, and writes it into the fingerprint library of the local configuration storage subunit. The new load fingerprint entry mode will automatically exit and reset if no effective load is detected within a preset timeout period, while the original fingerprint database remains unchanged.

6. A local energy-saving dispatching device based on power fingerprinting and electricity price collaborative decision-making as described in claim 1, characterized in that: The dimensionless fingerprint vector pre-stored in the local decision parameter storage unit is a four-dimensional vector, which is obtained by dividing the peak value of the starting current, the starting duration, the steady-state power value, and the steady-state power factor by their respective preset benchmark values.

7. A local energy-saving dispatching method based on power fingerprinting and electricity price collaborative decision-making, wherein the local energy-saving dispatching device described in any one of claims 1-6 is characterized in that, Includes the following steps: Voltage, current, and power information of the connected load are collected at fixed intervals; In the absence of external network connection, the peak starting current, starting duration, steady-state power value and steady-state power factor of the load are extracted, a dimensionless fingerprint vector is generated and compared with the locally stored power fingerprint database to determine the device identity; After the device is powered on and initialized, the control relay is closed by default to output mains power; Based on the device identity, the locally stored default scheduling priority is queried, and combined with the user's hardware priority settings, the final priority is determined; Dispatch instructions are generated based on the final priority and the current electricity price information, and on / off control is executed.

8. A local energy-saving dispatching method based on power fingerprinting and electricity price collaborative decision-making as described in claim 7, characterized in that: The final priority of the ruling includes: When the user hardware priority setting is consistent with the default scheduling priority, that priority shall be used as the final priority; If the user hardware priority setting is inconsistent with the default scheduling priority, the user hardware priority setting shall be the final priority, and a user-set priority prompt shall be issued.

9. A local energy-saving dispatching method based on power fingerprinting and electricity price collaborative decision-making as described in claim 7, characterized in that: The process of generating dispatch instructions based on the final priority and the current electricity price information includes: Ultimate priority is the necessary level: regardless of peak or valley periods, the relay remains closed; The final priority is the delay-level: the relay is disconnected after a preset time during peak hours and power is automatically restored during off-peak hours; The final priority is interruptible: the relay is disconnected after a preset time during peak hours and power is automatically restored during off-peak hours; When the final priority is interruptible and the peak operating power is greater than the preset overload power, the relay is immediately disconnected without any delay limit.