Mining power supply control method with wide voltage input

Abstract

The invention provides a mining power supply control method with wide voltage input, which comprises the following steps of: setting input capacitor voltage reference values of a control unit of a power module as a total output voltage reference value according to an actual output requirement of a mining power supply with wide voltage input; collecting an actual input voltage signal, an actual output current signal and an actual total output voltage of the power module; obtaining a given signal of a current loop by using a PI controller input into the grading ring; the single-neuron PI controller in the output voltage loop calculates a state variable to obtain the output quantity of the single-neuron PI controller at the next sampling moment; the difference value between the current loop output quantity of the control unit and the actual value is transmitted into a PI controller of the current loop, and an output signal of the current loop in the control unit is obtained; an output signal of the current loop is transmitted to a modulation drive circuit in the control unit, and finally a drive signal for controlling a power switch tube of the power module to work is generated. The problem that the PI parameter value cannot be adjusted timely is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of mine power supplies, and more specifically, to a control method for a mine power supply with wide voltage input. Background Art

[0002] Coal is still the basic energy source in China. During the coal production process, high-power electrical equipment such as coal shearers, roadheaders, and conveyors is required, and there are many power supply voltage levels for these electrical equipment. However, a single DC / DC power module cannot meet the power level and power supply voltage requirements of various electrical equipment.

[0003] Since the parameters of each power module in the DC / DC power supply system with wide voltage input cannot be exactly the same, this leads to differences in the output power of each power module, resulting in problems of uneven voltage of the input capacitor voltage values of each power module and instability of the total output voltage on the parallel side, reducing the service life. In response to this phenomenon, the mine-used DC / DC power supply with wide voltage input adopts a three-loop control strategy, an output voltage loop, an input voltage equalization loop, and a current loop, all using PI control. This method can ensure both input capacitor voltage equalization and output voltage stability. n power modules share a single output voltage loop and are easily affected by changes in the electrical performance of one or more power modules. At this time, the PI parameters of the output voltage loop are no longer applicable, and it is difficult to meet the purpose of real-time online tuning and optimizing control parameters following the changes in the power supply system. Summary of the Invention

[0004] The problem solved by the present invention is that in the three-loop control, the PI parameter values in the voltage loop controller are fixed and cannot be adjusted in a timely manner according to the working conditions.

[0005] To solve the above problems, the present invention provides a control method for a mine power supply with wide voltage input, and the mine power supply control method includes: Step 1: According to the actual output requirements of the mine power supply with wide voltage input, set the input capacitor voltage reference values of the control units of n power modules to be V in / n, and set the total output voltage reference value of the control unit to be V ref ; Step 2: Collect the actual input voltage values V in1 、V in2 、...、V inn of the n power modules, the actual output current values i o1 、i o2 、...、i on of the n power modules, and the actual total output voltage Vo of the mine power supply with wide voltage input; Step 3: Obtain the given value of the current loop using the PI controller of the input equalizing ring inside each of the control units; Step 4: The single-neuron PI controller in the output voltage loop receives the actual total output voltage and the total output voltage reference value at the previous sampling moment and at the current sampling moment k, calculates the obtained difference, which is converted into a state variable through a converter; the state variable enters the Hebb learning rule to correct the connection weight coefficient in the single-neuron PI controller. The connection weight coefficient adopts a normalization calculation method, uses the connection weight coefficient to perform weighted summation on the state variable, and multiplies by a gain K to obtain the output quantity output by the single-neuron PI controller at the next sampling moment ; K is the gain multiple of the neuron; Step 5: The sum of the output quantities of the current loops of the n control units is used as the given values i ref1 , i ref2 ,..., i refn , and are respectively subtracted from the actual output current values i o1 , i o2 ,..., i on of the n power modules collected in Step 2. The obtained differences are respectively input into the PI controller of the current loop, and after being operated by the PI controller of the current loop, the output signals of the current loops inside the n control units are obtained; Step 6: The output signals of the current loops inside the n control units are respectively input into the modulation drive circuits inside the n control units, and finally drive signals for controlling the power switching tubes of the power modules to work are generated.

[0006] Further, Step 3 specifically includes: inputting the difference obtained by subtracting the actual input voltage value V inn of the power module from the input capacitor voltage reference value V in / n into the input equalizing ring PI controller, and after being operated by the input equalizing ring PI controller, the given value of the output current loop is obtained.

[0007] Further, the calculation of converting the obtained difference into a state variable is specifically calculated by the following formula: ; where is the difference between the actual output voltage at the current sampling moment and the reference output voltage , is the difference between the actual output voltage at the previous sampling moment and the reference output voltage ; is the current sampling moment and the previous sampling moment in the change amount of the difference between the actual total output voltage and the total output voltage reference; the state variables include an integral term state variable and a proportional term state variable, is the corresponding integral term state variable, is the corresponding proportional term state variable.

[0008] Further, the state variables enter the Hebb learning rule to correct the connection weight coefficients in the single-neuron PI controller, specifically including: ; wherein, is the integral term state variable weight coefficient at the current sampling moment ; is the proportional term state variable weight coefficient at the current sampling moment ; is the integral term state variable weight coefficient at the next sampling moment ; is the proportional term state variable weight coefficient at the next sampling moment ; is the proportional learning rate; is the learning rate of integration; is the voltage error correction parameter specified at the current sampling moment ; is the output of the controller at the current sampling moment .

[0009] Further, the output quantity of the single-neuron PI controller at the next sampling moment is obtained, and the specific calculation formula is as follows: ; wherein, i takes 1 or 2. When i = 1, it corresponds to the integral term state calculation. When i = 2, it corresponds to the proportional term state calculation.

[0010] Further, the state variables are weighted and summed using the weight coefficients and multiplied by a gain K, and the specific calculation formula is as follows: ; wherein, is the output quantity of the single-neuron PI controller at the current sampling moment ; is the output quantity of the single-neuron PI controller output at the next sampling moment .

[0011] Further, the topology structure of a single power module of the mine power supply with wide voltage input is an isolated single-phase three-level H-bridge converter.

[0012] Further, the input ends of the n power modules are connected in series, and the output ends are connected in parallel.

[0013] Further, the n control units share one output voltage loop.

[0014] In summary, each of the above technical solutions of the present application may have one or more of the following advantages or beneficial effects: i) In the three-loop control strategy, the input voltage equalization loop and current loop of the n power modules can be based on their own input and output information, and modular design can be achieved without any mutual information exchange, with high system reliability; ii) The single-neuron voltage loop can adjust the parameter values of the controller online, which greatly improves the efficiency of the controller and meets the requirements of system real-time control; The combination of the single neuron and the traditional PI control constitutes a single-neuron PI control method with adaptive PI parameters; This control method obtains the self-learning ability of the neural network, enables the neural network to follow the system changes and correct the connection weights of the neurons in real time, and the controller obtains an adaptive function; The single-neuron PI control structure is simple, can quickly adapt to system parameter changes, and has strong anti-interference ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a control principle block diagram of a mine-used DC / DC power supply with wide voltage input provided by the present invention; Figure 2 is a block diagram of the output voltage loop based on single-neuron PI control of the present invention; Figure 3 is a topological structure diagram of the mine-used DC / DC power supply with wide voltage input of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The purpose of the present invention is to provide a control method for a mine-used power supply with wide voltage input, so as to achieve the effect that the single-neuron PI control structure is simple, can quickly adapt to system parameter changes, and has strong anti-interference ability.

[0017] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention is made with reference to the accompanying drawings.

[0018] See Figures 1 - 3 , the present invention provides a control method for a mine-used power supply with wide voltage input, and the mine-used power supply control method includes: Step 1: According to the actual output requirements of the mine-used power supply with wide voltage input, set the input capacitor voltage reference values of the control units of the n power modules to be V in / n, and set the total output voltage reference value of the control unit to be V ref ; Step 2: Collect the actual input voltage values V in1 , V in2 ,..., V inn of the n power modules, the actual output current values i o1 , i o2 ,..., i on of the n power modules, and the actual total output voltage Vo of the mine power supply with wide voltage input; Step 3: Obtain the given value of the current loop by using the PI controller of the input voltage equalization loop inside each control unit; Step 4: The single-neuron PI controller in the output voltage loop receives the actual total output voltage and the total output voltage reference value at the previous sampling moment and the current sampling moment k, calculates the obtained difference, converts it into a state variable through a converter; the state variable enters the Hebb learning rule to correct the connection weight coefficient in the single-neuron PI controller, adopts a normalization calculation method for the connection weight coefficient, performs weighted summation on the state variable by using the connection weight coefficient, and multiplies it by K times to obtain the output quantity output by the single-neuron PI controller at the next sampling moment ; K is the gain multiple of the neuron; Step 5: The sum of the output quantities of the current loops of the n control units is used as the given values i ref1 , i ref2 ,..., i refn , and are respectively subtracted from the actual output current values i o1 , i o2 ,..., i on of the n power modules collected in Step 2. The obtained differences are respectively input into the PI controllers of the current loops, and after being operated by the PI controllers of the current loops, the output signals of the current loops inside the n control units are obtained; Step 6: The output signals of the current loops inside the n control units are respectively input into the modulation drive circuits inside the n control units, and finally drive signals for controlling the power switch tubes of the power modules are generated.

[0019] It should be noted that the input ends of the n power modules are connected in series, and the output ends are connected in parallel; the n control units share one output voltage loop. The given value received by the input voltage equalization loop of each control unit is the difference between the actual input voltage value of the power module corresponding to this control unit and the input capacitor voltage reference value; the received signal of the output voltage loop is the difference between the actual total output voltage on the parallel side and the total output voltage reference value; the current loop receives the output signal of the input voltage equalization loop, the output signal of the output voltage loop, and the actual output current value of the power module corresponding to this control unit; the output voltage loop utilizes the self-learning and adaptive capabilities of the single neuron to enable the PI controller to obtain the function of online real-time adjustment of the PI parameter value, enhance the anti-interference ability of the system, and at the same time solve the problems of uneven voltage of the series capacitors on the input side and unstable output voltage on the output side of the mine-used DC / DC power supply system.

[0020] In the three-loop control strategy, the input voltage equalization loops and current loops of the n power modules can be modularly designed according to their own input and output information without any mutual information exchange, and have high system reliability.

[0021] In a specific embodiment, step 3 specifically includes: inputting the difference obtained by subtracting the input capacitor voltage reference value, both of which are V inn / n, from the actual input voltage signal V in of the power module into the input voltage equalization loop PI controller, and obtaining the given signal of the output current loop after the operation of the input voltage equalization loop PI controller.

[0022] In a specific embodiment, the calculated difference is converted into state variables through a converter, and the specific calculation formula is as follows: ; Among them, is the difference between the actual output voltage at the current sampling moment and the reference output voltage , is the difference between the actual output voltage at the previous sampling moment and the reference output voltage ; is the change amount of the difference between the actual total output voltage and the total output voltage reference at the current sampling moment and the previous sampling moment ; the state variables include an integral term state variable and a proportional term state variable, is the corresponding integral term state variable, is the corresponding proportional term state variable.

[0023] In a specific embodiment, the state variable enters the Hebb learning rule to correct the connection weight coefficient in the single-neuron PI controller, specifically including: ; wherein, is the integral term state variable weight coefficient at the current sampling moment ; is the proportional term state variable weight coefficient at the current sampling moment ; is the integral term state variable weight coefficient at the next sampling moment ; is the proportional term state variable weight coefficient at the next sampling moment ; is the proportional learning rate; is the integral learning rate; is the voltage error correction parameter specified at the current sampling moment ; is the output of the controller at the current sampling moment .

[0024] In a specific embodiment, the output of the single-neuron PI controller at the next sampling moment is obtained, and the specific calculation formula is as follows: ; where i takes 1 or 2. When i = 1, it corresponds to the integral term state calculation. When i = 2, it corresponds to the proportional term state calculation.

[0025] In a specific embodiment, the state variables are weighted and summed using the weight coefficients and multiplied by a gain factor K. The specific calculation formula is as follows: ; wherein, is the output of the single-neuron PI controller at the current sampling moment ; is the output of the single-neuron PI controller at the next sampling moment .

[0026] The single-neuron voltage loop can adjust the parameter values of the controller online, which greatly improves the efficiency of the controller and meets the requirements of system real-time control. The combination of the single neuron and the traditional PI control constitutes a single-neuron PI control method with adaptive PI parameters. This control method obtains the self-learning ability of the neural network, enabling the neural network to correct the connection weights of the neurons in real time following the system changes, and the controller obtains the adaptive function. The single-neuron PI control has a simple structure, can quickly adapt to the changes of system parameters, and has strong anti-interference ability.

[0027] In a specific embodiment, the topology structure of a single power module of the mine power supply with wide-voltage input is an isolated single-phase three-level H-bridge converter.

[0028] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be subject to the scope defined by the claims.

Claims

1. A mining power supply control method with wide voltage input, characterized in that: The mining power supply control method comprises: Step 1: According to the actual output requirements of the wide voltage input mining power supply, set the input capacitor voltage reference value of the control unit of n power modules to V in / n, and setting the total output voltage reference value of the control unit to V ref ; Step 2: Collect the actual input voltage values ​​V of n power modules in1 、V in2 , ..., V inn , the actual output current value i of the n power modules o1 、i o2 、...、i on , and the actual total output voltage Vo of the mining power supply with wide voltage input; Step 3: using the PI controller of the input voltage grading loop inside each of the control units to obtain a given value of the current loop; Step 4: The single neuron PI controller in the output voltage loop receives the previous sampling time The difference between the actual total output voltage at the current sampling time k and the total output voltage reference value is calculated and converted into a state variable by the converter; the state variable enters the Hebb learning rule to correct the connection weight coefficient in the single neuron PI controller, and the connection weight coefficient is calculated in a normalized manner. The state variable is weighted and summed using the connection weight coefficient, and the gain is K times to obtain the next sampling time. The output of the single neuron PI controller when ; K is the gain multiple of the neuron; Step 5: The sum of the output quantities of the current loops of the n control units is taken as a given value i ref1 、i ref2 、...、i refn and the actual output current value i of the n power modules collected in step 2 o1 、i o2 、...、i on Subtracting each other respectively, and the obtained differences are respectively transmitted to the PI controller of the current loop, and after being calculated by the PI controller of the current loop, the output signals of the n current loops inside the control unit are obtained; Step 6: The output signals of the current loops in the n control units are respectively transmitted to the modulation drive circuits in the n control units, and finally a drive signal for controlling the operation of the power switch tube of the power module is generated.

2. The mining power supply control method according to claim 1, characterized in that: The step 3 specifically includes: setting the actual input voltage value V of the power module inn and the input capacitor voltage reference value are both V in The difference obtained by subtracting θ from θ is transmitted to the input voltage balancing loop PI controller, and after being calculated by the input voltage balancing loop PI controller, the given value of the output current loop is obtained.

3. The mining power supply control method according to claim 1, characterized in that: The calculated difference is converted into a state variable by a converter, and the specific calculation formula is as follows: ； in, is the current sampling time The actual output voltage With reference output voltage The difference, is the previous sampling time The actual output voltage With reference output voltage The difference between is the current sampling time The previous sampling moment The difference between the actual total output voltage and the total output voltage reference; the state variable includes an integral term state variable and a proportional term state variable, is the corresponding integral term state variable, is the corresponding proportional term state variable.

4. The mining power supply control method according to claim 3, characterized in that: The state variable enters the Hebb learning rule to correct the connection weight coefficient in the single neuron PI controller, specifically including: ； in, is the current sampling time The integral state variable weight coefficient of ; is the current sampling time The weight coefficient of the proportional state variable; is the next sampling time The integral state variable weight coefficient of ; is the next sampling time The weight coefficient of the proportional state variable; is the proportional learning rate; is the learning rate of integration; is the current sampling time Specified voltage error correction parameters; is the current sampling time The output of the controller.

5. The mining power supply control method according to claim 4, characterized in that: The next sampling time is obtained The output of the single neuron PI controller is calculated as follows: ； Among them, i is 1 or 2, i=1 corresponds to the integral term state calculation, and i=2 corresponds to the proportional term state calculation.

6. The mining power supply control method according to claim 5, characterized in that: The weight coefficient is used to perform weighted summation on the state variables and gain K times. The specific calculation formula is as follows: ； in, is the current sampling time The output of the single neuron PI controller; is the next sampling time The output quantity of the single neuron PI controller.

7. The mining power supply control method according to claim 1, characterized in that: The topological structure of a single power module of the wide voltage input mining power supply is an isolated single-phase three-level H-bridge converter.

8. The mining power supply control method according to claim 1, characterized in that: The input ends of the n power modules are connected in series, and the output ends are connected in parallel.

9. The mining power supply control method according to claim 1, characterized in that: The n control units share one output voltage loop.

Images