Method for preventing misoperation through voice recognition based on artificial intelligence

Abstract

The invention relates to a method for preventing misoperation through voice recognition based on artificial intelligence. The power station serves as a power generation link of power production, a power generation dispatching unit issues a working instruction according to power grid operation requirements, and the power station optimally distributes unit output of the power station according to the working instruction and executes startup and shutdown operation. At present, a working instruction is issued between a power generation dispatching unit and a power station through a recording telephone, a receiver cannot accurately understand all the content of the instruction and is interfered by pronunciation habits, voice definition, background noise, signal interference and the like, the receiver needs to repeatedly confirm the content, and the communication cost and the risk of misoperation are increased. On the basis of artificial intelligence voice recognition, a voice library is established to check identity information of an instruction sender and an instruction receiver, and operation instruction information, locking information of operation and operation sequence control information which are automatically called are compared with operation instruction information, locking information of operation and operation sequence control information which are input by the receiver and judged.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power dispatching, and in particular to a method for preventing misoperation based on artificial intelligence voice recognition. Background Art

[0002] As the power generation link of power production, the power station is issued work instructions by the power generation dispatching unit according to the grid operation requirements. The power station optimizes the output of the units of the power station according to the work instructions and performs start and stop operations. At present, most of the work instructions between the power generation dispatching unit and the power station are issued through recorded calls. However, since there is usually no instant confirmation mechanism on the phone, the receiver fails to accurately understand or remember the full content of the instructions. At the same time, due to the pronunciation habits, voice clarity, background noise or signal interference of the issuer, the receiver needs to confirm the content repeatedly. When issuing instructions on the phone, it is difficult to ensure that all relevant personnel can hear and understand the instructions at the same time, which increases the communication cost and the risk of misoperation. Summary of the invention

[0003] The technical problem of the present invention is to reduce the communication cost between the issuer and the receiver of the work instruction and improve the work efficiency through a method of preventing misoperation based on artificial intelligence voice recognition. At the same time, the work instruction recognized by the method is used to lock the receiver's subsequent operation and control the operation process to prevent misoperation and reduce the risk of misoperation.

[0004] The purpose of the present invention is to solve the above problems, and discloses a method for preventing misoperation based on artificial intelligence speech recognition, comprising the following steps: S1: Obtain the voice pronunciation of all personnel, use artificial intelligence to process the voice pronunciation and extract features to form a voice library; S2: Collect the identity information of the issuer and receiver of the work order issued by voice, and obtain the data set D1 after processing with the voice library in step S1; S3: The system collects the work instruction information issued by voice, and retrieves the operation instruction information, the operation locking information and the operation sequence control information according to the work instruction information to obtain the data set D2; S4: The system identifies the content of the telephone recording of the issuer and the receiver of the work order, and extracts the voice and identity information of the issuer and the receiver to obtain the data set D3, and compares the voice and identity information of D3 with D1; S5: The system extracts the content related to the work instruction from the telephone recording content, and sends the operation instruction information, operation locking information and operation sequence control information related to the work instruction retrieved from D2 to the receiver; S6: The recipient executes the work instruction issued by the issuer over the phone. During the execution of the operation instruction, the system records the voice of the recipient executing the operation instruction. The system compares the recorded operation instruction with the operation instruction retrieved in D2 in step S5 item by item, and judges the locking information and operation sequence control information until all items of the operation instruction are compared.

[0005] Preferably, in step S1, the voice database refers to recording the voice information of all personnel, extracting the features of the voice information, and classifying them into the senders and receivers of instructions.

[0006] Furthermore, in step S2, the identity information specifically refers to the name and work unit of the issuer and receiver of the work order, and the processed data set D1 refers to a data set in which a one-to-one correspondence is established between the voice information and the identity information of the issuer and receiver of the work order.

[0007] Preferably, in step S3, the work instruction information includes unit startup, shutdown, line power supply, power outage and equipment maintenance information; the operation instruction information is the operation steps and content information required to complete the work instruction; the locking information of a work instruction information involving one or more operation instruction information operations is the remaining unit, line, switch, and equipment information not involved in the work instruction; the operation sequence control information is the relationship information of the execution order between multiple operation instructions.

[0008] Furthermore, in step S3, operation instruction information, operation locking information and operation sequence control information are retrieved according to the work instruction information, including entering the operation method, operation steps, operation locking and operation sequence control of unit startup, shutdown, line power supply, power outage and equipment maintenance into the system, and the system automatically extracts the operation instruction information, operation locking information and operation sequence control information related to the work instruction according to the work instruction.

[0009] Furthermore, in step S4, the data set D3 is compared with the voice and identity information of the sender and receiver of the data set D1 in step S2, including the following sub-steps: S41: If the voice and identity information of the sender and the receiver are successfully compared, step S5 is executed; S42: If the comparison of the voice and identity information of the sender and the receiver is unsuccessful, the sender and the receiver are prompted to re-send and accept the work instructions, and step S4 is executed until the comparison is successful, and step S5 is continued.

[0010] Preferably, in step S5, the system identifies the content of the telephone recording of the issuer and receiver of the work instruction, extracts the content related to the work instruction, and retrieves the operation instruction information, operation locking information and operation sequence control information involved in the work instruction in D2, and sends this information to the receiver.

[0011] Preferably, in step S6, the system compares the input operation instruction with the operation instruction retrieved in D2 in step S5 item by item, and determines whether a certain operation step and content meets the locking condition and the sequence control condition, including the following sub-steps: S61: The receiver executes the operation instruction, and the system sequentially records the receiver's operation instruction voice; S62: The system compares the steps and contents of the input operation instruction with the operation instruction retrieved from the data set D2 for consistency. If they are inconsistent, S61 is executed; if they are consistent, S63 is executed; S63: The system compares the steps and contents of the input operation instruction with the locking information of the operation retrieved from the data set D2. If locked, S61 is executed; if not locked, S64 is executed. S64: The system compares the consistency of the operation sequence control information retrieved from the sequence data set D2 of the entered operation instructions. If the control information is inconsistent, S61 is executed; if the control information is consistent, step S61 is repeated to enter the next operation instruction until all operation instructions are compared.

[0012] Compared with the prior art, the beneficial effects of the present invention include: 1) The present invention discloses a method for preventing misoperation based on artificial intelligence voice recognition, which uses artificial intelligence to collect and identify the identities of the sender and the receiver, and compares the input operation instructions with the retrieved operation instructions item by item, thereby improving the safety of the maintenance work of the power station units and reducing the risk of misoperation. 2) The present invention reduces the communication cost between the issuer and the receiver of the work instruction and improves work efficiency.

[0013] 3) This method locks and determines the operation process control for the identified work instructions and the instructions entered by the recipient to prevent misoperation and reduce risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0015] Figure 1 is a flowchart of an embodiment of the present invention; DETAILED DESCRIPTION like Figure 1 As shown, for example, issuer A issues a work order to start the No. 2 generator set.

[0016] The method for preventing misoperation based on artificial intelligence speech recognition comprises the following steps: S1: Obtain the voice pronunciation of all personnel, use artificial intelligence to process the voice pronunciation and extract features to form a voice library; S2: Collect the identity information of the issuer and receiver of the work order issued by voice, and obtain the data set D1 after processing with the voice library in step S1; S3: The system collects the work instruction information issued by voice, and retrieves the operation instruction information, the operation locking information and the operation sequence control information according to the work instruction information to obtain the data set D2; S4: The system identifies the voice and identity information of sender A and receiver B in the telephone recording content, extracts the voice and identity information of sender A and receiver B, obtains data set D3, and compares D3 with D1 for voice and identity information; S5: The system identifies the work instruction about starting the No. 2 generator set in the telephone recording between the sender A and the receiver B, and sends the operation instruction information, operation locking information and operation sequence control information related to the work instruction retrieved from D2 to the receiver B; S6: Receiver B accepts the start-up instruction for generator set No. 2 issued by issuer A over the phone, and executes the start-up instruction for generator set No. 2. The system sequentially records the voice of receiver B executing each operation instruction. The system compares the recorded operation instruction with the operation instruction retrieved in D2 in step S5 item by item, and judges the locking information and operation sequence control information until all items of the operation instruction are compared.

[0017] In step S1, the voice database records the voice information of all personnel, extracts the features of the voice information, and classifies them into the sender and receiver of the instruction.

[0018] In step S2, the identity information specifically refers to the name and work unit of the issuer and receiver of the work order, and the processed data set D1 refers to a data set in which a one-to-one correspondence is established between the voice information and the identity information of the issuer and receiver of the work order.

[0019] In step S3, the work instruction information includes unit startup, shutdown, line power supply, power outage and equipment maintenance information; the operation instruction information is the operation steps and content information required to complete the work instruction; a work instruction information involves one or more operation instruction information operations; the locking information is the remaining unit, line, switch, and equipment information not involved in the work instruction; the operation sequence control information is the relationship information between the execution order of multiple operation instructions.

[0020] In step S3, the operation instruction information, operation locking information and operation sequence control information are retrieved according to the work instruction information, including entering the operation method, operation steps, operation locking and operation sequence control of unit startup, shutdown, line power supply, power outage and equipment maintenance into the system. The system automatically extracts the operation instruction information, operation locking information and operation sequence control information related to the work instruction according to the work instruction.

[0021] In step S4, the data set D3 is compared with the voice and identity information of the sender and receiver of the data set D1 in step S2, including the following sub-steps: S41: If the voice and identity information of the sender and the receiver are successfully compared, step S5 is executed; S42: If the comparison of the voice and identity information of the sender and the receiver is unsuccessful, the sender and the receiver are prompted to re-enter the telephone recording content, and step S4 is executed until the comparison is successful, and step S5 is continued.

[0022] In step S4, the system extracts the following contents from the telephone recording: 1. Name of sender: A, work unit: C; 2. Name of receiver: B, work unit: D; In step S5, the system extracts the content of the telephone recording as follows: work instruction: start the generator set 2. The information sent to receiver B in D2 includes: a. Operation instruction: instruction 1: start the generator set 2 to the water system, instruction 2: start the generator set 2 to no-load, instruction 3: start the generator set 2 to the grid; b. Locking information of the start-up operation instruction 1, 2, and 3 of the generator set 2; c. Sequential control information between the start-up operation instruction 1, 2, and 3 of the generator set 2; d. The total number of items R of the start-up operation instruction 1, 2, and 3 of the generator set 2, in this example R=3; In step S6, the system compares the input operation instructions with the operation instructions retrieved in D2 in step S5 one by one, and determines whether a certain operation step and content meets the locking condition and the sequence control condition, including the following sub-steps: S61: The recipient executes operation instructions 1, 2, and 3 one by one, and the system records the recipient's voice of operation instructions 1, 2, and 3 in sequence; S62: The system compares the consistency of operation instructions 1, 2, and 3 with the operation instruction content retrieved from data set D2 one by one. If they are inconsistent, S61 is executed; if they are consistent, S63 is executed; S63: The system judges the locking information of the operation instructions 1, 2, 3 and the operation instructions 1, 2, 3 retrieved from the data set D2 one by one. If they are locked, S61 is executed; if they are not locked, S64 is executed. S64: The system compares operation instructions 1, 2, 3 one by one with the order of operation instructions one, two, three retrieved from data set D2. If the order is inconsistent, execute S61; if the order is consistent, repeat step S61 to enter the next operation instruction until all items R of the operation instructions are compared. In this example, R=3, that is, operation instructions 1, 2, 3.

[0023] In step S6, the receiver executes the operation instructions 1, 2, and 3 one by one, namely, a. starting unit 2 to the water system, b. starting unit 2 to no load, and c. starting unit 2 to the grid.

Claims

1. A method for preventing misoperation based on artificial intelligence speech recognition, characterized in that: The following steps are involved: S1: Obtain the voice pronunciation of all personnel, use artificial intelligence to process the voice pronunciation and extract features to form a voice library; S2: The system collects the identity information of the issuer and receiver of the work order issued by voice, and obtains the data set D1 after processing with the voice library in step S1; S3: The system collects the work instruction information issued by voice, and retrieves the operation instruction information, the operation locking information and the operation sequence control information according to the work instruction information to obtain the data set D2; S4: The system identifies the content of the telephone recording of the issuer and the receiver of the work order, and extracts the voice and identity information of the issuer and the receiver to obtain the data set D3, and compares the voice and identity information of D3 with D1; S5: The system extracts the content related to the work instruction from the telephone recording content, and sends the operation instruction information, operation locking information and operation sequence control information related to the work instruction retrieved from D2 to the receiver; S6: The recipient executes the work instruction issued by the issuer over the phone. During the execution of the operation instruction, the system records the voice of the recipient executing the operation instruction. The system compares the recorded operation instruction with the operation instruction retrieved in D2 in step S5 item by item, and judges the locking information and operation sequence control information until all items of the operation instruction are compared.

2. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S1, the voice database records the voice information of all personnel, extracts the features of the voice information, and classifies them into the sender and receiver of the instruction.

3. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S2, the identity information is the name and work unit of the issuer and receiver of the work instruction, and the data set D1 is a data set that establishes a correspondence between the language information and identity information of the issuer and receiver.

4. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S3, the work instruction information includes unit startup, shutdown, line power supply, power outage and equipment maintenance information; the operation instruction information is the operation steps and content information required to complete the work instruction; one work instruction information involves one or more operation instruction information; the operation locking information is the remaining unit, line, switch, and equipment information not involved in the work instruction; the operation sequence control information is the relationship information between the execution order of multiple operation instructions.

5. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S3, the operation instruction information, operation locking information and operation sequence control information are retrieved according to the work instruction information, including entering the operation method, operation steps, operation locking and operation sequence control of unit startup, shutdown, line power supply, power outage and equipment maintenance into the system, and the system automatically extracts the operation instruction information, operation locking information and operation sequence control information related to the work instruction according to the work instruction.

6. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S4, the sound and identity information of D3 and D1 are compared, including the following sub-steps: S41: If the voice and identity information of the sender and the receiver are successfully compared, execute step S5; S42: If the comparison between the voice and identity information of the sender and the receiver is unsuccessful, the sender and the receiver are prompted to re-send and accept the work instruction, and step S4 is executed until the comparison between the sender and the receiver is successful, and step S5 is executed.

7. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S5, the system identifies the content of the telephone recordings of the issuer and receiver of the work order, extracts the content related to the work order, and retrieves the operation instruction information, operation locking information and operation sequence control information involved in the work order in D2, and sends this information to the receiver.

8. The method for preventing misoperation based on artificial intelligence speech recognition according to claim 1, characterized in that: In step S6, the system compares the input operation instruction with the operation instruction retrieved in step S5 D2 item by item, and determines the locking information and operation sequence control information, including the following sub-steps: S61: The receiver executes the operation instruction, and the system sequentially records the receiver's operation instruction voice; S62: The system compares the steps and contents of the input operation instruction with the operation instruction retrieved from the data set D2 for consistency. If they are inconsistent, S61 is executed; if they are consistent, S63 is executed; S63: The system compares the steps and contents of the input operation instruction with the locking information of the operation retrieved from the data set D2. If locked, S61 is executed; if not locked, S64 is executed. S64: The system compares the order of the entered operation instructions with the operation sequence control information retrieved from the data set D2 for consistency. If the control information is inconsistent, S61 is executed; if the control information is consistent, step S61 is repeated to enter the next operation instruction until all operation instructions are compared.

Images