A CNC machine tool interactive control method and system based on speech recognition

Abstract

The present invention relates to the field of CNC machine tool control technology, and in particular to a CNC machine tool interactive control method and system based on voice recognition, comprising an association template establishment module, an operation instruction recognition module, an operation instruction feedback module, and an auxiliary recognition module. When an operator inputs a large number of operation instructions, the operation instruction currently input by the operator is judged through the operation process of the CNC machine tool to determine whether it can be executed. In this process, the operation instructions that have been executed by the CNC machine tool are eliminated, and the number of matching calculations of parsed feature vectors and multiple target feature vectors is reduced. Only the instructions that have not been executed and meet the current machine tool status are calculated and matched, so that the CNC machine tool can respond to new instructions faster and improve the overall production efficiency. According to the operation process of the CNC machine tool, the operation instruction corresponding to the missing voice signal in multiple parsed voice signals is determined, even if some voice information is missing in the voice recognition link.

Description

Technical Field

[0001] The present invention relates to the technical field of CNC machine tool control, and in particular to a CNC machine tool interactive control method and system based on speech recognition. Background Art

[0002] Under the general trend of intelligent development in today's manufacturing industry, the existing voice recognition CNC machine tool system has improved the convenience of operation to a certain extent. By pre-setting the associated templates corresponding to different voice information, a bridge between voice and machine tool control is built. After the operator issues a voice command, the system can quickly and accurately call out the corresponding control command based on the preset associated template, and input it into the CNC machine tool for execution in a timely manner;

[0003] However, during the voice control process, when the operator inputs too much voice information at one time, during the transmission from the voice acquisition device to the recognition system, if the data transmission rate cannot keep up with the voice input speed, it will cause data backlog, resulting in voice information loss, and then lead to incomplete control instructions, affecting the normal operation of the CNC machine tool. In view of this, we propose a CNC machine tool interactive control method and system based on voice recognition. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that part of the voice information is lost during the voice information recognition process when an operator outputs multiple voice information at one time.

[0005] To achieve the above object, the present invention provides a method for interactive control of a CNC machine tool based on speech recognition, comprising the following steps:

[0006] S1. Collect language commands corresponding to operation instructions, operation procedures corresponding to language commands, and the running order of the operation procedures of CNC machine tools from operators in the CNC machine tool production workshop, define the known language commands as target voice signals, associate the target voice signals, operation instructions, and operation procedures with each other through association functions, and establish an association template;

[0007] S2, sensing the voice command input by the operator, defining the voice signal to be parsed, respectively extracting the target voice signal, the target feature vector of the parsed voice signal, and the parsed feature vector, and matching and calculating the voice similarity between the multiple target feature vectors and the parsed feature vectors;

[0008] A speech recognition threshold interval is set by using a plurality of target feature vectors. If the speech similarity between a target feature vector and a parsing feature vector is within the speech recognition threshold interval, the target feature vector is determined to be the same as the parsing feature vector, and the parsed speech corresponding to the determined parsing feature vector is converted into the target speech.

[0009] S3, establish a connection with the CNC machine tool, sense the current working state of the CNC machine tool, the operation process and operation instructions of the CNC machine tool, and determine whether the operation instruction can proceed to the next operation process under the current working state;

[0010] S4, eliminating the operation instructions corresponding to the current operation state, extracting the target feature vectors corresponding to the remaining operation instructions, and determining again the operation instructions corresponding to the analyzed voice signal input by the operator;

[0011] S5. If there are undetermined operation instructions between the consecutive operation instructions, the operation instructions between the operation instructions are retrieved by using the instruction elimination method. If there is only one operation instruction between the operation instructions, it is determined to be the operation instruction corresponding to the analyzed voice signal output by the operator.

[0012] As a further improvement of the technical solution, the correlation function perceives the set of target speech signals , a collection of operation instructions and operating procedures , define the correlation function , indicating the association between the target voice signal, operation instruction and operation process.

[0013] As a further improvement of the technical solution, the working principle of extracting the target speech signal, analyzing the target feature vector corresponding to the speech signal, and analyzing the feature vector in S2 is as follows:

[0014] Step 1: Use the window function to divide the target speech signal and the parsed speech signal into the same short time periods;

[0015] Step 2: Divide the windowed speech signal into several small segments, each of which is used as an analysis unit;

[0016] Step 3: Use discrete Fourier transform to convert each speech signal from time domain to frequency domain to obtain the spectrum information of the speech signal;

[0017] Step 4: Use the modulus operation to calculate the modulus of the spectrum, and the modulus of the spectrum is used as the feature vector of the speech signal in this period.

[0018] As a further improvement of the technical solution, the speech similarity between multiple target feature vectors and parsed feature vectors is matched and calculated:

[0019] The set of perceived target feature vectors is , the set of analytical eigenvectors is: , then the speech similarity is: ;

[0020] in, is a vector and The dot product of and They are vectors and The value range of speech similarity is The closer the value is, , which means the more similar the two vectors are, is any vector in the set of perceived target feature vectors, is any vector in the set of analytic eigenvectors.

[0021] As a further improvement of the technical solution, the steps of setting the speech recognition threshold interval by using multiple target feature vectors are as follows:

[0022] Retrieving a plurality of target feature vectors corresponding to the collected target voice signals, analyzing the target feature vectors according to the corresponding operation instructions, each operation instruction corresponds to a group of target feature vectors related thereto;

[0023] For each operation instruction group, the similarity between the target feature vectors in the group is calculated, and the maximum and minimum values ​​of the similarity are retrieved.

[0024] As a further improvement of the technical solution, the judgment of whether the operation instruction can proceed to the next operation process in the current working state is based on rule matching, and the following parts are used to represent the relevant states and rules:

[0025] Use collection Indicates all working status of CNC machine tools ,in Indicates Working state, using collection Indicates all possible operation instructions ,in Indicates Operation instructions, using relations Indicates the legal conversion relationship between operation instructions and working status. ,if , it means in working state Next, operation instructions If it is legal, you can proceed to the next step. Otherwise, it is not allowed.

[0026] As a further improvement of this technical solution, the operation instructions corresponding to the operation process of the CNC machine tool are: ,in is the total number of operation instructions in the list, Indicates Operation instructions;

[0027] The current operating status of the CNC machine tool is , from the first operation instruction in the operation instruction corresponding to the operation flow Start with Compare, when At this time The value is Find the corresponding operation instructions in the operation process. After the corresponding operation instruction, the operation flow composed of the remaining operation instructions , that is, from the operation process The The operation instruction starts from the first one and ends at the end of the operation flow: .

[0028] As a further improvement of the technical solution, the instruction exclusion method checks the continuous operation instructions one by one. The operation instructions are , the determined operation instructions are and ,Will and and Compare the operation instructions in one by one and call out and Between are the judgment operation instructions ,determination For undetermined operation instructions, the connection established with the CNC machine tool through wireless connection technology will transmit the operation instructions Output to CNC machine tools.

[0029] The interactive control system of CNC machine tools based on speech recognition includes an associated template establishment module, an operation instruction recognition module, an operation instruction feedback module, and an auxiliary recognition module;

[0030] The association template establishment module associates the target voice signal, the operation instruction and the operation process with each other through an association function to establish an association template;

[0031] The operation instruction recognition module defines the language command input by the operator as a parsed voice signal, extracts the parsed feature vector and the target feature vector corresponding to the parsed voice signal and the target voice signal respectively, calculates the voice similarity between the parsed feature vector and the target feature vector, and determines the operation instruction corresponding to the parsed voice signal by setting a voice recognition threshold interval;

[0032] The speech recognition threshold interval is set by calling out a group of target feature vectors corresponding to each operation instruction, calculating the similarity between the target feature vectors corresponding to different operation instructions, and setting the speech recognition threshold interval by the average of multiple maximum values ​​and multiple minimum values, and when it is determined that the parsed speech signal corresponds to the operation instruction, setting the parsed speech signal as the target speech signal, and updating the speech recognition threshold interval;

[0033] The operation instruction feedback module determines whether the operation instruction corresponding to the analyzed voice signal can be used to perform the next operation process under the current working state after the operation instruction corresponding to the analyzed voice signal is input into the CNC machine tool, and feeds back the judgment result to the operator;

[0034] The auxiliary recognition module is used to eliminate the operation instructions corresponding to the current operation state of the CNC machine tool, and when multiple operation instructions are input at the same time, assist in determining the corresponding operation instructions according to the operation flow of the CNC machine tool.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] In the interactive control method and system of a CNC machine tool based on speech recognition, the speech recognition threshold interval is set by setting the maximum value and the minimum value between the target feature vectors, the speech recognition threshold interval is used to determine the control instruction corresponding to the parsed speech signal, and after the determination, the parsed speech signal is used as the target speech signal again, and the target feature vector corresponding to the target speech signal is extracted to update the speech recognition threshold interval, which has the ability of dynamic learning and optimization. As the operator continuously inputs the speech instruction, the recognition threshold can be continuously adjusted according to the new speech data, so as to improve the recognition efficiency and accuracy of the speech instruction, so that the speech recognition function is more and more in line with the actual use needs;

[0037] When the operator inputs a large number of control instructions, it is judged whether the control instructions currently input by the operator can be executed. In this process, the control instructions that have been executed by the CNC machine tool are eliminated, reducing the number of analytical feature vectors and multiple target feature vector matching calculations. Only the instructions that have not been executed and are consistent with the current machine tool status are calculated and matched, which improves the processing efficiency of the voice recognition of this system, enables the CNC machine tool to respond to new instructions faster, and improves the overall production efficiency;

[0038] According to the control process of the CNC machine tool, the control instructions corresponding to the lost voice signal in multiple analyzed voice signals are determined, and the control instructions corresponding to the lost voice signal can be restored as much as possible. Even if some voice information is missing in the voice recognition link, the integrity of the control instructions can be guaranteed to the greatest extent through analysis and inference of the machine tool control process, avoiding processing errors or machine tool failures caused by voice loss, thereby ensuring the stability of the CNC machine tool operation and the efficiency of processing products. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the overall module principle of the present invention;

[0040] Figure 2 A schematic diagram of the module principle for establishing the association template of the present invention;

[0041] Figure 3 This is a schematic diagram of the principle of the operation instruction recognition module of the present invention;

[0042] Figure 4 It is a schematic diagram of the working method of the present invention.

[0043] The meaning of each number in the figure is:

[0044] 100, association template establishment module; 200, operation instruction recognition module; 300, operation instruction feedback module; 400, auxiliary recognition module. DETAILED DESCRIPTION

[0045] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] CNC machine tools are automated machine tools equipped with a program control system. They can operate and process parts according to the programmed program. They integrate multiple technologies such as mechanics, automation, computers, measurement, and microelectronics. They have the advantages of high precision, high efficiency, and high flexibility. With the continuous development of the manufacturing industry, CNC machine tools are increasingly used in industrial production and have become key equipment in modern manufacturing. At the same time, the existing voice recognition technology is becoming more and more mature. Through voice recognition technology, operators can give instructions to CNC machine tools simply by speaking.

[0047] The interactive control method of a CNC machine tool based on speech recognition of this embodiment can be applied to a CNC machine tool. After the operator inputs the parsed speech signal, the operator calls out the target speech signal corresponding to the parsed speech signal and the operation instruction corresponding to the target speech signal in the association template, and inputs the operation instruction into the CNC machine tool through the connection established inside the CNC machine tool.

[0048] At the same time, the operation instructions are compared with the operation procedures. When the operation instructions are in line with the operation procedures, the CNC machine tool runs according to the operation instructions. When the operation instructions do not meet the operation procedures, the CNC machine tool does not run, and the results are output to the operator through the connection established inside the CNC machine tool.

[0049] S1. Collect language commands corresponding to operation instructions, operation procedures corresponding to language commands, and the running order of the operation procedures of CNC machine tools from operators in the CNC machine tool production workshop, define the known language commands as target voice signals, associate the target voice signals, operation instructions, and operation procedures with each other through association functions, and establish an association template;

[0050] The collection of correlation functions that perceive the target speech signal , a collection of operation instructions and operating procedures , define the correlation function , indicating the association between the target voice signal, operation instruction and operation process.

[0051] S2, sensing the voice command input by the operator, defining the voice signal to be parsed, respectively extracting the target voice signal, the target feature vector of the parsed voice signal, and the parsed feature vector, and matching and calculating the voice similarity between the multiple target feature vectors and the parsed feature vectors;

[0052] The speech recognition threshold interval is set by multiple target feature vectors. The setting of the threshold interval provides a reasonable judgment range for speech recognition, avoiding the misjudgment that may be caused by a single fixed threshold. If the speech similarity between a target feature vector and a parsing feature vector is within the speech recognition threshold interval, the target feature vector is judged to be the same as the parsing feature vector, and the operation instruction corresponding to the target speech signal in the associated template is called out to ensure the consistency of understanding of the speech instruction. During the entire speech recognition and instruction execution process, the target speech is used as the standard to avoid the instruction understanding deviation caused by the diversity of the parsed speech, and the parsed speech corresponding to the determined parsed feature vector is converted into the target speech.

[0053] The working principles of extracting the target speech signal, analyzing the target feature vector corresponding to the speech signal, and analyzing the feature vector are as follows:

[0054] Step 1: Use the window function (the function of the window function is to select a window of finite length on the time axis corresponding to the target speech signal and the analyzed speech signal, and extract the speech signal within the window) to divide the target speech signal and the analyzed speech signal into the same short time period. The expression is:

[0055] ;

[0056] in is the length of the window function, is a discrete time point, and the window function With speech signal Multiply them to get the windowed speech signal ,Right now , the value of the windowed speech signal outside the window is , which is equivalent to focusing only on the speech signal part within the window. The use of the window function avoids a general analysis of the entire speech signal and can carefully observe the speech characteristics within each short period of time;

[0057] Step 2: Divide the windowed speech signal into several small segments, each of which is used as an analysis unit. The specific calculation formula is:

[0058] Assume that the length of each segment is , the step length of the segment is (i.e. the interval between two adjacent segments), then Speech signal Expressed as ;

[0059] Step 3: Use discrete Fourier transform to convert each speech signal from time domain to frequency domain. Discrete Fourier transform represents discrete time domain speech signals as a linear combination of a series of sine and cosine functions of different frequencies to obtain the spectrum information of the speech signal. The calculation formula is:

[0060] ;

[0061] in is an imaginary unit, Indicates the frequency point, Indicates The frequency point of the speech signal The spectrum value at is calculated by discrete Fourier transform to obtain the spectrum of each speech signal, which reflects the frequency composition of the speech signal in this period.

[0062] Step 4: Use the modulus operation to calculate the modulus of the spectrum. The modulus of the spectrum is used as the feature vector of the speech signal in this period. The modulus of the spectrum is , which represents the energy of the speech signal at each frequency point, ignoring the phase information of the spectrum.

[0063] Match and calculate the speech similarity between multiple target feature vectors and parsed feature vectors:

[0064] The set of perceived target feature vectors is , the set of analytical eigenvectors is: , then the speech similarity is: ;

[0065] in, is a vector and The dot product of and They are vectors and The value range of speech similarity is The closer the value is, , which means the more similar the two vectors are, is any vector in the set of perceived target feature vectors, is any vector in the set of analytic eigenvectors.

[0066] The steps for setting the speech recognition threshold interval through multiple target feature vectors are as follows:

[0067] Retrieving a plurality of target feature vectors corresponding to the collected target voice signals, analyzing the target feature vectors according to the corresponding operation instructions, each operation instruction corresponds to a group of target feature vectors related thereto;

[0068] For each operation instruction group, calculate the similarity between the target feature vectors in the group, and call out the maximum and minimum values ​​of the similarity;

[0069] Assume that for an operation instruction group, there is target feature vector , for the feature vector and , and the similarity calculation formula is: ;

[0070] in is the dot product of the vectors, and They are vectors and Model.

[0071] The average of the maximum and minimum values ​​in multiple operation instruction groups is calculated respectively and set as the speech recognition threshold interval. The working principle is as follows:

[0072] Assume that The operation instruction groups are used to calculate the average of the maximum value and the minimum value respectively, and the upper and lower limits of the speech recognition threshold range are obtained;

[0073] Mean of the maximum values for: , minimum The mean of is: , then the speech recognition threshold interval is ,in, For the The maximum value, For the minimum value.

[0074] S3, establish a connection with the CNC machine tool through wireless connection technology, sense the working state, operation process and operation instructions of the CNC machine tool at this time, judge whether the operation instruction can carry out the next operation process under the current working state, and feed back the judgment result to the operator;

[0075] Wireless connection technologies include Bluetooth connection, Wi-Fi connection and Zigbee connection. Bluetooth technology uses the 2.4GHz ISM (industrial, scientific, and medical) frequency band and adopts frequency hopping spread spectrum technology to avoid interference and improve communication reliability. Before establishing a connection with the CNC machine tool, the device discovery and pairing process is performed first. During the discovery stage, broadcast data packets are sent periodically, and the CNC machine tool listens for connection requests. During pairing, encryption keys are exchanged with the CNC machine tool to establish a secure connection, thereby establishing a connection.

[0076] Determine whether the operation instruction can proceed to the next operation process under the current working state, based on rule matching judgment, and use the following parts to represent the relevant states and rules:

[0077] Use collection Indicates all working status of CNC machine tools ,in Indicates Working state, using collection Indicates all possible operation instructions ,in Indicates Operation instructions, using relations Indicates the legal conversion relationship between operation instructions and working status. ,if , it means in working state Next, operation instructions If it is legal, you can proceed to the next step. Otherwise, it cannot.

[0078] During the processing of CNC machine tools, if the operator outputs the parsed voice signal during the voice recognition process and there is no timely feedback, the operator will continue to operate the CNC machine tool according to the wrong instruction intention because of the lack of feedback on the recognition error. For example, the operator should have input "reduce cutting speed", but the "increase cutting speed" instruction was executed due to recognition error, which will cause excessive wear or even chipping of the tool, deterioration of the workpiece surface quality, and in severe cases, the processing may be interrupted, damaging the workpiece and tool, increasing production costs and processing time.

[0079] When the operator inputs a large number of parsed voice signals at one time, S2 is used to determine the operation instructions corresponding to the parsed voice signals one by one. At this time, if the language similarities between multiple target feature vectors and parsed feature vectors are calculated one by one, it will cause too long calculation time and reduce production efficiency. On some production lines with strict requirements on processing time, the long waiting time caused by calculating the similarities one by one may disrupt the entire production rhythm and affect product delivery time.

[0080] S4, sensing the operation process of the CNC machine tool, using the connection established with the CNC machine tool through the wireless connection technology to sense the current operation state of the CNC machine tool, eliminating the operation instructions corresponding to the current operation state, extracting the target feature vectors corresponding to the remaining operation instructions, and again determining the operation instructions corresponding to the operator's input parsed voice signal. In this way, the scope of determination is narrowed and the accuracy of determination is improved;

[0081] The operating instructions corresponding to the CNC machine tool operation process are: ,in is the total number of operation instructions in the list, Indicates Operation instructions;

[0082] The current operating status of the CNC machine tool is , from the first operation instruction in the operation instruction corresponding to the operation flow Start with Compare, when At this time The value is Find the corresponding operation instructions in the operation process. After the corresponding operation instruction, the operation flow composed of the remaining operation instructions , that is, from the operation process The The operation instruction starts from the first one and ends at the end of the operation flow: .

[0083] S5. When the operator inputs a large number of parsed voice signals at one time, in the process of determining the operation instructions corresponding to the parsed voice signals through S3, if there are undetermined operation instructions between consecutive control instructions, the operation instructions between the operation instructions are called out by the instruction elimination method. If there is only one control instruction between the operation instructions, it is determined to be the control instruction corresponding to the parsed voice signal output by the operator. By analyzing the relationship between the instructions, the target instruction is accurately locked. Compared with the traditional one-by-one matching determination method, the accuracy and speed of the determination are improved. At this time, the parsed voice corresponding to the determined parsed feature vector is deleted to avoid errors in the process of updating the voice recognition threshold interval.

[0084] The instruction elimination method checks the continuous operation instructions one by one. The operation instructions are , the determined operation instructions are and ,Will and and Compare the operation instructions in one by one and call out and Between are the judgment operation instructions ,determination For undetermined operation instructions, the connection established with the CNC machine tool through wireless connection technology will transmit the operation instructions Output to CNC machine tools.

[0085] The interactive control system of a numerical control machine tool based on speech recognition includes an associated template establishment module 100, an operation instruction recognition module 200, an operation instruction feedback module 300, and an auxiliary recognition module 400;

[0086] The correlation template establishing module 100 correlates the target voice signal, the operation instruction and the operation process with each other through a correlation function to establish a correlation template;

[0087] The operation instruction recognition module 200 defines the language command input by the operator as a parsed voice signal, extracts the parsed feature vector and the target feature vector corresponding to the parsed voice signal and the target voice signal respectively, calculates the voice similarity between the parsed feature vector and the target feature vector, and determines the operation instruction corresponding to the parsed voice signal by setting a voice recognition threshold interval;

[0088] The speech recognition threshold interval is set by calling out a group of target feature vectors corresponding to each operation instruction, calculating the similarity between the target feature vectors corresponding to different operation instructions, and setting the speech recognition threshold interval by the average of multiple maximum values ​​and multiple minimum values, and when it is determined that the parsed speech signal corresponds to the operation instruction, setting the parsed speech signal as the target speech signal, and updating the speech recognition threshold interval;

[0089] After the operation instruction corresponding to the analyzed voice signal is input to the CNC machine tool, the operation instruction feedback module 300 determines whether the operation instruction can be used to perform the next operation process under the current working state, and feeds back the judgment result to the operator;

[0090] The auxiliary recognition module 400 is used to eliminate the operation instructions corresponding to the current operation state of the CNC machine tool, and when multiple operation instructions are input at the same time, assist in determining the corresponding operation instructions according to the operation flow of the CNC machine tool.

[0091] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for interactive control of CNC machine tools based on speech recognition, characterized in that: The following steps are involved: S1. Collect the language commands corresponding to the operation instructions, the operation flow corresponding to the language commands, and the running order of the CNC machine tool operation flow, define the known language commands as the target voice signals, associate the target voice signals, the operation instructions and the operation flow with each other through the association function, and establish an association template; S2, sensing the voice command input by the operator, defining the voice signal to be parsed, respectively extracting the target voice signal, the target feature vector of the parsed voice signal, and the parsed feature vector, and matching and calculating the voice similarity between the multiple target feature vectors and the parsed feature vectors; A speech recognition threshold interval is set by using a plurality of target feature vectors. If the speech similarity between a target feature vector and a parsing feature vector is within the speech recognition threshold interval, the target feature vector is determined to be the same as the parsing feature vector, and the parsed speech corresponding to the determined parsing feature vector is converted into the target speech. S3, establish a connection with the CNC machine tool, sense the current working state of the CNC machine tool, the operation process and operation instructions of the CNC machine tool, and determine whether the operation instruction can proceed to the next operation process under the current working state; S4, eliminating the operation instructions corresponding to the current operation state, extracting the target feature vectors corresponding to the remaining operation instructions, and determining again the operation instructions corresponding to the analyzed voice signal input by the operator; S5. If there are undetermined operation instructions between the consecutive operation instructions, the operation instructions between the operation instructions are retrieved by using the instruction elimination method. If there is only one operation instruction between the operation instructions, it is determined to be the operation instruction corresponding to the analyzed voice signal output by the operator.

2. The interactive control method for CNC machine tools based on speech recognition according to claim 1 is characterized in that: The association function perceives the set of target speech signals , a collection of operation instructions and operating procedures , define the correlation function , indicating the association between the target voice signal, operation instruction and operation process.

3. The interactive control method for CNC machine tools based on speech recognition according to claim 2 is characterized in that: The working principle of extracting the target speech signal, analyzing the target feature vector corresponding to the speech signal, and analyzing the feature vector in S2 is as follows: Step 1: Use the window function to divide the target speech signal and the parsed speech signal into the same short time periods; Step 2: Divide the windowed speech signal into several small segments, each of which is used as an analysis unit; Step 3: Use discrete Fourier transform to convert each speech signal from time domain to frequency domain to obtain the spectrum information of the speech signal; Step 4: Use the modulus operation to calculate the modulus of the spectrum, and the modulus of the spectrum is used as the feature vector of the speech signal in this period.

4. The interactive control method for CNC machine tools based on speech recognition according to claim 3 is characterized in that: Match and calculate the speech similarity between multiple target feature vectors and parsed feature vectors: The set of perceived target feature vectors is , the set of analytical eigenvectors is: , then the speech similarity is: ; in, is a vector and The dot product of and They are vectors and The value range of speech similarity is The closer the value is, , which means the more similar the two vectors are, is any vector in the set of perceived target feature vectors, is any vector in the set of analytic eigenvectors.

5. The interactive control method for CNC machine tools based on speech recognition according to claim 4 is characterized in that: The steps for setting the speech recognition threshold interval through multiple target feature vectors are as follows: Retrieving a plurality of target feature vectors corresponding to the collected target voice signals, analyzing the target feature vectors according to the corresponding operation instructions, each operation instruction corresponds to a group of target feature vectors related thereto; For each operation instruction group, the similarity between the target feature vectors in the group is calculated, and the maximum and minimum values ​​of the similarity are retrieved.

6. The interactive control method for CNC machine tools based on speech recognition according to claim 5 is characterized in that: The judgment of whether the operation instruction can proceed to the next operation process in the current working state is based on rule matching, and the following parts are used to represent the relevant states and rules: Use collection Indicates all working status of CNC machine tools ,in Indicates Working state, using collection Indicates all possible operation instructions ,in Indicates Operation instructions, using relations Indicates the legal conversion relationship between operation instructions and working status. ,if , it means in working state Next, operation instructions If it is legal, you can proceed to the next step. Otherwise, it is not allowed.

7. The interactive control method for CNC machine tools based on speech recognition according to claim 1, characterized in that: The operating instructions corresponding to the CNC machine tool operation process are: ,in is the total number of operation instructions in the list, Indicates Operation instructions; The current operating status of the CNC machine tool is , from the first operation instruction in the operation instruction corresponding to the operation flow Start with Compare, when At this time The value is Find the corresponding operation instructions in the operation process. After the corresponding operation instruction, the operation flow composed of the remaining operation instructions , that is, from the operation process The The operation instruction starts from the first one and ends at the end of the operation flow: .

8. The interactive control method for CNC machine tools based on speech recognition according to claim 7 is characterized in that: The instruction elimination method checks the continuous operation instructions one by one. The operation instructions are , the determined operation instructions are and ,Will and and Compare the operation instructions in one by one and call out and The operation instructions are determined between ,determination For undetermined operation instructions, the connection established with the CNC machine tool through wireless connection technology will transmit the operation instructions Output to CNC machine tools.

9. A method for realizing an interactive control system for a CNC machine tool based on speech recognition, applied to an interactive control method for a CNC machine tool based on speech recognition as claimed in any one of claims 1 to 8, characterized in that: It comprises an associated template establishing module (100), an operation instruction identifying module (200), an operation instruction feedback module (300), and an auxiliary identifying module (400); The association template establishment module (100) associates the target voice signal, the operation instruction and the operation process with each other through an association function to establish an association template; The operation instruction recognition module (200) defines the language command input by the operator as a parsed voice signal, extracts the parsed feature vector and the target feature vector corresponding to the parsed voice signal and the target voice signal respectively, calculates the voice similarity between the parsed feature vector and the target feature vector, and determines the operation instruction corresponding to the parsed voice signal by setting a voice recognition threshold interval; The speech recognition threshold interval is set by calling out a group of target feature vectors corresponding to each operation instruction, calculating the similarity between the target feature vectors corresponding to different operation instructions, and setting the speech recognition threshold interval by the average of multiple maximum values ​​and multiple minimum values, and when it is determined that the parsed speech signal corresponds to the operation instruction, setting the parsed speech signal as the target speech signal, and updating the speech recognition threshold interval; The operation instruction feedback module (300) determines whether the operation instruction corresponding to the analyzed voice signal can be used to perform the next operation process under the current working state after the operation instruction corresponding to the analyzed voice signal is input into the numerical control machine tool, and feeds back the determination result to the operator; The auxiliary recognition module (400) is used to eliminate the operation instructions corresponding to the current operation state of the numerical control machine tool, and when multiple operation instructions are input simultaneously, to assist in determining the corresponding operation instructions according to the operation flow of the numerical control machine tool.

Images